The Role of Sensation Seeking and Need for Cognition on Web-Site Evaluations: A ResourceMatching Perspective

A second prerequisite for the Web sites was that they had to appeal to certain personality types. It was believed that high sensation seekers, due to their liking of visually complex stimuli, would prefer Web sites that displayed animations. Hence, a second test ascertained that the animations were more imagery eliciting and led to greater amounts of sensory processing relative to the static images. To this end, the data wereBrett A. S. Martin University of Auckland Business School, New Zealand

Michael J. Sherrard New Zealand Milk Ltd.

Daniel Wentzel University of St. Gallen, Switzerland

ABSTRACT The Internet theoretically enables marketers to personalize a Web site to an individual consumer. This article examines optimal Website design from the perspective of personality trait theory and resource-matching theory. The influence of two traits relevant to Internet Web-site processing—sensation seeking and need for cognition—were studied in the context of resource matching and different levels of Web-site complexity. Data were collected at two points of time: personality-trait data and a laboratory experiment using constructed Web sites. Results reveal that (a) subjects prefer Web sites of a medium level of complexity, rather than high or low complexity; (b) high sensation seekers prefer complex visual designs, and low sensation seekers simple visual designs, both in Web sites of medium complexity; and (c) high need-for-cognition subjects evaluated Web sites with high verbal and low visual complexity more favorably. © 2005 Wiley Periodicals, Inc.

Many researchers have predicted that the Internet may change the nature of advertising (Hofacker & Murphy, 2000; Hoffman & Novak, 1996). Unlike other mass media, the Internet potentially enables advertisers to tailor messages to individual consumers (Quinn, 1999). However, to date, only a few studies have produced empirical findings that can guide marketers in the design of effective Web sites (e.g., Dreze & Zufryden, 1997; Zhang, 2000).

Importantly, many Web sites now include tracking devices, called intelligent agents, that follow Web surfers and analyze patterns of behavior, such as which sites consumers have visited before arriving at a target site (Maes, 2000). LaBarbera (1999) suggests that this tracking software could be equipped with a personality component, which could predict users’ preferences based on the links they have selected. Because multiple designs of a Web site are relatively easy to achieve, such agents could segment users and expose them to a site version that is best suited to their personality type

The purpose of this study is to address this personalization issue by investigating the influence of relevant personality traits, in conjunction with the cognitive difficulty experienced by Web surfers in processing Website stimuli. Specifically, sensation seeking and need for cognition are studied and these traits are considered in relation to the visual and verbal complexity of the Web sites in a laboratory experiment. The following sections outline relevant literature. Next, research hypotheses, method, and results will be presented. Finally, the results will be discussed and future research directions presented.

CONCEPTUAL DEVELOPMENT 

Sensation Seeking

People engage in certain behaviors with the purpose of increasing their level of sensory stimulation. This need for sensory arousal, called sensation seeking, varies across individuals, with some people having a generally higher preferred stimulation level than others (Raju, 1980; Zuckerman, 1994). Berlyne (1971) suggested that stimuli vary in their capacity to increase arousal, a phenomenon that can be induced by stimulus complexity. Complexity, as defined by the number of independent units in a stimulus, can heighten arousal by increasing the cognitive demands that are necessary to appraise it (Zuckerman, 1994). Thus, based on the notion that high sensation seekers have higher levels of optimal arousal, they should show a greater liking for stimuli with high complexity.

Consistent with this perspective, Zuckerman, Bone, Neary, Mangelsdorf, and Brustman (1972) found that high sensation seekers liked visual designs that were complex and asymmetrical, whereas low sensation seekers preferred symmetrical and simple figures.

Research has also found that high sensation seekers like ambiguous and abstract paintings (Furnham & Bunyan, 1988; Zuckerman, Ulrich, & McLaughlin, 1993). Thus, with regard to the Internet, a high sensation seeker may find a Web site stimulating because it possesses arousing qualities. He or she may attach great importance to the sensory experience of the site, such as provided by complex graphic animations, whereas a low sensation seeker may prefer a graphically simple site.

Need for Cognition 

Need for cognition (NFC) refers to an individual’s tendency to seek and engage in effortful thinking (Cacioppo & Petty, 1982). People low in NFC prefer to avoid cognitively demanding activities, whereas those high in NFC possess an intrinsic motivation to think. A substantial stream of research has shown that NFC can be an important antecedent of attitude change (Cacioppo & Petty, 1982; Cacioppo, Petty, Kao, & Rodriguez, 1986; Zhang, 1996). For high NFC individuals, attitude change is a function of the strength and merit of the message. Conversely, low NFC individuals are more likely to prefer more heuristic strategies (Haugtvedt, Petty, & Cacioppo, 1992).

Further, high NFC individuals are also more likely to favor informationorientated media, and a verbal over a visual processing style (Ahlering, 1987; Heckler, Childers, & Houston, 1993). Owing to the Internet’s capacity to provide information, it may be of particular appeal to high NFC individuals who are likely to be more interested in the quality of the verbal information presented, than in execution characteristics like graphics or sound effects (Cacioppo, Petty, Feinstein, & Jarvis, 1996). Conversely, low NFC individuals may be more prone to the influence of symbolic cues in Web sites given that they will avoid elaborative processing. Hence, low NFC individuals may base their attitudes not on the actual informational content of the site, but on the attractiveness of the execution characteristics (e.g., graphics, sound effects).

Resource-Matching Theory

The final theoretical concept to be reviewed is resource-matching theory. This states that the persuasive impact of a message is maximized when the resources allocated to processing the communication match those required for the task. Stimuli that demand too much or too little of the resources that an individual makes available should undermine persuasion (Peracchio & Meyers-Levy, 1997).

Previous consumer research has studied resource matching in the context of promotional stimuli. For instance, Meyers-Levy and Peracchio (1995) examined the effects of color in terms of resource matching. They found that color ads, being more difficult to process than black-and-white ads, were more persuasive when the resource demands of the ad copy were low.

Because the low demands of the ad copy did not occupy all available resources, color had a beneficial impact by communicating sensory information that reinforced the ad claims. Yet for high resource demands, color undermined persuasion by directing attention away from the advertising claims.Although this research has furthered the understanding of resource matching, no published study to date has examined the relationship between stable individual differences, such as NFC, and resource-matching theory.

However, it seems plausible that perceptions of complexity, and the evaluation of sensory information such as color, may vary across individuals. Given that a high sensation seeker is more prone to sensory processing, he or she might be able to extract more relevant information out of sensory cues, in which case the use of color is more important for high sensation seekers. Conversely, the potentially distracting effects of color may be greater for low sensation seekers and may have a negative impact on attitudes. Similarly, high NFC individuals are likely to favor complex ad copy, and since they are more prone to employ a verbal processing style, the use of sensory information may be of limited use. Yet low NFC individuals may prefer simple verbal information and should attach greater importance to executional characteristics.

Resource-matching theory is applicable to the Internet, as individuals may evaluate visual and verbal elements in Web sites differently. A Web site that displays complex visual and verbal information may be too complex even for high sensation seekers or high NFC. Equally, a site that is visually and verbally simple may provide too little stimulation for even low sensation seekers or low NFC. However, trait differences may be more evident when stimuli complexity is at a medium rather than an extreme level (Cacioppo et al., 1996). This does not imply that high NFC individuals are unable or unwilling to process complex visual information. What it does imply is that they should show more positive attitudes toward a Web site with a modality that matches their processing style. In this study, by combining visual complexity (high, low) with verbal complexity (high, low) it is possible to create three broader levels of complexity—high, low, and medium. For example, combining static images (low visual complexity) with a spacious text version (low verbal complexity) should result in a low level of complexity. Conversely, animation (high visual complexity) paired with a crammed text version (high verbal complexity) is expected to be high complexity, with the two remaining conditions representing medium complexity (i.e., high visual–low verbal or low visual–high verbal). Thus, in accordance with resource-matching theory it is postulated that:

H1: Consumers will form more favorable brand attitudes, Web-site attitudes, and purchase intentions after being exposed to a Web site of a medium level of complexity rather than high or low complexity. Further, it is posited that certain combinations of design features will have unique appeal for the trait types. For sensation seeking, the arousal-inducing qualities of complex animations should mainly appeal to high sensation seekers (Zuckerman, 1994), whereas low sensation seekers should prefer static images to minimize arousal. This leads to the following hypotheses:

H2a: Subjects high in sensation seeking will exhibit more favorable brand attitudes, Web-site attitudes, and purchase intentions toward a Web site that combines high visual complexity with low verbal complexity.

H2b: Subjects low in sensation seeking will exhibit more favorable brand attitudes, Web-site attitudes, and purchase intentions toward a Web site that combines low visual complexity with high verbal complexity.

For need for cognition, high NFC subjects should prefer a complex verbal format, and the reverse should occur for low NFC subjects. However, these predicted effects should be subordinate to the effects of resource matching outlined in H1.Hence it is proposed that:

H3a: Subjects high in need for cognition will exhibit more favorable brand attitudes, Web-site attitudes, and purchase intentions toward a Web site that combines low visual complexity with high verbal complexity.

H3b: Subjects low in need for cognition will exhibit more favorable brand attitudes, Web-site attitudes, and purchase intentions toward a Web site that combines high visual complexity with low verbal complexity.

METHOD

Research Design and Sample 

The main study employed a 2 (visual complexity: high, low)  2 (verbal complexity: high, low) factorial design. The hypotheses were tested with Web sites created for the study. The data collection involved two separate phases: the collection of trait scores and the main study. These took place 6 weeks apart to avoid respondent fatigue and demand effects, which allowed a more precise evaluation of the impact of trait variables (Steenkamp & Baumgartner, 1992). Two hundred thirty-four undergraduate students enrolled in marketing classes (105 males, 129 females, age range: 20–23) participated in the collection of the personality scores, and 117 from this group participated in the main study (49 males, 68 females, age range: 20–23).

Subjects in this group of 117 rated their Web expertise in response to the statement “With regards to the Internet, I consider myself” on two 7-point items anchored by inexperienced–experienced and a novice–an expert (r .90). Results of the averaged scales yielded a mean of 4.68 (SD 1.35), with 69.2% of subjects rating their Web expertise above the midpoint. Thus, subjects may be considered as reasonably familiar with the Internet.

Stimuli Development

Product Category Selection. The product to be advertised on the Web site was determined after an extensive screening process. First, a focus group of 10 students was conducted to derive an initial list of product categories subjects were familiar and involved with (cameras, personal computers, running shoes, energy drinks, and mobile phones). A survey of 23 undergraduate marketing students excluded from the main study was then performed to determine the most appropriate product category for the experiment. Subjects rated involvement (Cronbach’s alpha 0.89) and familiarity (Cronbach’s alpha 0.86) using scales adapted from previous studies (Mittal, 1995; Oliver & Bearden, 1985). Results indicated that two products, energy drinks and personal computers, were the most familiar and involving to subjects. However, relative to energy drinks, the range of product attributes of computers that can be portrayed in a sensory manner appears limited, whereas more experiential products seem better suited to visual and verbal descriptions.

Further, two expert judges (one assistant professor, one associate professor of marketing) agreed that energy drinks were a more appropriate choice for visual and verbal representation, resulting in energy drinks being chosen. Energy drinks are commonly marketed as aiding reaction speed, concentration, and stamina levels through a product that often contains ingredients such as caffeine and vitamins. Examples of prominent energy drink brands include Red Bull, Shark, and V.

Web Sites. Web sites containing verbal and visual complexity manipulations for an energy drink called Dark Dog were then developed. By using images from a foreign brand (i.e., Austrian) it was possible to create a more realistic Web site. Researchers have highlighted that amateurish ads can undermine the external validity of findings (e.g., Mitchell, 1986). Pretests also revealed that the target population was unfamiliar with the brand, so the impact of prior brand knowledge was deemed minimal (Gardner, 1985).

Visual complexity was manipulated through the use of animations (high complexity) versus static images (low complexity). The animations included graphical elements that vibrated and bounced when clicked on, short storyboard movies, themes that appeared automatically, and an extensive introductory sequence. In the low complexity treatment all the images were static (see Appendix A).

Verbal complexity was manipulated by the amount of text that was included on a single page. Some advertisers prefer to crowd their Web pages with the maximum amount of text, whereas others limit the amount of information presented. Where the information is essentially the same, a crammed Web page should be more cognitively demanding because more information needs to be processed simultaneously. For low verbal complexity, all text pertaining to a topic was presented on a Web page. For high verbal complexity, information was grouped on the one page with subjects frequently having to use scroll-down buttons in order to read all of the information (see scroll-down feature in Appendix B). Web-site stimuli were loaded locally onto laboratory computers. Using standardized computers controlled for two potentially biasing effects: first, variety in the age and type of browser software, which may not be able to display advanced animations, and second, any temporal effects associated with variability in waiting times for information to download (see Davis & Hantula, 2001; Rajala & Hantula, 2000, for a discussion of the effects of longer download times) Thus, loading onto the hard drives resulted in comparable download times across conditions (i.e., 1.29–1.65 s).

Pretest

A pretest among 38 marketing undergraduate students excluded from the main study was conducted to verify the intended manipulations. Between 8 and 12 subjects were exposed to each of the four experimental Web sites and asked to browse through them as they normally would. Following this, subjects completed a questionnaire that contained various measures about perceived visual and verbal complexity.To measure visual complexity, subjects indicated on two 7-point scales anchored by difficult to understand–easy to understand and complicated–simple (r .62).These measures were also used to measure verbal complexity (r .85) and were adapted from Meyers-Levy and Peracchio (1995). As anticipated, a 2 (verbal complexity: high vs. low)  2 (visual complexity: high vs. low) analysis of variance (ANOVA) revealed a significant interaction; F(3,34) 3.28, p .05. Specifically, the low verbal–low visual complexity version (M 6.08, SD 0.44) was rated as less complex than both of the medium complexity conditions. Namely, the high verbal–low visual (M 5.58, SD 0.51) and low verbal–high visual conditions (M 5.56, SD 0.22). The high complexity condition (i.e., high verbal–high visual complexity) was also rated as the most complex (M 5.19, SD 0.54). Hence, the complexity manipulations were successful.

collapsed into two categories depending on whether respondents had seen stimuli of high or low visual complexity. Three 7-point scales adapted from Burnkrant and Unnava (1995) assessed to what extent subjects formed mental pictures while looking at the site (e.g., “I found myself thinking of images or pictures when I browsed through the site,” Cronbach’s alpha 0.86). As expected, subjects reported higher degrees of sensory processing [t(36) 0.46, p .05] when exposed to complex animations of high visual complexity (M 4.80, SD 1.28) relative to static images of low visual complexity (M 3.73, SD 1.02). Last, a series of items determined that the Web sites did not differ in comprehensibility, relevance, and appropriateness.

Procedure

For the measurement of the personality traits, 234 undergraduate marketing students participated voluntarily with a cash draw offered as an incentive. Subjects completed a booklet containing trait measures and demographic variables and provided their 7-digit ID number. This process took approximately 20 minutes. Six weeks later, these subjects were invited to participate in an experiment on Web-site effectiveness. To avoid demand effects, no indication was given that this experiment was connected to the trait measurement. The researcher invited students to come to a computer lab at certain times to participate in the experiment. Subjects were seated individually at a computer and asked to browse through the site as they normally would. Subjects viewed the stimuli on 667-MHz Pentium III-class IBM-PC compatible computers with monitors, keyboards, mice, and hard disk drives.These machines ran Windows 2000 and were situated at separate desks seated in rows. Each computer had a 17- inch color monitor with true color (24 bit) screen resolution and the screen area set at 1024  768 pixels. Subjects viewed the Web sites under Internet Explorer 5. The navigational functions of the Web browser had been disabled so that subjects could not access any Web site other than the experimental ones. After browsing through the site using a mouse, subjects filled out a questionnaire that contained several attitude scales and manipulation checks. By providing their ID number, subjects could be matched to the trait scores from the first stage. The entire procedure took less than 20 minutes to complete.

Measures

Need for cognition was measured with the use of the 18-item scale of Cacioppo, Petty, and Kao (1984). This consists of 18 items, such as “I would prefer complex to simple problems,” of which half are reverse scored (Cronbach’s alpha 0.86). For sensation seeking, subjects completed the 40-item scale of Steenkamp and Baumgartner (1992). Responses to items such as “I like to explore a strange city or section of town by myself” were scored on a 7-point scale anchored by extremely uncharacteristic–extremely characteristic (Cronbach’s alpha 0.87). For manipulation checks, two 7-point scales anchored by difficult to understand–easy to understand and complicated–simple for both visual (r .63) and verbal complexity (r .64) were presented. Attitude toward the Web site, brand attitudes, and purchase intentions were selected as the dependent measures. Attitude toward the Web site (Aws) and brand attitudes (Ab) were rated on four 7-point items anchored by very bad–very good, unfavorable–favorable, unappealing–appealing, not at all likeable–very likeable (Cronbach’s alpha Aws 0.93, Ab 0.88). Purchase intentions were rated on two 7-point items anchored by r(r .93; items from Miniard, Bhatla, Lord, Dickson, & Unnava, 1991).

RESULTS

Manipulation Checks 

A manipulation check for Web-site complexity was conducted by summating and averaging the two 7-point visual-complexity and two 7-point verbal-complexity items that were identical measures to those used in the pretest to form a composite measure (Cronbach’s alpha 0.71). An ANOVA revealed significant differences between the treatments [F(3,108) 2.54, p .05] with subjects rating the low complexity treatment (i.e., low verbal–low visual site) as being the least complex of the four treatments (M 5.62, SD 0.93). This was followed by the medium-complexity treatments of low verbal–high visual (M 5.02, SD 1.12) and high verbal–low visual (M 4.78, SD 0.74). The high complexity treatment (i.e., high verbal–high visual site) received the highest complexity ratings (M 4.52, SD 0.62), suggesting that the manipulation was successful.

Resource-matching Theory 

H1 predicted that Web sites of medium complexity are more persuasive than those of very low or very high complexity. Consistent with this expectation, the 2 (verbal complexity: high vs. low)  2 (visual complexity: high vs. low) between-subjects multivariate analysis of variance (MANOVA) revealed a significant verbal  visual interaction effect [F(3,108) 6.81, p .001, Wilks’s lambda 0.84] across all dependent variables. Specifically, in order of effect size, brand attitudes [F(1,108) 5.46, p .01, ω2 0.11, see Figure 1], purchase intentions [F(1,108) 4.77, p .01, ω2 0.09] and Web-site attitudes [F(1,108) 3.10, p .03, ω2 .05]. As expected, the main effects for verbal and visual complexity were not significant (Fs 1). Table 1 displays the means across each treatment and dependent variable.

Sensation Seeking 

For H2a and H2b, trait data were split into high and low categories with the use of median splitting.Although this results in lost data, this approach follows past research that has examined high and low categories of these traits (e.g., Batra & Stayman, 1990; Schoenbachler & Whittler, 1996). It was expected that sensation seeking would interact with visual complexity but not verbal complexity. High sensation seekers should favor Web sites of high visual complexity, and low sensation seekers should prefer simple graphics.

As expected, the 2 (sensation seeking: high, low)  2 (visual complexity: high, low) MANOVA revealed a significant interaction [F(3,106) 5.29, p .01, Wilks’s lambda 0.87) across all three dependent vari-

ables: brand attitudes [F(1,106) 7.91, p .01, ω2 0.06], Web-site attitudes [F(1,106) 15.29, p .001, ω2 .12], and, to a lesser degree, purchase intentions (p  .06,ω2 .02).Table 2 shows that high sensation seekers were more persuaded by Web sites that contained complex graphics and that they disliked simple graphics, whereas the reverse was true for low sensation seekers. Next, planned contrasts were performed within each of the four experimental cells between high and low sensation seekers. As anticipated, the sensation seeking–visual complexity interaction was only significant in conditions of medium complexity. High sensation seekers in the high visual–low verbal complexity condition expressed more favorable brand attitudes (Ms 5.46 vs. 4.80 for high and low sensation seekers, respectively; t –2.05, p .05, r .37) than low sensation seekers, but not for Web-site attitudes or purchase intentions (ps  .09), which were correlated across all conditions (r .43, p .001). Conversely, stronger effect sizes were evident for low sensation seekers. Specifically, the low visual–high verbal complexity condition revealed more positive brand attitudes (Ms 5.56 vs. 4.92 for low and high sensation seekers, respectively; t 2.57, p .02, r .44) and Web-site attitudes (Ms 5.68 vs. 4.83, t 2.91, p .01, r .48) for low sensation seekers. The other results were not significant, with the exception of the high visual–high verbal complexity condition, where high sensation seekers rated more favorable purchase intentions relative to low sensation seekers (Ms 4.91 vs. 3.73, t –2.28, p .05, r .40).

Need for Cognition

As anticipated, for H3a and H3b, a 2 (NFC: high, low)  2 (visual complexity: high, low) MANOVA was not significant [F(3,106) 1, p  .62]. Further, a 2 (NFC)  2 (verbal complexity) MANOVA revealed a significant interaction effect [F(3,106) 4.64, p .01, Wilks’s lambda .88]. The main effects for NFC and verbal complexity did not reach significance (F 1). Table 3 displays means for NFC across the verbal treat

ments. This shows that high NFC subjects preferred complex verbal layouts. On the other hand, low NFC subjects preferred simple verbal layouts to complex ones. As expected, planned contrasts revealed that high NFC subjects in the low visual–high verbal complexity condition rated more positive brand attitudes (Ms 5.57 vs. 4.90 for high and low NFC, respectively;t –2.74, p .01, r .46) and Web-site attitudes (Ms 5.65 vs. 4.88 for high and low NFC, respectively; t –2.61, p .02, r .44) than low NFC subjects, although this effect was not significant for purchase intentions (p  .35). In addition, in the low verbal–high visual complexity condition, the more favorable ratings for low NFC subjects were not significant (ps  .09). However, low NFC subjects in the low visual–low verbal condition expressed more favorable Web-site attitudes than high NFC subjects (Ms 5.09 vs. 4.35 for low and high NFC; t 2.08, p .05, r .41).

 DISCUSSION 

The data from this study lend support for resource-matching theory. Specifically,Web sites of medium complexity are evaluated more favorably than those of low or high complexity. Although this effect is small, in relation to the guidelines for effect sizes of Cohen (1977) for ω2, it was observed for all three dependent variables. Moreover, Fern and Monroe (1996) suggest that for experimental research, interactions typically produce smaller effect sizes. Presumably, subjects in the low or high complexity conditions were either bored or overwhelmed by the cognitive demands of the Web sites.

For the traits, the anticipated interaction between sensation seeking and visual complexity was confirmed for two of the dependent variables, brand attitudes, and attitudes toward the Web site. High sensation seekers expressed more favorable brand attitudes when exposed to complex visual.

elements relative to simple ones. Stronger effect sizes were evident for low sensation seekers, who exhibited the reverse pattern in support of H2b across brand attitudes and purchase intentions. As expected, this interaction was only significant when the cognitive demands of the Web site were at a medium level of complexity and was not significant for either of the extreme treatments. Thus, the stimulation provided by the low complexity version (i.e., low verbal–low visual complexity) was too low even for low sensation seekers, so that both groups evaluated the site less favorably. Likewise, the high complexity treatment tended to provide too much stimulation for either group.

The predicted effects did not emerge for the last dependent variable, purchase intentions. This may have been caused by the product category itself. Sensation seeking is significantly related to early stages of alcohol and drug abuse (Zuckerman, 1983). Although not as severe as drugs, energy drinks are a product designed to increase one’s energy level or wakefulness. Since low sensation seekers are arousal avoiders, they may be less likely to buy this class of product (Zuckerman, 1983). Hence, whereas they may have liked the Web site and the brand, the actual inclination to purchase was low. This rationale also helps to explain an unexpected effect that was significant. High sensation seekers in the high complexity treatment expressed greater purchase intentions than low sensation seekers. High sensation seekers may have associated the high level of arousal they experienced during Web-site exposure with the product. Given that high sensation seekers are more prone to use substances or drinks to increase sensory arousal levels (Zuckerman, 1994), this may have enhanced their purchase intentions.

As anticipated for H3a, high NFC subjects expressed more favorable attitudes toward a Web site that combined complex verbal with simple visual elements, relative to low NFC subjects. However, this effect did not materialize for purchase intentions. This may have occurred because the demand characteristics of a laboratory experiment may have prompted both high and low NFC subjects to engage in more detailed processing, despite their inherent trait predispositions.

Contrary to expectations for H3b, low NFC subjects did not evaluate a Web site with high visual and low verbal complexity more favorably than high NFC subjects. This suggests that high NFC subjects found this condition equally persuasive. From a resource-matching perspective, the less relevant peripheral visual cues may have influenced high NFC evaluations. Specifically, cognitive resources that high NFC subjects had available for Web-site processing may have been greater than that needed to process the verbal information (Peracchio & Meyers-Levy, 1997). Hence, although they are predisposed toward verbal information (Cacioppo & Petty, 1982), high NFC subjects may have utilized the visual stimuli as a central cue to aid their evaluations, rather than simply disregarding them in favor of verbal information (Meyers-Levy & Peracchio, 1995). However, a significant medium-sized effect (Sawyer & Ball, 1981) emerged in the low complexity treatment, where low NFC expressed more positive site attitudes than high NFC. This low verbal–low visual preference by low NFC subjects would appear intuitive, because this condition was the least difficult to process and should thus be preferred by people who generally avoid demanding cognitive stimuli.

Given the results of this study, an interesting avenue for future research would be to consider the potential interactive effects of complexity and download delay. Although this study suggests that medium complexity can enhance consumer attitudes, recent research on temporal effects suggests that longer download times can negatively influence attitudes (see Davis & Hantula, 2001; DiClemente & Hantula, 2003; Rajala & Hantula, 2000). Because greater complexity (e.g., more visual information) can be associated with more information to download, it would be useful to examine if and how consumers trade off between downloading time and levels of Web-site complexity.

REFERENCES

Ahlering, R. F. (1987). Need for cognition, attitudes, and the 1984 presidential election. Journal of Research in Personality, 21, 100–102. Batra, R., & Stayman, D. M. (1990). The role of mood in advertising effectiveness. Journal of Consumer Research, 17, 203–214. Berlyne, D. E. (1971). Aesthetics and psychobiology. New York: Appleton-Century-Crofts. Burnkrant, R. E., & Unnava, H. R. (1995). Effects of self-referencing on persuasion. Journal of Consumer Research, 22, 17–26. Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social Psychology, 42, 116–131. Cacioppo, J. T., Petty, R. E., Feinstein, J. A., & Jarvis, W. B. G. (1996). Dispositional differences in cognitive motivation: The life and times of individuals varying in need for cognition. Psychological Bulletin, 119, 197–253. Cacioppo, J. T., Petty, R. E., & Kao, C. F. (1984). The efficient assessment of need for cognition. Journal of Personality Assessment, 48, 306–307. Cacioppo, J. T., Petty, R. E., Kao, C. F., & Rodriguez, R. (1986). Central and peripheral routes to persuasion: An individual difference perspective. Journal of Personality and Social Psychology, 51, 1032–1043. Cohen, J. (1977). Statistical power analysis for the behavioral sciences. New York: Academic Press. Davis, E. S., & Hantula, D. A. (2001). The effects of download delay on performance and end-user satisfaction in an Internet tutorial. Computers in Human Behavior, 17, 249–268. DiClemente, D. F., & Hantula, D. A. (2003). Optimal foraging online: Increasing sensitivity to delay. Psychology & Marketing, 20, 785–809. Dreze, X., & Zufryden, F. (1997). Testing Web site design and promotional content. Journal of Advertising Research, 37, 77–91. Fern, E. F., & Monroe, K. B. (1996). Effect-size estimates: Issues and problems in interpretation. Journal of Consumer Research, 23, 89–105.Furnham, A., & Bunyan, M. (1988). Personality and art preferences. European Journal of Personality, 2, 67–74. Gardner, M. P. (1985). Does attitude toward the ad affect brand attitude under a brand evaluation set? Journal of Marketing Research, 22, 192–198. Haugtvedt, C. P., Petty, R. E., & Cacioppo, J. T. (1992). Need for cognition and advertising. Journal of Consumer Research, 37, 77–91. Heckler, S. E., Childers, T. L., & Houston, M. J. (1993). On the construct validity of the SOP scale. Journal of Mental Imagery, 17, 119–132. Hofacker, C. F., & Murphy, J. (2000). Clickable World Wide Web banner ads and content sites. Journal of Interactive Marketing, 14, 49–59. Hoffman, D. L., & Novak, T. P. (1996). Marketing in hypermedia computermediated environments: Conceptual foundations. Journal of Marketing, 60, 50–68. LaBarbera, P. (1999). Marketing on the Internet: The role of cognitive style in user attitudes and behaviour. In Proceedings of 28th European Marketing Academy Conference, Berlin (CD-ROM). Berlin: Humboldt University. Maes, P. (2000). Smart commerce: The future of intelligent agents in cyberspace. Journal of Interactive Marketing, 13, 66–76. Meyers-Levy, J., & Peracchio, L. A. (1995). Understanding the effects of color: How the correspondence between available and required resources affects attitudes. Journal of Consumer Research, 22, 121–138. Miniard, P. W., Bhatla, S., Lord, K. R., Dickson, P. R., & Unnava, H. R. (1991). Picture-based persuasion processes and the moderating role of involvement. Journal of Consumer Research, 18, 92–107. Mitchell, A. A. (1986). The effect of verbal and visual components of advertisements on brand attitudes and attitude toward the advertisement. Journal of Consumer Research, 13, 12–24. Mittal, B. (1995). A comparative analysis of four scales of consumer involvement. Psychology & Marketing, 12, 663–682. Oliver, R. L., & Bearden, W. O. (1985). Crossover effects in the theory of reasoned action: A moderating influence attempt. Journal of Consumer Research, 12, 324–340. Peracchio, L. A., & Meyers-Levy, J. (1997). Evaluating persuasion-enhancing techniques from a resource-matching perspective. Journal of Consumer Research, 24, 178–191. Quinn, C. (1999). How leading edge companies are marketing, selling, and fulfilling over the Internet. Journal of Interactive Marketing, 13, 39–50. Rajala, A. K., & Hantula, D. A. (2000). Towards a behavioral ecology of consumption: Delay reduction effects on foraging in a simulated online mall. Managerial and Decision Economics, 21, 145–158. Raju, P. S. (1980). Optimum stimulation level: Its relationship to personality, demographics, and exploratory behavior. Journal of Consumer Research, 7, 272–282. Sawyer, A. G., & Ball, A. D. (1981). Statistical power and effect size in marketing research. Journal of Marketing Research, 28, 275–290. Schoenbachler, D. D., & Whittler, T. E. (1996). Adolescent processing of social and physical threat communications. Journal of Advertising, 25, 37–54. Steenkamp, J. B. E., & Baumgartner, H. (1992). The role of optimum stimulation level in exploratory consumer behavior. Journal of Consumer Research, 19, 434–448.Zhang, P. (2000). The effects of animation on information seeking performance on the World Wide Web: Securing attention or interfering with primary tasks. Journal of the Association for Information Systems, 1, 1–19. Zhang, Y. (1996). Responses to humorous advertising: The moderating effect of need for cognition. Journal of Advertising, 25, 15–32. Zuckerman, M. (1983). Sensation seeking: The initial motive for drug abuse. In E. H. Gottheil, K. A. Druley, T. E. Skoloda, & H. M. Waxman (Eds.), Etiological aspects of alcohol and drug abuse (pp. 202–220). Springfield, IL: Charles C. Thomas. Zuckerman, M. (1994). Behavioral expressions and biosocial bases of sensation seeking. New York: Cambridge University Press. Zuckerman, M., Bone, R. M., Neary, R., Mangelsdorf, D., & Brustman, B. (1972). What is the sensation seeker? Personality trait and experience correlates of the sensation-seeking scales. Journal of Consulting & Clinical Psychology, 39, 308–321. Zuckerman, M., Ulrich, R. S., & McLaughlin, J. (1993). Sensation seeking and reactions to nature paintings. Personality & Individual Differences, 15, 563–576. Correspondence regarding this article should be sent to: Brett Martin, Department of Marketing, University of Auckland Business School, Private Bag 92019, Auckland, New Zealand (bas.martin@auckland.ac.nz).

 

The Influence of Gender on Mood Effects in Advertising

Pretest 1: Mood InductionBrett A. S. Martin University of Auckland

ABSTRACT The main objective of this article is to study the impact of gender on mood effects in relation to attitude toward the ad and brand attitudes. Specifically, gender, mood state, and ad affective tone are posited to interact. Data from an experiment support two hypotheses predicting the most favorable combinations of mood and affective tone for males and females for attitude toward the ad. Findings also support previous research that female gender and sad moods, respectively, result in more detailed processing. Limitations and future research directions are discussed.

An appreciable amount of consumer research has investigated how moods influence consumers. Studies have explored how moods impact on recall (Lee & Sternthal, 1999), shopping intentions (Swinyard, 1993), the amount of cognitive elaboration engaged in by consumers (Batra & Stayman, 1990), and evaluations of brand extensions (Barone, Miniard, & Romeo, 2000), advertisements (Goldberg & Gorn, 1987), and music (Holbrook& Gardner, 2000). Yet although recent psychological research has examined the impact of gender and affect (e.g., Cheng, 1999; Oliver, Sargent & Weaver, 1998; Seidlitz & Diener, 1998), consumer research regarding a gender influence on mood effects has been comparatively neglected (see Kellaris & Mantel, 1994, for an exception). This is surprising given that gender differences have been found in consumer information-processing strategies (Meyers-Levy & Maheswaran, 1991; Meyers-Levy & Sternthal, 1991). Indeed, it is the contention of this article that the predictions of mood research and gender research regarding depth of processing offer valuable insights into mood effects. Furthermore, Stern (1993) suggests that because of inherent predispositional characteristics, gender could influence mood responses, and she advocates the need for research in this area. “The mood variable then, requires additional research in relation to the gender of the consumer . . . and interaction effects not yet considered” (Stern, 1993, p. 562). Rusting (1999) has also advocated the need to study invariant characteristics of people to gain insight into mood states.

To this end, the current study examines the impact of gender, mood, and ad affective tone on consumer attitudes. Specifically, the structure of this article is as follows. First, mood theory will be addressed. Second, gender theory will be discussed. Third, research hypotheses, method, and results will be outlined. Finally, the results will be discussed, with limitations and future research directions presented.

CONCEPTUAL BACKGROUND

Mood Defined

Mood is defined as a consumer’s affective state that is relatively global in nature, as opposed to emotions, which tend to have a specific cause (Gardner, 1985; Luomala & Laaksonen, 2000; Rusting, 1998). Ad affective tone is defined as the affective valence of the content of the advertisement (Kamins, Marks, & Skinner, 1991). Within the field of mood research, a variety of moods are available for study. For example, in the context of negative moods, researchers have called attention to sad moods (Rusting & DeHart, 2000), anxious moods (Thayer, Newman, & McClain, 1994), and angry moods (Rusting, 1998; Sedikides, 1995). Recently, consumer research scholars have compared positive and neutral moods (e.g., Barone et al., 2000; Lee & Sternthal, 1999; Meloy, 2000). However, the present study examines positive and negative moods, specifically, happy versus sad moods.

The rationale for choosing these mood states is as follows. First, from a theoretical perspective, much of the research in this area draws upon the mood-congruency hypothesis, which is derived from an associative networkmodel of memory (Bower, 1981). This model posits that mood states prime the recall of memories of a similar affective valence. Positive moods prime positive memories; negative moods, negative memories. A substantial amount of research supports this notion that mood states influence judgments in a mood-congruent manner. Positive moods result in more favorable evaluations, whereas negative moods result in more negative evaluations (see Blaney, 1986; Forgas, 1992, 1995; Gardner, 1985; and Luomala & Laaksonen, 2000 for reviews). However, some studies have also found mood-incongruent results where negative moods result in favorable evaluations (e.g., Erber & Erber, 1994; Rusting & DeHart, 2000).

Recent research suggests that the role of negative moods is unclear in terms of when and why mood-congruent or mood-incongruent effects will occur (e.g., Rusting, 1998; Rusting & DeHart, 2000). Consequently, the study of happy and sad moods remains a topic of key concern to researchers (e.g., Larsen, McGraw, & Cacioppo, 2001; Park& Banaji, 2000; Wood, Michela, & Giordano, 2000). Thus, studying these mood states offers scope to contribute to the understanding of mood regulation, regarding the improvement of a sad mood to a happier mood state.

Second, from a promotional viewpoint, in the context of television advertising, which this study addresses, studying happy and sad moods is relevant. This is especially so since research suggests that viewing films and television programs can influence mood states (Curren & Harich, 1994; Gerrards-Hesse, Spies, & Hesse, 1994). Obviously a number of situations exist where consumers may be exposed to advertising when in a happy-mood state, such as watching a favorite television comedy. However, one can also expect that instances will occur when consumers will be exposed to advertising when in a sad-mood state. Consider, for instance, when advertising appears after a news item on some tragic event. Similarly, in this study the sad-mood manipulation involves a documentary where people speakabout relatives that have since died of cancer. By looking at sad-mood states, we not only address theoretical issues, but also provide insight into the viewing experience of consumers. Hence, this study examines mood effects in the context of happy and sad program-induced moods and the affective tone of the ads.

Mood and Information Processing

A substantial amount of research suggests that happy moods result in heuristic processing, whereas sad moods result in more effortful processing (see Bagozzi, Gopinath, & Nyer, 1999; Clore, Schwarz, & Conway, 1994 for reviews). Two perspectives support this view. The first suggests that moods influence cognitive capacity, where happy moods result in reduced capacity owing to their distracting nature (Mackie & Worth, 1989). Happy people are believed to prime a variety of memories, as positive concepts are posited to be more highly interconnected in memory (Isen, 1984; Mackie & Worth, 1989). Therefore, happy people have less cognitive capacity available for processing, which results in a greater likelihood of heuristic processing.

The other perspective, the mood-as-information model (Schwarz, 1990), also predicts heuristic processing for happy moods, but adopts a motivational rather than a capacity rationale. Here, moods indicate whether it is necessary to use heuristic or more detailed processing. Happiness indicates a pleasant environment, where there is little need for elaborative processing, unless called for by other goals (Bless,Schwarz, & Wieland, 1996). Thus, heuristic processing is likely. By contrast, sadness suggests a more problematic environment, which may need to be contemplated, which, in turn, leads to more elaborative processing. This approach is based on the mood-maintenance hypothesis that suggests people are generally motivated to maintain positive moods, and to repair negative moods (Isen, 1987; Morris & Reilly, 1987). When engaging in mood repair, people improve their moods by focusing on pleasant thoughts or memories (Rusting & DeHart, 2000). Thus, both models—cognitive capacity and mood-as-information—suggest that people engage in heuristic processing during happy moods, and more effortful processing during sad moods.

It is noted that although this is the dominant view in the literature, this perspective has been challenged by researchers such as Matmur and Chattopadhayay (1991), who found that positive moods actually increased the elaborative processing of happy ads. A variety of explanations have been proposed for such findings. These include viewing such results as reflecting (a) automatic affect-priming processes, which are not present for self-regulating motivational strategies (Luomala & Laaksonen, 2000), (b) the forced exposure nature of experiments, and (c) stimuli argument quality (Schwarz, Bless, & Bohner, 1991). Indeed some researchers have even suggested that negative moods can reduce processing capacity through the encroaching of negative thoughts during processing (e.g., Ellis & Ashbrook, 1988). However, this study will adopt the dominant view that happy program-induced moods enhance the use of less cognitively effortful processing strategies.

Gender and Information Processing

Another factor that can influence information processing is gender. Although one might assume that males and females process information in an equivalent manner, research reveals that substantial gender differences do exist. For example, females appear to have a superior ability in correctly recalling tasksequences (Nicholson & Kimura, 1996), in object recognition from studying visual stimuli (Harshman, Hampson, & Berenbaum, 1983), and have been found to perform better on colornaming tasks (Saucier, Elias, & Nylen, 2002). Further, research in psychophysiology suggests that for negative moods, males and females differ in terms of what parts of the brain are activated (i.e., frontal electroencephalographic activation) when judgments are being made (Blackhart, Kilne, Donohue, LaRowe, & Joiner, 2001). Current research in this field indicates that sex hormones may influence the development and function of the brain, thereby influencing processing (e.g., Duff & Hampson, 2001; Pogun, 2001). Indeed Pogun (2001, p. 205), has recently stated: “The differences between males and females transcend reproductive functions, are evident in the structural and functional organization of the brain, and are reflected in cognitive abilities and behavior.”Insights into these cognitive differences in a marketing context have been provided by Meyers-Levy (e.g., Meyers-Levy, 1988; Meyers-Levy & Maheswaran, 1991). In particular, Meyers-Levy and Sternthal (1991) found gender differences in levels of cognitive elaboration. Their research suggests that females have lower elaboration thresholds whereby they engage in detailed processing more readily than males. Males limit the cognitive effort they expend, and thus use heuristic processing, whereas females prefer more detailed processing. However, these differences disappear when males are sufficiently motivated (Dube & Morgan, 1996; Meyers-Levy & Maheswaran, 1991).

Research in psychology tends to offer support for this gender difference in processing. For example Seidlitz and Deiner (1998), in a study of the recall of affectively valenced life events, attribute the superior recall of females to engaging in more detailed processing than males. Psychophysiological research also provides insights. Duff and Hampson (2001), in a study of prefrontal cortex differences, found for a series of working memory tasks that females made fewer errors and needed less time to perform the tasks. This was despite there being no differences in the sample for intellectual ability and attention. Working memory was defined by these authors as the ability to store and manipulate data on-line for cognitive tasks. The results were interpreted as evidence for gender differences in working memory that could influence processing. In a similar vein, other research suggests that females prefer a sequential, elaborative strategy, whereas males prefer an impulsive, global strategy to cognitive processing (e.g., Klinteberg, Levander, & Schalling, 1987; Pogun, 2001).

Overall then, gender research tends to suggest that females have a greater propensity for detailed processing, whereas males tend toward more heuristic processing. However, this view offers contradictions with mood research and the influence of happy and sad moods. In other words, if females are detailed processors and males are heuristic processors, what happens when these inclinations conflict with the influence of their mood state? For example, do happy females engage in detailed processing, as is suggested by gender research, or the heuristic processing of happy-mood research? What about sad males?

Gender and Mood

To resolve this issue, one needs to consider the nature of the moods and compare them to gender predispositions. A key feature of mood states is their transient temporal nature (Luomala & Laaksonen, 2000) and their ability to greatly affect a person’s perspective and judgments. For instance, affect-priming theory suggests mood congruency where, as noted, mood states prime the recall of memories of a similar affective valence. Hence, although a mood persists, it has the potential to not only influence the amount of elaboration, but also the valence of the material accessed from memory that can be used as an input for making judgments.

By contrast, gender differences in processing reflect inherent characteristics that may be overridden by situational factors, such as the nature of the ad stimulus or the motivational state of the subject (Meyers-Levy & Maheswaran, 1991). Similarly, Tice, Bratslavsky, and Baumeister (2001), in a study of impulse control, suggest that negative moods generate a motivation to feel better that overrides internal restraints. Priority is given to mood regulation ahead of the long-term focus required for self-denial. An analogy here can be drawn from research on personality traits, where a central perspective is that the in- fluence of traits can be overridden by short-term states. For example, someone may be low in need for cognition (i.e., the trait, Cacioppo & Petty, 1982) but when exposed to the stimuli of interest, they may experience a high-involvement processing state.

Alternatively, someone may be high in positive affectivity (i.e., an optimist, Watson, Clark, & Carey, 1988), but owing to recent events in their life, may experience a negative mood state, and hence a more negative view of the world. Likewise, it is suggested here that mood states can override gender predispositions given the personal idiosyncrasies of the consumer being studied where those predictions conflict.

 

Research Hypotheses

Overall then, because happy moods result in less effortful processing, and males are predisposed to heuristic processing, a happy mood should cause them to engage in a minimalist level of processing. According to Forgas (1992, 1995), four processing strategies exist for the influence of affect: direct-access, heuristic, substantive, and motivated strategies. Of these, direct-access strategies represent the lowest form of effort minimalization, even below heuristic processing. The judge is uninvolved, there is no motivation, and the target (i.e., ad in this instance), has prototypical features. For effort minimalizers, this represents a preferred strategy (Forgas, 1995) where a preexisting evaluation is used for prototypical stimuli. Hence, happy males should exhibit no difference in preferences for happy and sad ads.1 Under sad moods, males will have their level of mental effort raised, and will thus more closely attend the nature of the ads. To this end, it is posited that sad males will prefer ads with happy content over ads with sad content. The rationale for this prediction is twofold. First, mood-repair research suggests a gender difference in how males and females attain a positive mood state when in a negative mood. This research indicates that males achieve mood repair in sad moods by using a distraction strategy, where they prefer to concentrate on something other than the cause of their mood to distract themselves, and thereby elevate their mood state (e.g., Nolen-Hoeksema, Larson, & Grayson, 1999; Sethi & Nolen-Hoeksema, 1997). Hence, given this gender difference, sad males should prefer the happy ad as a convenient means of repairing their mood. These gender–mood effects should have a direct effect upon attitude toward the ad (Aad), which is defined as the affective reactions to an ad (Brown & Stayman, 1992; MacKenzie, Lutz, & Belch, 1986).

From a theoretical perspective, mood has been posited in models to act as an affective antecedent that impacts directly upon Aad (e.g., MacKenzie & Lutz, 1989). Mood at the time of ad exposure is theorized as being itself a product of variables such as stimuli reception context and individual differences. This mood in turn influences a consumer’s attitude toward the ad. Likewise, the Murry, Lastovicka, and Singh (1992) model of the influence of television program-induced affect suggest that moods have a direct effect upon Aad. This discussion leads to the following hypotheses:

H1: (a) Males, under happy-mood conditions, will exhibit equal attitude toward the ad (Aad) for happy or sad affectively toned ads.

H1: (b) Under sad moods, males will exhibit more favorable attitudes (Aad) for happy ads than for sad ads.

For females, when they are in a happy mood, which discourages extensive processing, they should engage in less effortful heuristic processing rather than the minimalist direct access strategy of males. For heuristic processing, simple decision rules are used for judgments (Forgas, 1995). Yet in contrast to direct access strategies, an attempt is made to formulate an evaluation. Previous research suggests that under happy-mood heuristic processing, consumer evaluations are influenced in a manner congruent to simple valence cues, such as the affective tone of the ad (e.g., Kamins et al., 1991). Therefore, ad affective tone should act as a cue for attitudinal responses so that they prefer happy ads to sad ads.

However, when in a sad mood, which encourages more mental effort, females as detailed, comprehensive processors (Kiecker, Palan, & Areni, 2000), should process the happy and sad ads extensively. Thus, unlike sad males, who will prefer the happy ad as a convenient means of distraction, sad females will attain mood repair from viewing either the happy or sad ads. For these ads, the comprehensive processing of females should mean that the advantages offered by the products in both ads should be equally apparent, regardless of ad valence. For the happy ad, the positive nature of the ad will make these benefits readily apparent. Equally, for the sad ad, the detailed processing of females should mean that they comprehend how the negative events in the ad can be remedied by the advertised product. Therefore, unlike sad males, who should prefer to distract themselves with the happy ad, sad females should regard happy and sad ads equally favorably. On the basis of this discussion, the following hypotheses are offered for testing:

H2: (a) Females, under happy-mood conditions, will exhibit more favorable attitude toward the ad (Aad) for happy ads, than for sad ads. H2: (b) Under sad moods, females will exhibit equal preference (Aad) for happy or sad affectively toned ads.

Brand attitudes (Ab) were also measured in the present study, as research on affect typically includes Aad and Ab as dependent measures (Brown, Homer, & Inman, 1998), and a variety of empirical research has shown that Aad acts as an antecedent for Ab (e.g., Laczniak& Muehling, 1993; MacKenzie et al., 1986; Mitchell & Olson, 1981). Although some studies have shown that mood effects can influence Ab for evaluations of low personal relevance (Batra & Stephens, 1994; Curren & Harich, 1994)—presumably owing to an affect-as-information cue bias where mood-state valence is treated as a cue for low-effort judgments (Forgas, 1995)—a stream of empirical research indicates that the effect of feelings is direct upon Aad but indirect upon Ab (e.g., Brown et al., 1998; Burke & Edell, 1989). Aad mediates the effect of feelings on Ab. This effect has been found to be stronger and more consistent than the effect on Ab (Murry et al., 1992). Theoretical models of the influence of mood upon Aad also indicate a direct effect upon Aad and indirect effect on Ab (e.g., Mackenzie & Lutz, 1989). Consequently it posited that the gender–mood effects have a direct influence upon Aad but not Ab. This leads to the following hypothesis:

H3: The interaction of gender and mood will be evident for Aad but not for Ab. To summarize, it is expected that mood, ad affective tone, and gender interact and influence consumer attitudes.

PRETESTING OF TREATMENTS

Pretest 1: Mood Induction

The first pretest sought to identify happy and sad mood-inducing television programs for the main study. Gerrards-Hesse et al. (1994), in areview of close to 250 psychological studies, concluded that the film-clip mood-induction procedure represents the only method that is equally suited for the induction of happy and sad moods. For happy moods, the programs tested were The Simpsons, an animated situation comedy featuring the humorous antics of Homer Simpson, Marge Simpson and their children, and Seinfeld, a situation comedy featuring comedian Jerry Seinfeld. For sad moods, the programs were Caraline’s Story, a documentary item on a young woman who suffers from anorexia nervosa resulting from childhood abuse, and A Lot of Love, A Lot of Pain, a networkdocumentary item on child cancer, featuring interviews with grieving parents and home videos of the children they have lost.Eighty-three subjects, participating in groups of 17–25 subjects, then rated one of the programs on a 7-point mood scale (1  happy, 7  sad), followed by six items on the extent to which they experienced certain feelings when they viewed the program. Three were filler items, the other three measuring anger, derived from Izard’s Differential Emotions Scale (Izard, Dougherty, Bloxom, & Kotsch, 1974).2 That is, enraged, angry, and mad (1  not at all, 7  to a great extent). As a principal-axis factor analysis revealed that the three anger items loaded strongly on a single factor and formed reliable scales (Cronbach’s alpha  0.92), an anger index was created for the analysis. A significant difference was found in mood scores (t  14.56, p  .0001), with the happy programs (M  2.37) rated as significantly happier than sad programs (M  5.38). Although no significant difference existed between Seinfeld and The Simpsons (t  1.26, p .21), a judgment decision was made to select The Simpsons (M  2.24) as the relevant happy program for the main study. For sad programs there was no significant difference in mood scores (t  0.11; p .91). However, Caraline’s Story rated significantly higher than A Lot of Love, A Lot of Pain on the anger index (Ms  10.65 vs. 5.12, respectively, t  4.50, p  .0001). No gender differences were present in the mood ratings (Fs 1). Thus, The Simpsons was selected for happy-mood induction, and A Lot of Love, A Lot of Pain was selected for the sad-mood induction.

Pretest 2: Affective Tone

A second pretest was conducted to ensure selected ads had the appropriate affective tone. One hundred thirty-five subjects viewed five ads presented in one of three sequences to test for order effects. The ads presented the same number and type of benefits and were selected for pretesting from a pool of 43 (generously provided by Saatchi & Saatchi). After exposure, subjects rated the ad’s affective tone on a 7-point scale (1  happy, 7  sad), as well as liking (1  dislike very much, 7  like very much), interest (1  interesting, 7  boring) and attitude (1  good, 7  bad). A significant treatment effect for ad type was evident across the dependent variables (F 3.51, p .02, 2  0.07) with the dancing ad (young people dance, chat with friends, and wave to the camera), and rain ad (a morose young woman is jostled in a rain-swept street, while having flashbacks of a man who is presumably her ex-boyfriend) selected for the study. No main effects or interactions for gender or order were evident across the dependent variables (Fs 2.28).

METHOD

 

Subjects and Product Context

Two hundred eighty-two undergraduate business students participated in the study (165 females, 117 males). Subjects participated in groups of 8–18, with groups randomly assigned to treatment conditions. Cellular phones were chosen as the product category based on three criteria:

1. Subject knowledge: The product had to be one subjects were familiar enough with to be able to make a judgment on, to avoid nonsense responses and process the ad information effectively (Homer & Yoon, 1992).

2. Equal Gender Relevance (Gainer, 1993): Ninety-two subjects in an earlier pretest from the same target population rated cell phones on two 7-point bipolar semantic-differential scales, “handheld cellular phones are important to me” and “handheld cellular phones do not have anything to do with me or my needs,” anchored by “strongly agree”– “strongly disagree” (adapted from Celsi & Olson, 1988). No gender differences eventuated (p .21)

. 3. Commercially successful: The cellular market has recently been estimated at over $16 billion (Kupfer, 1999). In 1998, AT&T paid $1.5 billion for the largest independent cellular provider in America (Kupfer, 1998).

 

Procedure

Subjects were told that the purpose of the study was to find out how people evaluate television programs and ads. Next, subjects were asked to seat themselves in a comfortable manner to view the television. The television program was then played. When it was over, subjects completed a scale that measured post mood-induction mood scores (1  Happy, 7  Sad), excluding filler items. Such a global measure of mood is consistent with past research (Abele & Hermer, 1993; Hertel & Fielder, 1994; Hornik, 1993; Kramer, Newton & Pommerenke, 1993; Schwarz et al., 1991). Pursuant to the recommendations of psychometric theory (Paulhus, 1991), filler items were used to disguise the purpose of the questionnaire. The ad was then played, after which subjects completed the remainder of the questionnaire at their own pace. Booklets were collected. Subjects were thanked for their participation. For sadmood subjects, chocolate biscuits were handed out (a potent smile inducer). The entire procedure tookless than 37 minutes to complete.

Measures

Attitude toward the ad (Aad) was rated on four 7-point items anchored by: “bad”– “good,” “uninteresting”– “interesting,” “dislike”– “like,” and “not irritating”– “irritating.” Three 7-point items measured attitude toward the brand (Ab) anchored by: “bad”– “good,”, “unpleasant”– “pleasant,” and “dislike”– “like.” The reliability of these scales was suf- ficiently high (Cronbach’s alpha  0.79 and 0.90 for Aad and Ab, respectively), which is consistent with past research that has used these items (e.g., Yi, 1990, 1993). The questionnaire included two manipulation checks. First, a mood-manipulation check asked subjects to rate how they felt after having watched the program (1  happy, 7  sad). Second, a checkwas performed for affective tone, with subjects rating the extent to which the ad seemed happy or sad (1  happy, 7  sad).

Given previous research on gender and elaboration thresholds (Meyers-Levy & Sternthal, 1991), a measure of consumer involvement with the ad was included as a covariate. The present study utilized Mittal’s (1995) five-item adaptation of Zaichowsky’s (1985) Personal Involvement Inventory (PII), which contained the statement: “For me, the advertisement was…” along with the anchors: “important”– “unimportant,” “of no concern”– “of concern to me,” “means a lot to me”– “means nothing to me,” “matters to me”– “does not matter,” and, “significant”– “insignificant” (Cronbach’s alpha  0.93). This measure was chosen for three reasons: (a) the support for measure unidimensionality and validity presented by Mittal (1995), (b) because this revised scale has been used successfully in recent consumer research (e.g., Dean, 1999; De Wulf, Odekerken-Schroder, & Iacobucci, 2001), and (c) because this scale is more parsimonious and simpler than the 20-item PII. This aided questionnaire length and reduced the potential for nonresponses by subjects.

A measure for message framing was also included. In a comprehensive review of the framing literature, Levin, Schneider, and Gaeth (1998) suggest that when affect and the affective nature of ads are being studied, such as with the ad affective tone of the present study, then the framing valence of ad information should also be considered. Consequently, a two-item framing measure derived from Maheswaran and Meyers-Levy (1990) was included as a covariate (r  0.61, p .01).

RESULTS

Manipulation Checks

Consistent with expectations, subjects in the happy program-induced mood condition reported happier moods (M  2.77) than those in the sad-mood condition (M  4.45, F  117.56, p .0001, 2  0.30). Likewise, a very strong effect was found for affective tone with the content of happy ads rated happier (M  2.64) than sad ads (M  4.20, F  89.59, p  .0001, 2  0.24). This indicates that the intended factors were manipulated successfully. Furthermore, a principal-axis factor analysis performed on all measures that comprised three or more items showed that all measures loaded onto single factors.

Hypothesis Testing

H1(a) posits that happy males will show no preference for a particular ad affective tone type, whereas H1(b) predicts that sad males will exhibit more favorable attitudes for happy ads than for sad ads. H2(a) and H2(b) predict the converse, namely, that happy females will prefer happy ads (H2[a]), whereas sad females will show no particular preference for ad type (H2[b])

To ascertain whether gender interacted with mood state and ad affective tone, a 2 (mood: happy, sad)  2 (ad affective tone: happy, sad)  2 (gender: male, female) MANOVA was performed on Aad and Ab. MANCOVA was not used, as the covariates were either significantly correlated with the independent variable manipulation checkscores (i.e., message framing with affective tone, r  0.43, p  .01), or a significant difference in the covariate existed across levels of treatment variables (i.e., gender differences in involvement F  4.01, p  .0001). Consequently, analysis of variance was deemed more appropriate than analysis of covariance (Huitema, 1980; Wildt & Ahtola, 1978).

This analysis revealed a main effect for affective tone (F  5.54, p  .01, 2  0.02) for Aad, as displayed in Table 1. Overall, subjects rated the happy ads more favorably than the sad ads (Mhappy ad  19.13 vs. Msad ad  17.69). The main effects for mood and gender were not significant (Fs 1). Importantly, the interaction for mood, affective tone, and gender was significant for Aad (F  5.41, p  .02, 2  0.02). Although the guidelines of Cohen (1977) for 2 suggest this is a small

effect, Peterson, Albaum, and Beltrami (1985), in a review of effect sizes in major psychological and marketing journals from 1970 to 1982, found that 62.5% of studies reported a significant effect with an 2 of between 0.01 and 0.09. Furthermore, Iacobucci (1994) suggests that small significant 2s can occur for complex research issues, and Fern and Monroe (1996) assert that large 2s should not be expected for experimental research, and that interactions typically produce smaller effect sizes.

As displayed in Figure 1 and consistent with H1(a) and H1(b), males did not differ in their Aad evaluations of happy and sad ads when they were happy. However, when sad, males clearly favor happy ads. Consistent with H2(a) and H2(b), happy females report more favorable Aad when exposed to happy ads. Yet this difference disappears when females are sad.

These results were verified by a series of planned contrasts (Rosenthal & Rosnow, 1985). Specifically, for males under happy mood, Aad

Figure 1. Plot of the interaction of gender, mood, and affective tone on attitude toward the ad.

scores were not significantly different across the affective tone conditions (p .87). Yet for sad-mood males, as displayed in Figure 1, happy ads rated significantly more favorably than sad ads (t  3.02, p  .004). For females under happy mood, the more favorable ratings for happy ads were marginally significant (t  1.88, p  .065), whereas no differences were found between ads in the sad-mood condition (p .79). Subjects’ means scores and standard deviations for each of these conditions are presented in Table 2.

To further investigate the hypotheses, a 2 (mood: happy, sad)  2 (ad affective tone: happy, sad)  2 (gender: male, female) ANOVA was performed on involvement. If sad moods, as the literature suggests, result in more substantive, elaborative processing, then one should expect a main effect for mood on involvement. Likewise, gender theory suggests females engage in elaborative processing more often than males, which may result in a main effect for gender. Furthermore, under happy mood, it is expected that females will exhibit higher levels of involvement than males, thus suggesting a gender difference may be apparent for happymood data.

Consistent with this view, a significant main effect for mood indicated that subjects’ involvement was higher during sad moods (M  22.71), and lower during happy moods (M  20.06, F  12.84, p  .0001, 2  .04). As displayed in Table 3, a significant main effect for gender was also present (F  4.01, p .05, 2  .01), with females (M  22.07) displaying higher involvement than males (M  20.63). Planned contrasts revealed that no gender differences in levels of involvement were evident under sad moods (p .52), whereas the expected gender difference in involvement for happy mood was marginally significant (t  1.81, p  .07, 2  .02) with happy females (M  21.03) displaying greater involvement than males (M  18.77). Consistent with H3, no main effects (F 2.78, p .09) or interactions (F 2.38, p .12) were statistically significant for Ab. This pattern of null results indicates that the effects of mood and gender influence Aad rather than consumer brand attitudes. Further, consistent with previous research (MacKenzie et al., 1986), Aad and Ab were positively correlated (r  .49, p  .01), suggesting an indirect effect for affective states upon Ab that operated through Aad. DI

DISCUSSION

The objective of this study was to assess whether gender interacted with mood effects, specifically, to examine the interaction of gender, television-program-induced mood states, and ad affective tone in the context of Aad and Ab. The findings yielded support for the hypotheses that sad males would exhibit more favorable Aad evaluations for happy ads, whereas happy males would show no distinction for ad affective tone type for Aad. Likewise, the hypotheses that happy females would prefer happy ads, yet show no preference for ad type when sad, was also supported for Aad. Consistent with expectations, gender–mood effects had a direct effect on Aad but not Ab which supports research suggesting that the influence of affect on Ab is mediated by Aad (Brown et al., 1998).

Findings also supported the mood and gender processing differences that were posited to underlie the Aad findings. Specifically, consistent with previous research (Bagozzi et al., 1999), sad-program-induced moods resulted in higher ad involvement than happy moods. Females also yielded higher involvement than males, which supports previous consumer research (Meyers-Levy & Sternthal, 1991). However, the effect sizes were small. Although a measure of involvement with the ad was used, it would be useful to replicate these findings with a processing measure such as cognitive responses, and recall (see Johar & Simmons, 2000; Meyers-Levy & Sternthal, 1991, for examples of these measures), to see if this improves the explanatory power of the findings.

Furthermore, although the present study provides support for differences in involvement levels for gender–mood effects, it would be of interest to see what sort of processing was occurring. For instance, Grunert (1996) offers two kinds of cognitive processing: automatic processing, which is more subconscious, and strategic processing, which requires more elaboration by the consumer. Which, if either, process is applicable to gender–mood effects? In answer to this question, insight can be gained from the recent comprehensive review of mood research by Luomala and Laaksonen (2000). They suggest that mood studies can be viewed by distinguishing between backdrop moods and motivational moods. Backdrop moods operate at the automatic level, and tend to be global and diffuse. With motivational moods, “individuals are conscious of their mood experiences” (Luomala & Laaksonen, 2000, p. 204), and they are stimulus specific (i.e., people are aware of the cause of their mood). This dichotomy not only aligns with Grunert (1996), but also fits literature on mood repair. For instance, Forgas (1992, 1995) suggests that mood repair follows a motivated processing strategy that is distinct from other less-focused mood strategies.

Following this approach, the results of the present study suggest that males exhibit backdrop mood effects when happy, as characterized by lower involvement, but are in a motivational mood state when in a sad mood. The sad-program-induced mood initiates a desire for mood repair, which is then acted upon by males preferring the happy ad, consistent with research that males prefer distraction mood-repair strategies (Sethi & Nolen-Hoeksema, 1997). Females employ higher-intensity backdrop moods when happy, consistent with mood-congruency theory (Bower, 1981; Luomala & Laaksonen, 2000), and motivated moods when sad. It may be that females have more intense backdrop moods and utilize both happy and sad ads for mood repair under sad moods. Thus, future research should assess the intensity of elaboration that influences backdrop or motivational moods. In summary, this study suggests that in mood research, gender should also be considered.

An alternative explanation of the results should be acknowledged, however, given the idiosyncratic nature of moods. For example, moods can be transient and as fleeting as a few minutes in duration (Luomala & Laaksonen, 2000; Park & Banaji, 2000). This idiosyncrasy has been highlighted by Rusting (1998), who suggests that because the duration and intensity of moods varies between individuals, individual characteristics should be taken into account to yield greater insights. Although the present study suggests that one such individual characteristic— gender— influences mood effects, another characteristic offers an alternative explanation for the results. In particular, it is possible that the mood manipulations not only influenced mood states, but also individual levels of arousal (Clark, 1982; LaTour & Rotfeld, 1997; Mano, 1997).

A significant stream of consumer research has examined the effect of arousal in a persuasive communication context. In these studies,arousal is often highlighted as a dimension of emotion (e.g., Bagozzi, 1996; LaTour & Rotfeld, 1997; Olney, Holbrook, & Batra, 1991). Indeed, in a comprehensive review of affect research, Bagozzi et al. (1999) argue that physiological arousal fulfills an integral role in emotion. Likewise, Sanbonmatsu and Kardes (1988) assert that arousal can be affected by emotionally valenced stimuli.

In the context of the present study, arousal may have influenced the manner of processing used by subjects. Previous research suggests that high arousal can result in a decrease of cognitive effort (Mano, 1992; Sanbonmatsu & Kardes, 1988). Here, high arousal reduces attentional capacity, resulting in a greater reliance on heuristic processing (Mano, 1992, 1997; Pham, 1992). Thus, even though the mood-manipulation checks were significant, it could be that arousal influenced subjects to prefer happy ads owing to heuristic processing. In other words, females were aroused by the happy-mood treatment, whereas males may have been highly aroused by the sad-mood treatment, resulting in both genders preferring happy ads in these conditions.

Although some researchers have suggested that it is unclear that gender differences exist in arousal responses to affective stimuli (e.g., Kring & Gordon, 1998), recent research suggests this avenue is worthy of exploration. For example, physiological arousal, as measured by skin conductance activity, has been shown to be higher for males than females for negative stimuli (e.g., Brewster, Nelson, McCanne, Lucas, & Milner 1998). Consequently one can speculate that males may have been more highly aroused than females by the sad-mood condition, thus resulting in heuristic processing of the happy ad.

Importantly, arousal was not measured in the current study, which represents a limitation that should be recognized. Future research should examine gender and mood effects with arousal measures included to investigate this issue (see LaTour & Rotfeld, 1997, for an example of incorporating arousal into affect research).

A second limitation is the student sample, which restricts the results from being generalized to other populations. Although student samples are prevalent in consumer affect research (e.g., Barone et al., 2000; Holbrook& Gardner, 2000) they can lessen external validity. However, given the range of variables and covariates that were studied, a laboratory experiment was deemed appropriate. Furthermore, college-aged consumers have been identified as a promising segment for cell-phone promotions (Steward, 1995). Second, the use of real ads may have resulted in inferences based on prior ad evaluations (Forgas, 1992). However, in recognition of this issue, foreign ads were utilized. For future research, other affective states could also be studied, such as anger (Sedikides, 1995). From a marketing viewpoint, research of this nature could study angry-mood effects in the interests of customer complaints and retention.

A further limitation is the focus on biological sex. Future research may wish to consider the theory of gender-role socialization of affect (Zillman, Weaver, Mundorf, & Aust, 1986). This predicts that the people are socialized according to traditional cultural gender roles, especially regarding how they publicly display the influence of affect (Basow, 1986; Lott, 1987). Thus, activated sex roles may influence how a consumer responds to ads, whereby males are guided by assertive agentic roles, and females by communal goals (see Meyers-Levy, 1988). To this end, research could consider the role of gender schematicity. Gender schematicity (Bem, 1976; Frable & Bem, 1985) draws upon the bidimensional construct of masculinity and femininity, which are derived from an androgynous model where they represent orthogonal, independent factors (Marusic & Bratko, 1998; Watson, Biderman, & Sawrie, 1994), rather than the bipolarity of a unidimensional factor. In genderschematic terms, individuals can be classified as sex-typed, cross-sextyped or as gender aschematic based upon self-perceived gender-role characteristics. Sex-typed individuals conform to societal gender-role expectations. By contrast, cross-sex-typed individuals present opposite profiles (e.g., a low masculine, highly feminine male), and genderaschematic people are undifferentiated in their role perceptions. It would be interesting to assess what insights could be gained regarding the biological sex differences of the present study when gender schemacity is taken into account.

Another avenue for future research pertains to a strategy employed by people in negative moods—distraction and rumination. A distraction strategy involves avoiding thinking about why one is depressed (Lyubormirsky & Nolen-Hoeksema, 1993). Here if one is sad, the response is to go out, to go shopping, to immerse one’s self in work—anything that takes one’s mind off the sad mood and the reasons for that sad mood. Research by Nolen-Hoeksema reveals that the use of a distraction strategy results in a shorter duration of sad moods than a ruminative strategy (e.g., Nolen-Hoeksema & Morrow, 1991; Nolen-Hoeksema, Morrow, & Fredrickson, 1993). This may offer additional explanatory power for why some consumers, particularly males, respond most favorably to the sad-mood–happy-ad combination. Such consumers may be employing a distraction strategy to achieve mood repair, whereby the affective tone of the happy ad is used to facilitate the priming of a positive mood. Future research of this nature could employ the negativemood response-style questionnaire (Butler & Nolen-Hoeksema, 1994), which consists of 20 items that can be used to classify subjects as distractors or ruminators (see Cheng, 1999).

Another intriguing avenue would be to examine gender–mood effects in relation to sex-type self-discrepancy. Self-discrepancy occurs where there is a difference between a person’s actual behavior and the perceived appropriate behavior for that person’s sex type (Grimmell & Stern, 1992). Previous research has suggested that self-discrepancy is a significant predictor of negative mood (Grimmell, 1998). Thus, researchers could explore how self-discrepancy in a consumption context influences consumer mood effects.

REFERENCES

Abele, A., & Hermer, P. (1993). Mood influences on health-related judgments: Appraisal of own health versus appraisal of unhealthy behaviours. European Journal of Social Psychology, 23, 613 – 625. Asuncion, A. G., & Lam, W. F. (1995). Affect and impression formation: Influence of mood on person memory. Journal of Experimental Social Psychology, 31, 437 – 464. Bagozzi, R. P. (1996). The role of arousal in the creation and control of the halo effect in attitude models. Psychology & Marketing, 13, 235 – 264. Bagozzi, R. P., Gopinath, M., & Nyer, P. U. (1999). The role of emotions in advertising. Journal of the Academy of Marketing Science, 27, 184 – 206. Barone, M. J., Miniard, P. W., & Romeo, J. B. (2000). The influence of positive mood on brand extension evaluations. Journal of Consumer Research, 26, 386 – 400. Basow, S. A. (1986). Gender stereotypes. Monterey, CA: Brooks-Cole. Batra, R., & Stayman, D. M. (1990). The role of mood in advertising effectiveness. Journal of Consumer Research, 17, 203 – 214. Batra, R., & Stephens, D. (1994). Attitudinal effects of ad-evoked moods and emotions: The moderating role of motivation. Psychology & Marketing, 11, 199 – 215. Bem, S. L. (1976). Probing the promise of androgyny. In A. G. Kaplan & J. P. Bean (Eds.), Beyond sex-role stereotypes: Readings towards a psychology of androgyny. Boston: Brown, Little. Blackhart, G. C., Kilne, J. P., Donohue, K. F., LaRowe, S. D., & Joiner, T. E. (2001). Affective responses to EEG preparation and their linkto resting anterior EEG symmetry. Personality & Individual Differences, 32, 167 – 174. Blaney, P. H. (1986). Affect and memory: A review. Psychological Bulletin, 99, 229 – 246. Bless, H., Schwarz, N., & Wieland, R. (1996). Mood and the impact of category membership and individuating information. European Journal of Social Psychology, 26, 935 – 959. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129 – 148. Brewster, A. L., Nelson, J. P., McCanne, T. R., Lucas, D. R., & Milner, J. (1998). Gender differences in physiological reactivity to infant cries and smiles in military families. Child Abuse & Neglect, 22, 775 – 788. Brown, S. P., Homer, P. M., & Inman, J. (1998). A meta-analysis of relationships between ad-evoked feelings and advertising responses. Journal of Marketing Research, 35, 114 – 126. Brown, S. P., & Stayman, D. M. (1992). Antecedents and consequences of attitude toward the ad: A meta-analysis. Journal of Consumer Research, 19, 34 – 51.Burke, M. C., & Edell, J. A. (1989). The impact of feelings upon ad-based affect and cognition. Journal of Marketing Research, 26, 69 – 83. Butler, L. D., & Nolen-Hoeksema, S. (1994). Gender differences in responses to depressed mood in a college sample. Sex Roles, 30, 331 – 346. Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social Psychology, 42, 116 – 131. Celsi, R. L., & Olson, J. C. (1988). The role of involvement in attention and comprehension processes. Journal of Consumer Research, 15, 210 – 224. Cheng, C. (1999). Gender-role differences in susceptibility to the influence of support availability on depression. Journal of Personality, 67, 439 – 467. Clark, M. S. (1982). A role for arousal in the link between feeling states, judgements and behavior. In M. S. Clark& S. T. Fiske (Eds.), Affect and cognition (pp. 263 – 289). Hillsdale, NJ: Erlbaum. Clore, G. L., Schwarz, N., & Conway, M. (1994). Cognitive causes and consequences of Emotion. In R. S. Wyer & T. K. Srull (Eds.), Handbookof social cognition (pp. 323 – 417).Hillsdale: Erlbaum. Cohen, J. (1977). Statistical power analysis for the behavioral sciences. New York: Academic Press. Curren, M. T., & Harich, K. R. (1994). Consumers’ mood states: The mitigating influence of personal relevance on product evaluations. Psychology & Marketing, 11, 91 – 107. Dean, D. H. (1999). Brand endorsement, popularity, and event sponsorship as advertising cues affecting consumer pre-purchase attitudes. Journal of Advertising, 28, 1 – 12. De Wulf, K., Odekerken-Schroder, G., & Iacobucci, D. (2001). Investments in consumer relationships: A cross-country and cross-industry exploration. Journal of Marketing, 65, 33 – 50. Dube, L., & Morgan, M. S. (1996). Trend effects and gender differences in retrospective judgments of consumption emotions. Journal of Consumer Research, 23, 156 – 162. Duff, S. J., & Hampson, E. (2001). A sex difference on a novel spatial working memory taskin humans. Brain and Cognition, 47, 470 – 493. Ellis, H. C., & Ashbrook, T. W. (1988). Resource allocation model of the effects of depressed mood on memory. In K. Fielder & J. P. Forgas (Eds.), Affect, cognition and social behavior (pp. 25 – 43). Gottingen, Germany: Hogrefe. Erber, R., & Erber, M. W. (1994). Beyond mood and social judgment: Mood incongruent recall and mood regulation. European Journal of Social Psychology, 24, 79 – 88. Fern, E. F., & Monroe, K. B. (1996). Effect-size estimates: Issues and problems in interpretation. Journal of Consumer Research, 23, 89 – 105. Forgas, J. P. (1992). Affect in social judgments and decisions: A multiprocess model. Advances in Experimental Social Psychology, 25, 227 – 275. Forgas, J. P. (1995). Mood and judgment: The affect infusion model (AIM). Psychological Bulletin, 117, 39 – 66. Frable, D. E. S., & Bem, S. L. (1985). If you are gender schematic, all members of the opposite sex lookalike. Journal of Personality and Social Psychology, 49, 459 – 468. Gainer, B. (1993). An empirical investigation of the role of involvement with a gendered product. Psychology & Marketing, 10, 265 – 283.Gardner, M. P. (1985). Mood states and consumer behavior: A critical review. Journal of Consumer Research, 12, 281 – 300. Gerrards-Hesse, A., Spies, K., & Hesse, F. W. (1994). Experimental inductions of emotional states and their effectiveness: A review. British Journal of Psychology, 85, 55 – 78. Goldberg, M. E., & Gorn, G. J. (1987). Happy and sad TV programs: How they affect reactions to commercials. Journal of Consumer Research, 14, 387 – 403. Grimmell, D. W. (1998). Effects of gender-role self-discrepancy on depressed mood. Sex Roles, 39, 203 – 214. Grimmell, D. W., & Stern, G. S. (1992). The relationship between gender role ideals and psychological well-being. Sex Roles, 27, 487 – 497. Grunert, K. G. (1996). Automatic and strategic processes in advertising effects. Journal of Marketing, 60, 88 – 101. Harshman, R. A., Hampson, E., & Berenbaum, S. A. (1983). Individual differences in cognitive abilities and brain organization. Neuropsychologica, 36, 37 – 43. Hertel, G., & Fielder, K. (1994). Affective and cognitive influences in a social dilemma game. European Journal of Social Psychology, 24, 131 – 145. Holbrook, M. B., & Gardner, M. P. (2000). Illustrating a dynamic model of the mood-updating process in consumer behavior. Psychology & Marketing, 17, 165 – 194. Homer, P. M., & Yoon, S. (1992). Message framing and the interrelationships among ad- based feelings, affect, and cognition. Journal of Advertising, 21, 19 – 33. Hornik, J. (1993). The role of affect in consumers’ temporal judgments. Psychology & Marketing, 10, 239 – 255. Huitema, B. E. (1980). The analysis of covariance and alternatives. New York: Wiley. Iacobucci, D. (1994). Analysis of experimental data. In R. P. Bagozzi (Ed.), Principles of marketing research (pp. 224 – 278). Cambridge, MA: Basil Blackwell Ltd. Isen, A.M. (1984). Toward understanding the role of affect in cognition. In R. Wyer, T. Srull, & A. Isen (Eds.), Handbookof social cognition (Vol. 3, pp. 179 – 236). Isen, A. M. (1987). Positive affect, cognitive processes, and social behavior. In L. Berkowitz (Ed.), Advances in experimental psychology (Vol. 20, pp. 203 – 253). San Diego: Academic Press. Izard, C. E., Dougherty, F. E., Bloxom, B. M., & Kotsch, W. E. (1974). The differential emotions scale: A method of measuring the subjective experience of discrete emotions. Unpublished manuscript, Vanderbilt University, Nashville. Johar, G. V., & Simmons, C. J. (2000). The use of concurrent disclosures to correct invalid inferences. Journal of Consumer Research, 26, 307 – 322. Kamins, M. A., Marks, L. J., & Skinner, D. (1991). Television commercial evaluation in the context of program induced mood: Congruency versus consistency effects. Journal of Advertising, 20, 1 – 14. Kellaris, J. J., & Mantel, S. P. (1994). The influence of mood and gender on consumers’ time perceptions. In C. T. Allen & D. Roedder John (Eds.), Ad-vances in consumer research (Vol. 21, pp. 514 – 518). Provo, UT: Association for Consumer Research. Kiecker, P., Palan, K. M., & Areni, C. S. (2000). Different ways of ‘seeing’: How gender differences in information processing influence the content analysis of narrative texts. Marketing Letters, 11, 49 – 65. Klinteberg, B. A., Levander, S. E., & Schalling, D. (1987). Cognitive sex differences: Speed and problem-solving strategies on computerized neuropsychological tasks. Perceptual and Motor Skills, 65, 683 – 697. Kramer, R. M., Newton, E., & Pommerenke, P. L. (1993). Self-enhancement biases and negotiator judgment: Effects of self-esteem and mood. Organizational Behavior & Human Decision Processes, 56, 110 – 133. Kring, A. M., & Gordon, A. H. (1998). Sex differences in emotion: Expression, experience, and physiology. Journal of Personality and Social Psychology, 74, 686 – 703. Kupfer, A. (1998). AT&T gets lucky. Fortune, 138, 98 – 102. Kupfer, A. (1999). AT&T goes cable crazy. Fortune, 139, 79 – 81. Laczniak, R. N., & Muehling, D. D. (1993). Toward a better understanding of the role of advertising message involvement in ad processing. Psychology & Marketing, 10, 301 – 319. Larsen, J. T., McGraw, A., & Cacioppo, J. T. (2001). Can people feel happy and sad at the same time? Journal of Personality & Social Psychology, 81, 684 – 696. LaTour, M. S., & Rotfeld, H. J. (1997). There are threats and (maybe) fearcaused arousal: Theory and confusions of appeals to fear and fear arousal itself. Journal of Advertising, 26, 45 – 59. Lee, A. Y., & Sternthal, B. (1999). The effects of positive mood on memory. Journal of Consumer Research, 26, 115 – 127. Levin, I. P., Schneider, S. L., & Gaeth, G. J. (1998). All frames are not created equal: A typology and critical analysis of framing effects. Organizational Behavior and Human Decision Processes, 76, 149 – 188. Lott, B. (1987). Women’s lives: Themes and variation in gender learning. Monterey, CA: Brooks Cole. Luomala, H. T., & Laaksonen, M. (2000). Contributions from mood research. Psychology & Marketing, 17, 195 – 233. Lyubormirsky, S., & Nolen-Hoeksema, S. (1993). Self-perpetuating properties of dysphoric rumination. Journal of Personality and Social Psychology, 65, 339 – 349. Mackenzie, S. B., & Lutz, R. J. (1989). An empirical examination of the structural antecedents of attitude toward the ad in an advertising pretesting context. Journal of Marketing, 53, 48 – 65. Mackenzie, S. B., Lutz, R. J., & Belch, G. E. (1986). The role of attitude toward the ad as a mediator of advertising effectiveness: A test of competing explanations. Journal of Marketing Research, 23, 130 – 143. Mackie, D. M., & Worth, L. T. (1989). Processing deficits and the mediation of positive affect in persuasion. Journal of Personality and Social Psychology, 57, 27 – 40. Maheswaran, D., & Meyers-Levy, J. (1990). The influence of message framing and issue involvement. Journal of Marketing Research, 27, 361 – 367.Mano, H. (1992). Judgments under distress: Assessing the role of unpleasantness and arousal in judgment formation. Organizational Behavior and Human Decision Processes, 52, 216 – 245. Mano, H. (1997). Affect and persuasion: The influence of pleasantness and arousal on attitude formation and message elaboration. Psychology & Marketing, 14, 315 – 335. Marusic, I., & Bratko, D. (1998). Relations of masculinity and femininity with personality dimensions of the five-factor model. Sex Roles, 38, 29 – 44. Matmur, M., & Chattopadhayay, A. (1991). The impact of moods generated by television programs on responses to advertising. Psychology & Marketing, 8, 59 – 77. Meloy, M. G. (2000). Mood-driven distortion of product information. Journal of Consumer Research, 27, 345 – 359. Meyers-Levy, J. (1988). The influence of sex roles in judgment. Journal of Consumer Research, 14, 522 – 530. Meyers-Levy, J., & Maheswaran, D. (1991). Exploring differences in males’ and females’ processing strategies. Journal of Consumer Research, 18, 63 – 70. Meyers-Levy, J., & Sternthal, B. (1991). Gender differences in the use of message cues and judgments. Journal of Marketing Research, 28, 84 – 96. Mitchell, A. A., & Olson, J. C. (1981). Are product attribute beliefs the only mediators of advertising effects on brand attitudes? Journal of Marketing Research, 18, 318 – 322. Mittal, B. (1995). A comparative analysis of four scales of consumer involvement. Psychology & Marketing, 12, 663 – 682. Morris, W. N., & Reilly, N. P. (1987). Toward the self-regulation of mood: Theory and research. Motivation and Emotion, 11, 215 – 249. Murry, J. P., Lastovicka, J. L., & Singh, S. (1992). Feeling and liking responses to television programs: An examination of two explanations for media-context effects. Journal of Consumer Research, 18, 441 – 451. Nicholson, K. G., & Kimura, D. (1996). Sex differences for speech and manual skill. Perceptual and Motor Skills, 82, 3 – 13. Nolen-Hoeksema, S., Larson, J., & Grayson, C. (1999). Explaining the gender difference in depressive symptoms. Journal of Personality and Social Psychology, 77, 1061 – 1072. Nolen-Hoeksema, S., & Morrow, J. (1991). A prospective study of depression and distress following a natural disaster: The 1989 Loma Prieta earthquake. Journal of Personality and Social Psychology, 62, 105 – 121. Nolen-Hoeksema, S., Morrow, J., & Fredrickson, B. L. (1993). Response styles and the duration of episodes of depressed mood. Journal of Abnormal Psychology, 102, 20 – 28. Oliver, M. B., Sargent, S. L., & Weaver, J. B. (1998). The impact of sex and gender role self-perception on affective reactions to different types of film. Sex Roles, 38, 45 – 62. Olney, T. J., Holbrook, M. B., & Batra, R. (1991). Consumer responses to advertising: The effects of ad content, emotions, and attitude toward the ad on viewing time. Journal of Consumer Research, 17, 440 – 453. Park, J., & Banaji, M. R. (2000). Mood and heuristics: The influence of happy and sad states on sensitivity and bias in stereotyping. Journal of Personality and Social Psychology, 78, 1005 – 1023.Paulhus, D. L. (1991). Measurement and control of response bias. In J. P. Robinson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measures of personality and social psychological attitudes (pp. 17 – 60). Peterson, R. A., Albaum, G., & Beltrami, R. F. (1985). A meta-analysis of effects in consumer behavior experiments. Journal of Consumer Research, 12, 97 – 103. Pham, M. T. (1992). Effects of involvement, arousal, and pleasure on the recognition of sponsorship stimuli. In J. Sherry & B. Sternthal (Eds.), Advances in consumer research, (Vol. 19, pp. 85 – 93). Provo, UT: Association for Consumer Research. Pogun, S. (2001). Sex differences in brain and behavior: Emphasis on nicotine, nitric oxide and place learning. Personality and individual differences. International Journal of Psychophysiology, 42, 195 – 208. Rosenthal, R., & Rosnow, R. L. (1985). Contrast analysis: Focused comparisons in the analysis of variance. New York: Cambridge University Press. Rusting, C. L. (1998). Personality, mood, and cognitive processing of emotional information: Three conceptual frameworks. Psychological Bulletin, 124, 165 – 196. Rusting, C. L. (1999). Interactive effects of personality and mood on emotioncongruent memory and judgment. Journal of Personality and Social Psychology, 77, 1073 – 1086. Rusting, C. L., & DeHart, T. (2000). Retrieving positive memories to regulate negative mood: Consequences for mood-congruent memory. Journal of Personality and Social Psychology, 78, 737 – 752. Sanbonmatsu, D. M., & Kardes, F. R. (1988). The effects of physiological arousal on information processing and persuasion. Journal of Consumer Research, 15, 379 – 385. Saucier, D. M., Elias, L. J., & Nylen, K. (2002). Are colours special? An examination of the female advantage for speeded colour naming. Personality and Individual Differences, 32, 27 – 35. Schwarz, N. (1990). Feelings as information: Informational and motivational functions of affective states. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbookof motivation and cognition: Foundations of social behavior (pp. 527 – 561). New York: Guilford. Schwarz, N., Bless, H., & Bohner, G. (1991). Mood and persuasion: Affective states influence the processing of persuasive communications. Advances in Experimental Social Psychology, 24, 161 – 199. Sedikides, C. (1995). Central and peripheral self-conceptions are differentially influenced by mood: Tests of the differential sensitivity hypothesis. Journal of Personality and Social Psychology, 69, 759 – 777. Seidlitz, L., & Diener, E. (1998). Sex differences in the recall of affective experiences. Journal of Personality and Social Psychology, 74, 262 – 271. Sethi, S., & Nolen-Hoeksema, S. (1997). Gender differences in internal and external focusing among adolescents. Sex Roles, 37, 687 – 700. Smith, R. B., & Petty, R. E. (1995). Personality moderators of mood congruency effects on cognition: The role of self-esteem and negative mood regulation. Journal of Personality and Social Psychology, 68, 1092 – 1107. Stern, B. B. (1993). Feminist literary criticism and the deconstruction of ads: A postmodern view of advertising and consumer responses. Journal of Consumer Research, 19, 556 – 566. Steward, S. (1995). Tapping the well for new customers. Cellular Business, 34 – 38. Swinyard, W. R. (1993). The effects of mood, involvement, and quality of store experience on shopping intentions. Journal of Consumer Research, 20, 271 – 280. Thayer, R. E., Newman, J. R., & McClain, T. M. (1994). Self-regulation of mood: Strategies for changing a bad mood, raising energy, and reducing tension. Journal of Personality and Social Psychology, 67, 910 – 925. Tice, D. M., Bratslavsky, E., & Baumeister, R. F. (2001). Emotional distress regulation takes Precedence over impulse control: If you feel bad, do it! Journal of Personality and Social Psychology, 80, 53 – 67. Watson, D., Clark, L. A., & Carey, G. (1988). Positive and negative affectivity and their relation to anxiety and depressive disorders. Journal of Abnormal Psychology, 97, 346 – 353. Watson, P. J., Biderman, M. D., & Sawrie, S. M. (1994). Empathy, sex role orientation, and narcissism. Sex Roles, 30, 701 – 723. Wildt, A. R., & Ahtola, O. T. (1978). Analysis of covariance. Beverly Hills, CA: Sage. Wood, J. V., Michela, J. L., & Giordano, C. (2000). Downward comparison in everyday life: Reconciling self-enhancement models with the mood-cognition priming model. Journal of Personality & Social Psychology, 79, 563 – 579. Yi, Y. (1990). Cognitive and affective priming effects of the context for print advertisements. Journal of Advertising, 19, 40 – 48. Yi, Y. (1993). Contextual priming effects in print advertisements: The moderating role of prior knowledge. Journal of Advertising, 22, 1 – 10. Zaichowsky, J. L. (1985). Measuring the involvement construct. Journal of Consumer Research, 12, 341 – 352. Zillman, D., Weaver, J. B., Mundorf, N., & Aust, C. F. (1986). Effects of an opposite-gender companion’s affect to horror on distress, delight, and attraction. Journal of Personality and Social Psychology, 51, 586 – 594. Correspondence regarding this article should be sent to Brett Martin, Department of Marketing, University of Auckland, Private Bag 92019, Auckland, New Zealand (bas.martin@auckland.ac.nz).

 

CONCLUSION EXPLICITNESS IN ADVERTISING

The Moderating Role of Need for Cognition (NFC) and Argument Quality (AQ) on Persuasion

Brett A. S. Martin, Bodo Lang, and Stephanie Wong

ABSTRACT: Previous research into the use of explicit and implicit conclusions in advertising has yet to demonstrate consistent effects for both brand attitudes and purchase intentions. While research has examined the role of involvement, this study contributes by examining the trait called need for cognition (NFC), which addresses a person’s propensity to engage in effortful thinking. In addition, this study introduces argument quality (AQ) as another potential moderator of conclusion explicitness effects. In a 2 × 2 experiment of 261 subjects, conclusion explicitness (explicit conclusion, implicit conclusion) and AQ (strong, weak) are manipulated, with NFC (high NFC, low NFC) as a third measured variable. Results indicate more favorable evaluations for implicit conclusions over explicit conclusions for high-NFC individuals. Further, implicit conclusions result in more favorable brand attitudes and purchase intentions when linked with strong AQ for high-NFC individuals. The findings confirm that conclusion explicitness does not differentially affect the evaluations of low-NFC subjects. Results suggest that NFC may represent an important moderating variable for future conclusion explicitness research.

One of the ultimate aims of advertising is to persuade consumers to buy certain brands over others. To achieve this goal, many advertisers utilize advertisements with a clear conclusion (e.g., Beardi 2001; Halliday 2001). For obvious conclusions, however, it may be more effective to imply rather than state the intended conclusion, as this may be viewed as less of a “hard sell.” Indeed, in the field of comparative advertising, the use of explicit and implicit conclusions is becoming increasingly common (Barone et al. 1999). For example, a recent print advertisement by Saab presents the performance of a Saab and a BMW on a number of attributes. The ad invites consumers to “compare the value you will get,” before stating, “and then you make the decision.” Although early conclusion explicitness research found explicit conclusions to be more effective (e.g., Fine 1957; Hovland, Janis, and Kelley 1953), recent research has shown the benefits of implicit conclusions in advertising (e.g., Ahearne, Gruen, and Saxton 2000; Sawyer and Howard 1991). Thus, conclusion explicitness offers practitioners a way of formulating ad copy to enhance advertising effectiveness.

For academic researchers, theoretical understanding of conclusion explicitness effects is promising but underdeveloped. For instance, research indicates that a consumer’s motivation to process ad information has a key impact on whether implicit conclusions are effective. Motivated consumers tend to be more likely to be persuaded by implicit conclusions. This research has studied the impact of situational states, such as involvement (e.g., Chebat, Charlebois, and Gélinas-Chebat 2001; Sawyer and Howard 1991). However, little research has examined the effects of personality traits on conclusion explicitness effects. This gap in the literature is important given that the trait called need for cognition (NFC) (Cacioppo and Petty 1982) relates to a person’s motivation to process information. Further, researchers have identified NFC as a potentially important moderator of conclusion explicitness effects, and have called for research on this topic (e.g., Ahearne, Gruen, and Saxton 2000; Kardes, Kim, and Lim 1994). Thus, the present research offers three contributions, which are outlined in the paragraphs that follow.

First, the role of NFC is studied, thereby extending the literature from situational states (e.g., involvement) to predispositional traits (e.g., NFC). This contributes by exploring the cognitive processes involved in conclusion explicitness effects, specifically the impact of individual differences on a person’s motivation to think about an ad. This tests a boundary condition of conclusion explicitness research, and explores the role of individual differences in advertising. Individual differences are currently being highlighted as offering useful perspectives for marketers (e.g., Baumgartner 2002; Luna and Peracchio 2002), and research into traits has offered useful insights for advertising research (e.g., Moore and Har-ris 1996; Zhang and Buda 1999). Research into ad avoidance also reveals individuals’ differences in motivation to process ads, with motivated people less likely to avoid print ads (e.g., Speck and Elliott 1997). Consequently, NFC appears to be a useful construct to consider.

Second, we explore the moderating role of argument quality (AQ) on conclusion explicitness effects. Previous research suggests that the effect of AQ is influenced by the amount of elaboration engaged in by a consumer (Batra and Stayman 1990). Since NFC relates to a person’s inherent tendency to engage in elaboration, researchers have suggested that AQ is an important factor to consider in conjunction with NFC (Batra and Stayman 1990; Cacioppo et al. 1986). As stated by Petty, Unnava, and Strathman (1991, p. 246), “People who enjoy thinking (high in ‘need for cognition’) tend to form attitudes on the basis of the quality of the arguments in a message.” Consequently, studying AQ contributes to our theoretical understanding of the role of NFC in conclusion explicitness effects.

Third, as the majority of studies in conclusion explicitness have examined simple product categories, such as toothbrushes (Sawyer 1988; Sawyer and Howard 1991), this study contributes by examining conclusion explicitness effects for a complex product category. A complex product is defined as a product possessing several functions or features (Griffin 1997). Studying complex products allows us to examine the generalizability of conclusion explicitness effects. From a managerial perspective, insights gained from a complex product context is highly relevant, since many comparative ads compare brands on a variety of attributes.

LITERATURE REVIEW

Defining Explicit and Implicit Conclusions

Within the conclusion explicitness literature, there is variation in the terminology used to describe an ad format that provides a conclusion or no conclusion. Researchers have referred to closed-ended and open-ended messages (e.g., Ahearne, Gruen, and Saxton 2000; Sawyer and Howard 1991) or explicit and implicit conclusions (e.g., Kardes 1988; Kardes, Kim, and Lim 1994). As this study addresses message conclusion explicitness, the latter terms are used.

Explicit conclusions involve the direct statement of a conclusion within an ad (Sawyer and Howard 1991), such as “Brand X is better than the rest.” An advantage of this type of message is that the chances of a consumer misinterpreting the ad are minimized (Ahearne, Gruen, and Saxton 2000). However, Kardes, Kim, and Lim (1994) suggest that explicit conclusions are a form of “hard sell” that allow little scope for individual interpretation, which can result in distrust and less favorable evaluations. By contrast, implicit conclusions do not directly state a conclusion. This format relies on an implied set of arguments that are designed to lead an audience toward the intended conclusion. Phrases commonly seen in such ads ask the consumer to “compare for yourself” and suggest that “you make the decision.” Such messages allow consumers to form their own conclusion based on the information provided (Kardes, Kim, and Lim 1994; Sawyer and Howard 1991). They also encourage consumers to read the ad prior to making their own decision, thus prompting higher levels of message processing. Advantages of implicit conclusions include enhanced advertiser credibility owing to being perceived as less coercive. Yet the risks of using implicit conclusions include consumers failing to form a conclusion (see Sawyer and Howard 1991), or reaching the wrong conclusion (Ahearne, Gruen, and Saxton 2000; Kardes, Kim, and Lim 1994).

Previous Research on Conclusion Explicitness

Early research indicated that explicit conclusions resulted in greater opinion change (for reviews, see Sawyer 1988; Sawyer and Howard 1991). More recently, scholars have begun to address how conclusion explicitness relates to a consumer’s motivation to process the message. Kardes (1988) was the first study to consider the role of involvement and conclusion explicitness. Involvement is defined as the level of personal relevance that motivates individuals to engage in effortful processing (Batra and Stayman 1990). Kardes’s (1988) findings suggest that when an audience is confronted with an implicit conclusion ad, high-involvement subjects are likely to spontaneously generate inferences about the missing conclusion. This facilitates highly accessible brand attitudes, whereas brand attitudes for low-involvement subjects are relatively inaccessible due to an insufficient motivation to infer missing conclusions. However, Kardes (1988) did not find a differential effect for involvement on brand attitudes for implicit messages.

This lack of a persuasive advantage for implicit conclusions prompted further exploration by Sawyer and Howard (1991). Rather than processes, their main focus was on the relative persuasive impact of implicit and explicit messages, and the moderating role of involvement. They presented subjects with information on the relative performance of four brands across a set of attributes. Such a design made a conclusion about each attribute, and a global conclusion, clearer. They found a persuasive advantage for implicit conclusions over explicit conclusions for brand attitudes, purchase intentions, and choice behavior under high involvement. Recently, Ahearne, Gruen, and Saxton (2000) replicated Sawyer and Howard (1991) in two experiments—one with low product complexity and one with high product complexity. They found similar results to Sawyer and Howard (1991) for the simple product category, but nonsignificant results for the complex product (a compact disc player). However, their use of real brand names and attributes with complex terminology (e.g., bump immunity, programmability) may have affected subjects’ ability to process the message, and consequently, the results. Due to such methodological issues in previous studies, a complex product is used as the basis for this study. Overall, the research on conclusion explicitness and involvement highlights involvement as a moderator, and suggests that the advantage of implicit conclusions may be conditional upon individual traits or message-specific factors.

NFC, AQ, and Conclusion Explicitness

Many studies have suggested that need for cognition (NFC) is an important variable to consider for conclusion explicitness (e.g., Ahearne, Gruen, and Saxton 2000; Sawyer 1988). NFC refers to an individual’s propensity to engage in and enjoy cognitively demanding tasks (Cacioppo, Petty, and Kao 1984). High-NFC individuals enjoy solving complex problems and report greater cognitive effort relative to low-NFC individuals (Batra and Stayman 1990).

Previous research suggests that high-NFC individuals are more influenced by argument quality (AQ) (e.g., Inman, McAlister, and Hoyer 1990; Zhang 1996). Indeed, strong arguments tend to be more persuasive for high-NFC individuals. Further, they are more likely to seek out and elaborate on information, since they enjoy doing so (Luna and Peracchio 2002). Conversely, low-NFC individuals are less motivated to study a message in depth. As a result, they are more influenced by humor (Zhang 1996), promotion signals (Inman, McAlister, and Hoyer 1990), and positive mood (Batra and Stayman 1990), which suggests that attitude change relates to simple cues in the advertising message. In much of this research (e.g., Cacioppo et al. 1986; Inman, McAlister, and Hoyer 1990; Zhang 1996), the Elaboration Likelihood Model (ELM) (Petty and Cacioppo 1981) has been used to interpret the findings, with high-NFC individuals following a central, cognitively effortful route to persuasion, and lowNFC individuals following a simpler peripheral, cue processing route.

Stayman and Kardes (1992) were the first to consider NFC in a conclusion explicitness context. Although the focus of their study was spontaneous inference generation rather than persuasion, they found that inferences about implicit conclusions were more likely to be spontaneous for high-NFC individuals, owing to a greater processing motivation. This is consistent with previous research where highly involved audiences generated inferences about implicit conclusions (Kardes 1988). However, it is unclear whether high-NFC individuals respond differently implicit conclusions in terms of persuasion. This study seeks to explore this important issue.

HYPOTHESES

Effects for NFC and Conclusion Explicitness

Hypothesis 1 predicts that the NFC levels of consumers (high, low) and the conclusion explicitness of the ad (implicit, explicit) will interact, resulting in differences in persuasion. Specifically, given that previous research has found involvement to be a moderator of conclusion explicitness (e.g., Kardes 1988; Sawyer and Howard 1991), we might expect NFC to be a moderator of conclusion explicitness effects. From an ELM perspective, Petty, Unnava, and Strathman (1991) classify involvement as a situational factor, and NFC as an individual factor, with both influencing an individual’s motivation to process a message. Thus, since high-NFC individuals enjoy elaborating upon a message, they are more likely to evaluate implicit messages more favorably than explicit conclusions. By contrast, conclusion explicitness effects are not expected for low-NFC consumers, as these effects require the effortful processing of arguments presented in an ad. Hence, the following hypothesis is proposed:

H1: There will be a significant interaction between NFC and conclusion explicitness across dependent measures. Specifically, for high-NFC individuals, implicit conclusion ads will lead to more favorable attitudes toward the ad (Aad), attitudes toward the brand (Ab ), brand beliefs, and purchase intentions, when compared with an ad with an explicit conclusion. For low-NFC individuals, there will be no difference between implicit conclusion ads and explicit conclusion ads for Aad, Ab , brand beliefs, or purchase intentions.

Effects for NFC and AQ

Argument quality (AQ) is defined as the valence of thoughts evoked by an argument (Batra and Stayman 1990). Strong arguments elicit more favorable thoughts about an advocated position; weak arguments elicit more unfavorable thoughts. Despite its potential importance to persuasion, the application of AQ to conclusion explicitness has not been examined. Indeed, much of the previous research has focused on the most important product attributes for comparison between brands, constituting a relatively strong AQ. In this study, we examine the role of AQ for strong and weak arguments. Previous research suggests that AQ has a greater impact on persuasion under high involvement (Petty and Cacioppo 1981) and for high-NFC individuals (Batra and Stayman 1990; Cacioppo et al. 1986). Thus, AQ is expected to be an important persuasion variable for high-NFC individuals. Since high-NFC consumers are more likely to scrutinize the arguments in a message (Cacioppo et al. 1986), it is expected that a strong AQ will be more persuasive than a weaker argument. By contrast, lowNFC individuals are expected to be unaffected by AQ, owing to a lack of motivation to process the information. Thus, the second hypothesis is posited:

H2: There will be a significant interaction between NFC and AQ across dependent measures. Specifically, for high-NFC individuals, a strong AQ ad will lead to more favorable Aad, Ab , brand beliefs, and purchase intentions when compared with a weak AQ ad. For low-NFC individuals, there will be no difference between a strong AQ ad and a weak AQ ad for Aad, Ab , brand beliefs, or purchase intentions.

Effects for NFC, Conclusion Explicitness, and AQ

Drawing upon H1 and H2, we expect that high-NFC individuals will be persuaded by ads with implicit conclusions over explicit conclusions, and that this effect will be stronger for ads with a strong AQ. Sawyer (1988) suggests that an implicit conclusion with weak arguments may result in the wrong conclusion being reached. Hence, we expect that conclusions drawn from strong arguments will be perceived as more valid than conclusions based on weak arguments. For low-NFC individuals, no preference should be evident. Specifically, the following hypothesis is suggested.

H3: There will be a significant interaction between NFC, conclusion explicitness, and AQ across dependent measures. Specifically, high-NFC individuals will exhibit more favorable Aad, Ab , brand beliefs, and purchase intentions toward implicit conclusion ads with strong AQ. For low-NFC individuals, no preference in ad type will be exhibited.

METHOD

Subjects, Design, and Procedure

A total of 275 students recruited from an undergraduate marketing class were randomly assigned to one of eight conditions. Of these, 14 students were excluded owing to incomplete responses, resulting in a final sample of 261 students. Subjects participated in groups of 85 to 100 and received the chance to win free movie passes.

The design of the experiment was a 2 (conclusion explicitness: explicit, implicit) × 2 (AQ: strong, weak) between-subjects factorial design, with NFC (high, low) used as a measured independent variable, following a median split procedure as in previous research (e.g., Batra and Stayman 1990; Luna and Peracchio 2002). Subjects were informed that a study was being conducted on cellular phone advertisements. They then read a booklet containing an ad and the questionnaire. Subjects were asked to read the ad as they would normally do if they were reading it in a magazine. The entire procedure took 20 minutes to complete. Subjects were later debriefed in a follow-up session.

Experimental Stimulus Development

Pretest 1

This pretest identified an appropriate product based on two criteria: (1) the product offered a range of attributes for the AQ manipulation, and (2) the product was relevant to a student sample. First, 20 undergraduates were asked to create a list of complex products. Next, 32 subjects rated the four most frequently mentioned products from stage one on five, seven-point scales (e.g., unimportant/important) for involvement (a list of past research from which the measures used in this study were sourced is available from the first author upon request), from which an average score was derived. Prior knowledge, ownership, and frequency of use measures were also administered. It was found that cell phones had the highest involvement score (M = 5.04), most subjects had previously or currently owned a cell phone (93.8%), and a large number use a cell phone more than four times a week (87.6%), suggesting a moderate to high frequency of use. Thus, cell phones were selected.

Pretest 2

This pretest sought to determine product attributes for the AQ manipulation. Twenty-two subjects rated a list of 15 attributes derived from a content analysis on six, seven-point scales for AQ (e.g., not compelling/compelling). These scores were summed to form an index. The five attributes with the highest means (talk time, standby time, vibrating alert, weight, and security features) and the five lowest means (ringing alert options, internal antennae, range of colors, Personal Information Manager, and interchangeable face plates) were chosen for the strong and weak AQ manipulations, respectively.

Pretest 3

This pretest identified fictitious brand names. Eight brand names were created that (1) did not sound phonetically similar to existing brands, and (2) did not include a relevant attribute or benefit (e.g., PicturePerfect) that can lead to higher recall (Keller, Heckler, and Houston 1998). Thirty-one undergraduates rated these names on five, seven-point items (e.g., bad/good). The three brands rated as most similar were Tectron TZ (M = 4.10), Samsonic SX (M = 4.19), and Norden NT (M = 4.16). A paired-samples t test confirmed that there were no significant differences between these evaluations (p > .10). Hence, Samsonic was chosen as the target brand, with Tectron and Norden as the two competitors. Furthermore, the comparative advertising format adopted reflects the format of recent advertising by well-known brands (e.g., Toyota, Saab),but not of any cellular phone brands that may prime, and thus bias, responses.

Independent Variables

For conclusion explicitness, the conclusion was clearly stated (explicit) or implied (implicit). Explicit conclusion conditions contained a statement that the target brand (Samsonic) was superior: “Now that you’ve seen the facts, choose Samsonic— the cellular phone which is best for you.” For the implicit conclusion, the following statement invited subjects to infer their own conclusion about which brand was superior: “Now that you’ve seen the facts, decide for yourself which cellular phone is best for you.” This approach was adapted from Sawyer and Howard (1991). An example of the ad stimuli is provided in the Appendix.

NFC was measured using the 18-item scale devised by Cacioppo, Petty, and Kao (1984). For AQ, strong AQ ads contained the five most important attributes (pretest 2), with the target brand outperforming competitors on four of the five attributes. By contrast, weak AQ ads contained the five least important attributes, with the target brand outperforming competitors on three of the five attributes. The performance of the target brand relative to competitors was varied to provide for a more comprehensive AQ manipulation than simply changing the nature of attributes promoted in the ad copy. AQ manipulations emphasized attributes, which follows recommendations for long advertising copy (e.g., Westphal 2001).

Dependent Variables

Brand attitudes were measured on four, seven-point scales (bad/good, dislike quite a lot/like quite a lot, unpleasant/ pleasant, poor quality/good quality; α = .90). Aad was measured on four, seven-point scales (bad/good, dislike/like, not irritating/irritating, not interesting/interesting; α = .78). Brand attribute beliefs assessed how realistic subjects felt about the performance of the three brands on each attribute. Subjects rated the likelihood (very unlikely/very likely) that each of the brands had the attribute in question on a fivepoint scale. An overall belief rating was summed and averaged for each brand. Two measures of purchase intentions were used: a four-point scale (definitely would not buy/definitely would buy) and a constant-sum scale where subjects allocated 100 points indicating the likelihood they would buy each of the brands. These intention measures were identical to those of Sawyer and Howard (1991), and were combined to form an overall intention score by calculating a standardized score for the target brand on each of the items and analyzing the mean of the two standardized scores (α = .88).

Covariates

Potential covariates—involvement and product knowledge— were measured to control for the influence of extraneous variables (Hair et al. 1998). Involvement and product knowledge were chosen because previous conclusion explicitness research suggests that these factors may have an effect (Chebat, Charlebois, and Gélinas-Chebat 2001; Kardes 1988; Sawyer and Howard 1991). Involvement was measured on four, sevenpoint scales (unimportant/important, irrelevant to me/relevant to me, means nothing to me/means a lot to me, not needed/ needed). Product knowledge was measured on four, sevenpoint scales (know very little/know very much, inexperienced/ experienced, uninformed/informed, novice buyer/expert buyer).

RESULTS

Manipulation Checks

A conclusion explicitness manipulation check was conducted using two, seven-point scales rating subjects’ level of agreement with the following statements: (1) I think that the advertisement for Samsonic SX ends with an explicit conclusion about which brand is superior, and (2) I think that the advertisement for Samsonic SX ends with an obvious conclusion about which brand is superior (r = .65). It was found that explicit conclusions (Mexplicit = 4.48) were regarded as being more explicit and obvious than implicit conclusions (Mimplicit = 3.89, p < .05). An AQ manipulation check was performed using four, seven-point items (weak/strong, unpersuasive/persuasive, not convincing/convincing, bad/good; α = .93). These results indicate that the AQ manipulation was effective (MstrongAQ = 4.16, MweakAQ = 3.76, p < .05).

Assumption Testing

Prior to examining treatment effects with multivariate analysis of covariance (MANCOVA), a variety of assumptions were tested. First, skewness and kurtosis statistics verified that the assumption of normality was satisfied for the dependent variables and covariates. Second, a nonsignificant Box’s M test confirmed that homogeneity of variance existed among the covariance matrices (Box’s M = 38.50, p > .16). Third, a requirement of covariance analysis is that covariates must be correlated with the dependent variables (Hair et al. 1998). A correlation matrix suggested that involvement was a significant covariate (r > .33, p < .01). However, product knowledge was uncorrelated with any dependent variable (r < .10, p > .13), and was hence excluded from the analysis. Fourth, prior to using MANCOVA it is important to identify any outliers that impact the level of type I error anddistort the results (Hair et al. 1998). An examination of studentized residuals across the dependent variables revealed 13 cases as outliers. Hence, the sample size was reduced to 248 observations.

Hypothesis Testing

Hypothesis 1: Effects for NFC and Conclusion Explicitness Hypothesis 1 predicts that for high-NFC individuals, implicit conclusions will be more effective than explicit conclusions. For low-NFC individuals, no such differences will be evident. A two-way MANCOVA did not produce a significant NFC × conclusion explicitness interaction across any of the dependent measures (Fs < 2.30, ps > .13). To further investigate this result, a planned comparison MANCOVA was run across dependent measures for high-NFC individuals only. This yielded a significant main effect for conclusion explicitness for brand attitudes and purchase intentions. Specifically, for high-NFC individuals, implicit conclusions (Mimplicit = 4.66) result in more favorable brand attitudes than explicit conclusions, Mexplicit = 4.08, F(1, 118) = 7.06, p < .01. A similar pattern is evident for purchase intentions, Mimplicit = .33, Mexplicit = –.11, respectively; F(1, 118) = 5.58, p < .05. Although the means for Aad and brand beliefs were in the predicted direction, they approached but did not reach significance (ps > .07). Furthermore, as expected, an analysis of covariance (ANCOVA) performed for low-NFC individuals yielded no preference for different types of conclusion explicitness in ads (F < 1.02). Thus, these results generally support the hypothesis.

Hypothesis 2: Effects for NFC and AQ

Hypothesis 2 posits that for high-NFC individuals, ads with strong arguments will be more effective than ads with weak arguments. Low-NFC individuals, on the other hand, should not be affected by AQ. Consistent with the hypothesis, for high-NFC individuals, a significant positive main effect for AQ was revealed for brand beliefs, F(1, 118) = 8.68, p < .01, and purchase intentions, F(1, 118) = 10.96, p < .01. It is important to note that the means were in the expected direction, supporting the hypothesis. For example, high-NFC individuals showed more favorable brand beliefs for strong ad arguments (MstrongAQ = 3.92) than for weak arguments (MweakAQ = 3.68). Likewise, for purchase intentions, high-NFC individuals rated more favorable purchase intentions for strong ad arguments (MstrongAQ = .34) than for ads containing weak arguments (MweakAQ = –.08). Yet while significant positive effects were evident for brand beliefs and purchase intentions, the differences for Aad and brand attitudes did not reach significance (Fs < .80, ps > .37). As expected, however, low-NFC individuals showed no preferences for AQ, as a MANCOVA for the low-NFC group revealed no significant main effects or interactions (F < 1.61). Thus, overall there is partial support for this hypothesis.

Hypotheses 3: Effects for NFC, Conclusion Explicitness, and AQ

Hypothesis 3 proposes that high-NFC individuals will prefer ads with implicit conclusions and strong AQ, whereas there will be no effect for low-NFC individuals. A conclusion explicitness × AQ × NFC MANCOVA with involvement as a covariate on all dependent variables revealed significant main effects for NFC, Wilks’s λ = .96, F(4, 236) = 2.76, p <. 05; conclusion explicitness, Wilks’s λ = .95, F(4, 236) = 3.00, p < .05; and AQ, Wilks’s λ = .91, F(4, 236) = 5.97, p < .01. No significant three-way interactions were evident, however. Yet further analysis of high-NFC data revealed a significant positive conclusion explicitness × AQ interaction for brand attitudes, F(1, 118) = 4.30, p < .05, and the result for purchase intentions approached, but did not reach, significance (p = .06).

As can be seen in Table 1, planned contrasts revealed that implicit conclusions (Mimplicit = 4.82) generated more favorable brand attitudes than explicit conclusions (Mexplicit = 3.91, p = .01) for ads with strong arguments, but not for ads with weak arguments. Similarly, for purchase intentions, implicit conclusions (Mimplicit = .77) were more effective than explicit conclusions (Mexplicit = –.01, p < .01) for strong argument ads, but not for weak argument ads. Nevertheless, although the means for Aad and brand beliefs were in the predicted direction, the differences were not statistically significant, and thus can not be considered supportive of the hypotheses. Overall, implicit conclusions used with strong AQ were the most persuasive ads for high-NFC individuals. As expected, no such result was evident for low-NFC individuals (p > .14). Hence, H3 is partially supported.

DISCUSSION

The results of this study offer several contributions to the issue of conclusion explicitness. First, this study is the first to provide empirical evidence that NFC moderates the persuasive impact of conclusion explicitness. Although NFC has been identified as a potential moderating factor in previous research (e.g., Kardes, Kim, and Lim 1994; Sawyer and Howard 1991), its influence on the persuasive impact of conclusion explicitness has not been examined. Although most of our hypotheses received only partial support, our results suggest that implicit conclusions are more effective for high-NFC individuals for brand attitudes and purchase intentions. This suggests that personality trait antecedents can have an influence on conclusion explicitness effects.

 

 

Note: AQ = argument quality; NFC = need for cognition. Standard deviations are in parentheses. a Significant effects represent a statistically significant difference for conclusion explicitness for high-NFC/ strong AQ data, for that dependent variable. * p = .01. ** p < .01

Second, this study examined the effect of AQ. Previous conclusion explicitness research has examined only strong arguments. Hence, it is unknown whether the persuasion advantages of implicit conclusions can be extended to ads containing weaker arguments, such as when a brand advertises outperforming competitors on attributes that consumers deem to be relatively unimportant. The present study hypothesized that implicit conclusions and strong arguments would be more persuasive for high-NFC individuals, owing to the greater perceived validity of a conclusion drawn from a strong argument as opposed to a weak argument. Overall, the findings partially supported this hypothesis for brand beliefs and purchase intentions. Furthermore, the lack of interaction between conclusion explicitness and AQ for Aad suggests that high-NFC individuals evaluate these two factors independently of each other. Indeed, conclusion explicitness does not appear to affect Aad. In a metanalysis of comparative advertising research, however, Grewal and colleagues (1997) found that comparative ads created more negative Aad than noncomparative ads, as they can be viewed as more impersonal and unfriendly. Hence, future research could consider conclusion explicitness and Aad in a noncomparative format.

Third, this study used a complex product category. With the exception of Ahearne, Gruen, and Saxton (2000), the majority of conclusion explicitness studies have focused on simple products. Ahearne, Gruen, and Saxton (2000) suggested that conclusion explicitness effects in advertising do not apply to complex product categories, but methodological factors may have influenced their findings. In contrast, this study found significant conclusion explicitness effects for a complex product. Finally, this study confirms that the persuasion of low-NFC individuals is unaffected by conclusion explicitness or AQ. Although not tested in this research, it could be that low-NFC consumers avoid effortful cognitive tasks and are relatively unmotivated to process arguments. Hence, low-NFC consumers did not respond differently when presented with an implicit conclusion versus an explicit conclusion. These results concur with Sawyer and Howard (1991), who found that an uninvolved audience tended not to draw a conclusion, especially when an explicit conclusion is not provided in the message. Similarly, consistent with previous research (Cacioppo et al. 1986), AQ did not have an effect on low-NFC persuasion.

This study offers several implications that may be of interest to advertisers. First, in designing ads, it may not always be effective to explicitly state the conclusion of the message. An implicit conclusion with strong arguments may cause highNFC consumers to evaluate the advertiser’s brand more favorably and have higher purchase intentions than if an explicit conclusion is used. In other words, if the target market is comprised of high-NFC individuals and the advertising brand has a competitive advantage on the most important attributes (i.e., has strong AQ), then implicit conclusions offer a useful alternative for promotion.

Furthermore, implicit conclusions may prove useful when comparing several competing brands. As comparative advertising can be regarded as an aggressive attack on competitors, using an implicit conclusion may be perceived as less of a hard sell. Practitioners should be aware of the risks with implicit conclusions, however. First, there is the risk that consumers may fail to draw a conclusion. Hence, the ad should be designed to encourage consumers to draw a conclusion. For example, asking consumers to “decide for yourself” or “compare for yourself” should complement an implicit conclusion strategy. Second, consumers may draw the incorrect conclusion. Thus, ad information must be presented in an obvious way to enable consumers to draw the correct conclusion. As such, implicit conclusions are likely to work best in print as opposed to television or radio, since print ads allow consumers sufficient time to process the information and reach a conclusion.

Yet while conclusion explicitness lies well within the control of advertisers as an ad design factor, how can advertisers make use of findings relating to NFC? In answer to this question, NFC offers additional information for market segmentation (Luna and Peracchio 2002), where segments can be classified as high-NFC or low-NFC. How can this be done? One approach is to study the nature of the media vehicle in which the ad is to be placed. A judgment can be made regarding the NFC level of the target market reader, based on preferred content and featured articles. For instance, readers of investment magazines that offer company case studies (e.g., Forbes), should enjoy—or at least process—in-depth information. Such formats tend to be more demanding of cognitive resources (Meyers-Levy and Peracchio 1995). Thus, the readership profile for these magazines may be closer to high-NFC than the readership for less text-based magazines, which focus on, for example, photos of movie celebrities. In the latter case, we could assume a low-NFC readership profile and, thus, use conclusion explicitness accordingly.

This study does have a variety of limitations. First, in addition to the usual limitations associated with student subjects, the study was conducted in an artificial setting, which  may have raised subjects’ involvement levels. Second, although cellular phones represent a complex product, future research might examine other complex products that contain a higher level of risk and complexity (see Darley and Smith 1995 for an application of risk types).

Future research should also study the effects of repetition. Research by Ray and Sawyer (1971) suggests that hard-sell ads perform poorly over repetition relative to soft-sell ads. Thus, research could examine whether the advantage of an implicit conclusion over an explicit conclusion holds over repeated exposures. Furthermore, while this study used a comparative advertising format with the ad conclusion related to a target brand’s superiority, noncomparative formats could be studied, if the ad conclusion is not contingent upon such information (e.g., presenting research on product performance). In addition, this study focused on a verbal manipulation of ambiguity by varying whether or not an explicit conclusion was stated. It would be useful, however, to study visual and audio formats to examine whether similar effects can be found for the effectiveness of implicit over explicit conclusions.

REFERENCES

Ahearne, Michael, Thomas Gruen, and M. Kim Saxton (2000), “When the Product is Complex, Does the Advertisement’s Conclusion Matter?” Journal of Business Research, 48 (1), 55–62. Barone, Michael J., Randall L. Rose, Paul W. Miniard, and Kenneth C. Manning (1999), “Enhancing the Detection of Misleading Comparative Advertising,” Journal of Advertising Research, 39 (September/October), 43–50. Batra, Rajeev, and Douglas M. Stayman (1990), “The Role of Mood in Advertising Effectiveness,” Journal of Consumer Research, 17 (September), 203–214. Baumgartner, Hans (2002), “Toward a Personology of the Consumer,” Journal of Consumer Research, 29 (September), 286– 292. Beardi, Cara (2001), “Zippo’s Eternal Flame,” Advertising Age, 72 (33), 4. Cacioppo, John T., and Richard E. Petty (1982), “The Need for Cognition,” Journal of Personality and Social Psychology, 42 (January), 116–131. ———, ———, and Chuan Feng Kao (1984), “The Efficient Assessment of Need for Cognition,” Journal of Personality Assessment, 48 (9), 306–307. ———, ———, ———, and Regina Rodriguez (1986), “Central and Peripheral Routes to Persuasion: An Individual Difference Perspective,” Journal of Personality and Social Psychology, 51 (5), 1032–1043. Chebat, Jean-Charles, Mathieu Charlebois, and Claire GélinasChebat (2001), “What Makes Open Vs. Closed Conclusion Advertisements More Persuasive? The Moderating Role of Prior Knowledge and Involvement,” Journal of Business Research, 53 (2), 93–102. Darley, William K., and Robert E. Smith (1995), “Gender Dif-ferences in Information Processing Strategies: An Empirical Test of the Selectivity Model in Advertising Response,” Journal of Advertising, 24 (Spring), 41–56. Fine, Bernard J. (1957), “Conclusion-Drawing, Communicator Credibility, and Anxiety as Factors in Opinion Change,” Journal of Abnormal and Social Psychology, 54 (May), 369– 374. Grewal, Dhruv, Sukumar Kavanoor, Edward F. Fern, Carolyn Costley, and James Barnes (1997), “Comparative Versus Noncomparative Advertising: A Meta-Analysis,” Journal of Marketing, 61 (October), 1–15. Griffin, Abbie (1997), “The Effect of Project and Process Characteristics on Product Development Cycle Time,” Journal of Marketing Research, 24 (February), 24–35. Hair, Joseph F., Rolph E. Anderson, Ronald L. Tatham, and William C. Black (1998), Multivariate Data Analysis, 5th ed., Upper Saddle River, NJ: Prentice Hall. Halliday, Jean (2001), “Of Hummers and Zen,” Advertising Age, 72 (32), 29. Hovland, Carl I., Irving L. Janis, and Harold H. Kelley (1953), Communication and Persuasion, New Haven: Yale University Press. Inman, J. Jeffrey, Leigh McAlister, and Wayne D. Hoyer (1990), “Promotion Signal: Proxy for a Price Cut?” Journal of Consumer Research, 17 (1), 74–81. Kardes, Frank R. (1988), “Spontaneous Inference Processes in Advertising: The Effects of Conclusion Omission and Involvement on Persuasion,” Journal of Consumer Research, 15 (September), 225–233. ———, John Kim, and Jeen-Su Lim (1994), “Moderating Effects of Prior Knowledge on the Perceived Diagnosticity of Beliefs Derived from Implicit Versus Explicit Product Claims,” Journal of Business Research, 29 (3), 219–224. Keller, Kevin L., Susan E. Heckler, and Michael J. Houston (1998), “The Effects of Brand Name Suggestiveness on Advertising Recall,” Journal of Marketing, 62 ( January), 48– 57. Luna, David, and Laura A. Peracchio (2002), “‘Where There Is a Will . . .’: Motivation as a Moderator of Language Processing by Bilingual Consumers,” Psychology and Marketing, 19 (July/August), 573–593.Meyers-Levy, Joan, and Laura A. Peracchio (1995), “Understanding the Effects of Color: How the Correspondence Between Available and Required Resources Affects Attitudes,” Journal of Consumer Research, 22 (September), 121–138. Moore, David J., and William D. Harris (1996), “Affect Intensity and the Consumer’s Attitude Toward High Impact Emotional Advertising Appeals,” Journal of Advertising, 25 (Summer), 37–50. Petty, Richard E., and John T. Cacioppo (1981), Attitudes and Persuasion: Classic and Contemporary Approaches, Boulder, CO: Westview Press. ———, Rao H. Unnava, and Alan J. Strathman (1991), “Theories of Attitude Change,” in Handbook of Consumer Behavior, T. S. Robertson and Harold H. Kassarjian, eds., Englewood Cliffs, NJ: Prentice Hall, 241–280. Ray, Michael L., and Alan G. Sawyer (1971), “Repetition in Media Models: A Laboratory Technique,” Journal of Marketing Research, 8 (February), 20–29. Sawyer, Alan G. (1988), “Can There Be Effective Advertising Without Explicit Conclusions? Decide for Yourself?” in Nonverbal Communication in Advertising, Susan Heckler and D. W. Steward, eds., Lexington, MA: D. C. Heath. ———, and Daniel J. Howard (1991), “Effects of Omitting Conclusions in Advertisements to Involved and Uninvolved Audiences,” Journal of Marketing Research, 28 (November), 467–474. Speck, Paul Surgi, and Michael T. Elliott (1997), “Predictors of Advertising Avoidance in Print and Broadcast Media,” Journal of Advertising, 26 (Fall), 61–76. Stayman, Douglas M., and Frank R. Kardes (1992), “Spontaneous Inference Processes in Advertising Effectiveness,” Journal of Consumer Psychology, 1 (2), 125–142. Westphal, Linda (2001), “‘Isn’t the Copy a Bit Long?’ Remarked the Advertiser,” Direct Marketing, 64 (August), 60. Zhang, Yong (1996), “Responses to Humorous Advertising: The Moderating Effect of Need for Cognition,” Journal of Advertising, 15 (1), 15–32. ———,and Richard Buda (1999), “Moderating Effects of Need for Cognition on Responses to Positively Versus Negatively Framed Advertising Messages,” Journal of Advertising, 28 (Summer), 1–15.

 

Self-referencing and consumer evaluations of larger-sized female models: A weight locus of control perspective

Brett A. S. Martin & Ekant Veer & Simon J. Pervan

Received: 28 February 2006 / Accepted: 21 February 2007 /

Published online: 21 March 2007

# Springer Science + Business Media, LLC 2007

 

Abstract In two experiments, we show that the beliefs women have about the controllability of their weight (i.e., weight locus of control) influences their responses to advertisements featuring a larger-sized female model or a slim female model. Further, we examine self-referencing as a mechanism for these effects. Specifically, people who believe they can control their weight (“internals”), respond most favorably to slim models in advertising, and this favorable response is mediated by self-referencing. In contrast, people who feel powerless about their weight (“externals”), self-reference larger-sized models, but only prefer larger-sized models when the advertisement is for a non-fattening product. For fattening products, they exhibit a similar preference for larger-sized models and slim models. Together, these experiments shed light on the effect of model body size and the role of weight locus of control in influencing consumer attitudes.

Keywords Larger-sized models . Self-referencing . Weight locus of control . Brand and advertising attitudes.

 

1 Introduction

A common advertising tactic is to use slim female models in advertisements. Yet in recent times, companies, such as Dove and The Body Shop have used larger-sized female models (LMs). Model body size is important given the changing shape of many consumers in today’s society. For example, 64% of adults in America are overweight or obese, an increase of over 36% since 1980 (National Center for Health Statistics 2005). However, despite research on models as idealized images (e.g., Richins 1991), the systematic examination of LM effects in a marketing context isunder researched. An exception is Peck and Loken (2004) who suggest that female consumers respond positively to LMs in advertising, and that they self-reference LMs when viewing them in advertisements. This highlights the need to study whether the pervasive use of slim female models (SMs) in advertising is the only alternative for marketers.

The purpose of this article is to investigate how female consumers react to print advertising featuring SMs and LMs. Specifically, our research addresses two key questions. First, as suggested by Peck and Loken (2004), do all female consumers wish to see LMs in advertising? Second, what cognitive process underlies a female consumer’s attitude towards SMs and LMs in advertising? We suggest that while some female consumers respond positively to LMs, this view should not be generalized to all female consumers. Specifically, we contribute by showing that the responses of female consumers to model body size is moderated by their individual beliefs regarding their ability to control their own weight (i.e., weight locus of control, WLOC). WLOC (Saltzer 1982) has provided a useful basis for studying how people differ in their perceptions of what constitutes an ideal female body shape (Furnham and Nordling 1998; Saltzer 1982). In addition, we contribute by showing how the cognitive process of self-referencing (Martin et al. 2004) acts as a mediating variable between model body size and attitudes. Our results build on the research of Peck and Loken (2004) by showing that the extent of self-referencing that female consumers engage in when viewing a SM or LM depends upon their weight locus of control beliefs.

 

2 Background and hypotheses

 

2.1 Model body size effects

Research suggests that viewing SMs may have negative effects on female consumers. Richins (1991) found that exposure to highly attractive models in ads resulted in female college students reporting lower satisfaction with their own appearance. Similarly, a meta-analysis of experimental studies shows that young women report more negative body satisfaction after exposure to SMs than other types of models (Groesz et al. 2002). Women are particularly relevant as a consumer group as research shows that weight is regarded by many women as a defining aspect of their value (Grover et al. 2003). Yet Western society has progressively moved towards the use of ever thinner depictions of women as physically attractive (Furnham and Nordling 1998; Weeden and Sabini 2005).

Owing to these concerns, Peck and Loken (2004) examined more realistic model body sizes. They found that exposure to LMs in a context that primed non-traditional stereotypes (i.e., a non-traditional women’s magazine with LMs), resulted in higher ratings of LM attractiveness, than exposure to LMs in a traditional context (i.e., a traditional women’s magazine). Women rather than men also engaged in more positive thoughts when exposed to LMs, and more negative self-relevant thoughts when exposed to SMs, than did men. This research shows that LMs can result in positive effects in advertising. Importantly, LMs were large (sizes 16 to 18) but not obese. Likewise, we study LMs who are heavier than a SM, but who are not obese.1 Instead, the emphasis is on a realistic portrayal of body size.

However, exposure to LMs does not always result in positive effects. In psychology, Mills, Polivy, Herman, and Tiggemann (2002) found that exposure to SMs resulted in dieters reporting a thinner current body size than exposure to LMs. This self-enhancement effect did not extend to non-dieters, but it does suggest that individual differences in weight-related beliefs may offer useful insights to research in this area. More recently, Smeesters and Mandel (2006) found that consumer selfevaluations are enhanced after exposure to moderately slim (but not extremely slim) models, when a free-response measure of self-esteem is used. Yet interestingly, they show that when a rating scale (e.g., a 7-point item) is used to measure self-esteem, consumers provide lower ratings after exposure to SMs. Drawing upon social comparison research (Mussweiler 2003), they suggest that this contrast effect is the result of the ad model being used as a reference point to anchor the scale. Thus, a SM results in females contrasting away from the slim standard and reporting lower subjective ratings of self-esteem. They also recommend that future research into model body size effects examines consumer-oriented variables, such as purchase intention. In this research, we answer this call by studying model size effects in relation to attitudinal variables. In addition, given the importance of weight to assessments of female attractiveness (Weeden and Sabini 2005), we study individual differences in perceptions of weight by examining weight locus of control.

2.2 Weight locus of control

Locus of control (LOC, Rotter 1966) refers to the degree to which a person believes in self-determination, and being able to influence events in their lives through their own actions, such as success through planning (internal LOC), as opposed to their lives being influenced by chance, fate and external influences (external LOC). Research in marketing and psychology has examined the influence of LOC, yet findings for body size-related issues, such as weight management, have been mixed. Consequently, researchers have advocated the use of domain-specific LOC measures, rather than using a general measure (e.g., Holt et al. 2001). A variety of domain-specific measures exist, ranging from parenting LOC, work LOC, to the measure pertinent to this research—weight LOC.

Weight locus of control (WLOC) relates to LOC expectancies regarding an individual’s personal weight (Saltzer 1982). Internals believe their weight is influenced by their own actions, whereas externals believe a person’s body weight is more matter of fate and outside a person’s control. WLOC has provided useful insights regarding actual weight loss. For instance, Saltzer (1982) demonstrated that WLOC was associated with completion of a medical weight loss program. Internals were more likely to succeed with their weight loss goals than externals. In addition to weight loss behavior, WLOC has also shed light on attitudinal responses. Holt et al. (2001) studied attitudes towards health education materials advocating weight management. They found that internals viewed the materials as more informative. Externals tended to believe people are overweight owing to genetics, and from a lack of support from family and friends.

 

Importantly, internals and externals differ in how they view people of different body sizes. Internals place a high emphasis on body shape (Saltzer 1982) and believe that weight is controllable, owing to the effect of a person’s diet and their physical activity (Holt et al. 2001). Thus, internals should prefer SMs, who are a body size which they feel is desirable and achievable. Internals should also have a more negative attitude toward a LM, relative to the SM, given their greater preoccupation with physical appearance and their view that being overweight reflects a lack of effort by an individual (Holt et al. 2001; Tiggemann and Anesbury 2000). Research indicates internals have negative attitudes towards overweight people, since they view body weight as a controllable condition (e.g., Tiggemann and Rothblum 1997). Thus, we expect internals to react more favorably towards the SM.

On the other hand, externals believe that there is nothing they can do to alter their body shape, which they feel is influenced by chance or genetics (Holt et al. 2001). Externals also tend to experience greater body dissatisfaction and feel discriminated against by others regarding their weight (Holt et al. 2001). This dissatisfied, powerless view towards weight, and given the salience of weight to females in society (Grover et al. 2003), and that exposure to slim, attractive models can result in negative feelings towards the self (Richins 1991), suggests that externals may react more favorably to the LM. Indeed, in a study of different ideal body sizes, Furnham and Nordling (1998) found that whereas female internals preferred a slim, buxom figure (i.e., large breasts–small waist–small hips female body shape), female externals preferred a more overweight figure (i.e., the medium breasts–large waist– small hips combination). Therefore, we suggest that externals will evaluate a LM more favorably than a SM.

H1 Weight locus of control and model size will interact to affect attitudes and purchase intent. Specifically, for internals, the use of a slim model leads to a more favorable attitude toward the ad (Aad), brand attitudes (Ab), and purchase intent (PI) than using a larger-sized model. For externals, the use of a larger-sized model leads to a more favorable Aad, Ab and PI than using a slim model.

2.3 Self-referencing as a mediator of attitudes

We expect the findings based on attitudes to show that WLOC moderates model body size effects (H1). Yet such results do not provide insight into the psychological mechanism underlying these effects. We contend that self-referencing offers such a mechanism that provides useful insights into these effects. Self-referencing is defined as a cognitive processing strategy where a consumer relates message information to his or her self structure (Burnkrant and Unnava 1995). From this perspective, the self represents a frequently-used construct in memory that aids the elaboration of encoded information. Hence, self-referenced information is more easily associated with previously stored information.

In marketing, self-referencing has been successfully induced by exposure to pictures of female models. For example, Martin et al. (2004) showed that Asian consumers exhibit greater levels of self-referencing when exposed to print ads featuring an Asian model, as compared to ads featuring a White model. Thus, featuring a self-relevant model in an advertisement can result in consumers spontaneously self-referencing the degree to which they relate to the model in the ad. The affect associated with the self is then transferred to the ad, resulting in positive attitudes (Martin et al. 2004). In addition, self-referencing has been shown to mediate attitudes in response to ads featuring a single model (Martin et al. 2004). We predict that self-referencing will be convergent with the persuasive advantage in H1. Specifically, we expect internals to engage in self-referencing in response to the SM. Externals, with their preference for larger body sizes, should exhibit more self-referencing in response to the LM.

The mediation of WLOC on model size effects by self-referencing is studied using path analysis (Baron and Kenny 1986). Specifically, internals are expected to engage in greater levels of self-referencing when viewing the SM, resulting in a negative association between model body size (dummy variable: SM=0, LM=1), and levels of self-referencing. Since externals should self-reference the LM, they should exhibit a positive model body size-self-referencing association.

H2a For internals, self-referencing acts as a mediator between the effect of the body size of the model on Aad, Ab, and PI. Specifically, model size should be negatively associated with self-referencing, which in turn should be positively associated with Aad, Ab and PI

H2b For externals, model size should be positively associated with self-referencing, which in turn should be positively associated with Aad, Ab and P.

3 Study 1

3.1 Pretests

A pretest (n=61 undergraduates) rated model attractiveness and body size of three SMs or three LMs, which were tested in independent groups (30 and 31 participants, respectively) to avoid body size assimilation-contrast effects. Attractiveness was rated on five 7-point items (Ohanian 1990) and body size on the Pictorial Body Image Scale adapted from Stunkard et al. (1983) which displays thin to large female body shapes (1=slim, 9=large). The selected SM and LM did not differ on attractiveness (p=0.07). Yet, the SM was seen as significantly slimmer than the LM (Mslim=3.81, Mlarge=6.23, F1,59=109.90, p<0.001, 52 =0.64). No gender differences were present (p>0.22). Based on the findings of a separate pretest (n=28) showing that hamburgers were familiar and related to putting on weight, burgers were chosen for the main study.

3.2 Method

3.2.1 Participants, design and procedure

One-hundred and fifty eight female undergraduate business students were randomly assigned to the cells of a 2 (model: large, slim) between subjects design with WLOC (internal, external) used as a measured independent variable, following a median split (Median=4.75, 7-point scale).2 Participants were informed that a study was being conducted on print advertisements. Next, they read a booklet containing an ad and the questionnaire. Participants were asked to read the ad as they would normally do so if reading a magazine. The entire procedure took 15 min to complete. At the conclusion of the data collection, participants were debriefed.

3.2.2 Measures

All measures used 7-point scales. Aad used three items (e.g., good–bad, alpha= 0.87). Ab used three items (e.g., like–dislike, alpha=0.94). Likewise, PI used three items (e.g., likely–unlikely, alpha=0.97). WLOC was measured using the four item scale of Saltzer (1982) which included statements such as, “Whether I gain, lose, or maintain my weight is entirely up to me,” (alpha=0.68).3 Self-referencing was assessed on seven items (e.g., “I can easily relate myself to the advertising model,” alpha=0.91) anchored by strongly disagree–strongly agree. For all multi-item measures, mean scores were calculated and were used in subsequent analyses. Finally, measures were included for fear of fat, dislike of fat people, and willpower (Crandall 1994), as well as measures for attractiveness, expertise, trustworthiness (Ohanian 1990), and pressures to be thin (Netemeyer 1997). However, since they yielded almost no relevant insights they are not discussed any further.

3.3 Results

3.3.1 Manipulation check

Participants rated the ad model’s perceived body size on the Pictorial Body Image Scale. LMs were rated significantly larger (M=6.55) than SMs (M=3.40, F1,155= 488.09, p<0.001, 52 =0.76). No significant main effect or WLOC X Model size interaction were evident (Fs<1).

2 Average age was 21.97 years and mean body mass index (BMI) was 21.38 kg/m2 . The sample can be classified as 16.7% underweight (i.e., 26 participants with a BMI less than 18.5), 67.3% normal weight (105 participants, BMI 18.5 to 24.9), 14.7% overweight (25, BMI 25 to 30) and 1.3% obese (2, BMI over 30). 3 While widely used by researchers, the WLOC scale has reported instances of low reliability (e.g., 0.49 to 0.58 Holt et al. 2001; Saltzer 1982). Thus, we included a related scale on weight control beliefs which used four 7-point items (e.g., “People have control over their weight,” strongly disagree–strongly agree, alpha=0.75), adapted from Tiggemann and Anesbury (2000). Analysis indicated that WLOC was positively correlated with weight control beliefs (r=0.55, p<0.001) as well as with the willpower dimension of the Crandall antifat scale (r=0.28, p<0.01). Study 2 replicated these results (i.e., WLOC— weight control beliefs, r=0.38, p<0.01; WLOC—willpower, r=0.31, p<0.01)

3.3.2 Tests of effects of weight locus of control and model size on attitudes (H1)

A MANOVA revealed a significant WLOC X Model size interaction for Aad (F1,137=4.84, p<0.05, 52 =0.02), Ab (F1,137=3.97, p<0.05, 52 =0.02) and PI (F1,137= 8.56, p<0.01, 52 =0.05). Planned contrasts revealed that internals preferred SMs over LMs (Aad: Ms=3.71 versus 3.00, p=0.01; Ab: Ms=4.08 versus 3.38, p=0.01; PI: Ms=3.83 versus 2.58, p<0.001). In contrast, externals exhibited equal preference for LMs and SMs. (Aad: Ms=3.34 versus 3.18; Ab: Ms=3.51 versus 3.44; PI: Ms=3.28 versus 3.15, Fs<1 for all). The results for internals are consistent with H1, which suggests that they will respond more favorably to SMs, yet the results for externals do not support H1, as they show no specific model body size preference. Thus, there is partial support for H1 (for a summary see Table 1).

3.3.3 Tests of mediation (H2a and H2b)

To test the mediating effect of self-referencing, we conducted regression analyses for internals and externals (Baron and Kenny 1986). First, we regressed Aad on model size. Second, we regressed self-referencing on model size. Third, we regressed Aad on model size and self-referencing. Overall, the results are consistent with H2a and partially consistent with H2b. For internals (H2a), a significant effect for model body size was evident for Aad (b=−0.31, p<0.01), Ab (b=−0.35, p<0.01) and PI (b=−0.47, p<0.001). Model size also had a significant effect on selfreferencing for internals (b=−0.45, p<0.001). Importantly, the effect of model size was reduced or eliminated when self-referencing was included in the model for Aad (b=−0.20, NS), Ab (b=−0.25, NS) and PI (b=−0.29, p<0.05). These results are consistent with H2a. For externals, the effect for model size was not significant (Aad: b=0.06, NS, Ab: b=0.03, NS, and PI: b=0.04, NS). Yet model size resulted in a positive association with self-referencing (b=0.23, p<0.05). Model size had no significant effect when self-referencing was included in the model. These findings offer partial support for H2b.

 

3.4 Discussion

The findings show how model evaluations are influenced by a consumer’s WLOC. Internals respond most favorably to SMs, an effect which is mediated by selfreferencing. In contrast, externals exhibit a similar preference for LMs and SMs, as well as self-referencing in response to the LMs. However, there is an issue that merits further attention. It is possible that the findings are driven by differences in physical weight rather than WLOC. Internals may weigh less than externals, and may prefer slim models who are a similar weight to them. Yet an ANOVA showed that internals and externals did not differ in self-reported weight (p=0.52) or BMI (p> 0.94). Further, when classified as an independent variable by median split (i.e., weight: heavy, slim), no significant Weight X Model size interaction was evident (Aad: F1,126= 1.19, p>0.27, Ab: F1,126=1.66, p=0.20, PI: F1,126=3.27, p>0.07), nor did weight interact with WLOC (Aad: F1,126=0.41, p>0.52, Ab: F1,126=0.03, p>0.86, PI: F1,126= 0.40, p>0.53). These results suggest that differences in physical weight do not represent an alternative explanation for the findings.

While the hypotheses were generally supported, two questions arise: first, are the results generalizable to non-fattening products; and second, does self-referencing merely reflect perceived similarity to the model? For the first question, it could be that internals judge LMs harshly when they are used to advertise a fattening product. In Study 2 we explore product type as a boundary condition to the generalizability of the results of Study 1. With regards to the latter question, in Study 2 we measure perceived similarity and test its overlap with self-referencing. Given that participants may contrast self-evaluations away from a model (Smeesters and Mandel 2006), we also measure appearance self-esteem and normalcy of the model (Bower and Landreth 2001) to provide additional insights.

4 Study 2

Study 2 tests the generalizability of the results found in Study 1 using a nonfattening product. We also measure perceived similarity, appearance self-esteem and perceived normalcy of the model.

4.1 Method

4.1.1 Participants, design, procedure and measures

Eighty seven female undergraduates participated in the study.4 The design, procedure and measures were identical to Study 1. For perceived similarity, participants rated their own body shape on the Stunkard et al. (1983) scale (i.e., “The figure that reflects the way you think you look”). This score was subtracted from the model rating manipulation check. Scores were then reversed and converted to an absolute value (1=low similarity, 9=high similarity). We measured appearance self-esteem on a 7-point scale (Heatherton and Polivy 1991, alpha=0.81). Perceived normalcy was measured on one 7-point item (“I would consider this model to be normal-looking,” strongly agree–strongly disagree) adapted from Bower and Landreth (2001).

4.2 Results

4.2.1 Manipulation check

As intended, LMs were rated as significantly larger in body size (M=6.03) than SMs on the Pictorial Body Image Scale (M=4.09, F1,77=35.17, p<0.001, 52 =0.30).

4.2.2 Weight locus of control and model size on attitudes (H1)

A significant WLOC X Model size interaction was again evident for Aad (F1,62=20.42, p=0.001, 52 =0.23), Ab (F1,62=5.30, p<0.05, 52 =0.06) and PI (F1,62=12.89, p= 0.001, 52 =0.15). Planned contrasts revealed that internals preferred SMs over LMs (Aad: Ms=4.41 versus 2.91, p=0.001; Ab: Ms=4.11 versus 3.49, NS; PI: Ms=4.15 versus 2.82, p<0.01). Importantly, in contrast to Study 1, externals preferred LMs to SMs for ads featuring a non-fattening product (Aad: Ms=4.29 versus 3.54, p<0.03; Ab: Ms=4.21 versus 3.70, NS; PI: Ms=4.27 versus 3.36, p<0.05). Thus, the results for Study 2 support H1 (see Table 1).

4.2.3 Tests of mediation (H2a and H2b)

For internals (H2a), a significant effect for model body size was evident for Aad (b= −0.59, p<0.001) and PI (b=−0.54, p<0.01), but not for Ab (b=−0.30, NS). Model size also had a significant effect on self-referencing for internals (b=−0.53, p<0.01). The effect of model size was reduced or eliminated when self-referencing was included in the model for Aad (b=−0.41, p<0.05), Ab (b=−0.25, NS) and PI (b=−0.31, NS). These findings are consistent with H2a (see Table 2). For externals (H2b), the effect for model body size was generally significant (Aad: b=0.37, p<0.05, Ab: b=0.27, NS, and PI: b=0.34, p<0.05). Further, while model size had a significant positive association with self-referencing (b<0.31, p<0.05), this effect was eliminated when self-referencing was included in the model (Aad: b=0.29, NS; Ab: b=0.20, NS; PI: b=0.22, NS). These findings support H2b.

4.2.4 Secondary analysis

Consistent with Study 1, no differences in weight (p>0.51) and BMI (p=0.71) were evident between internals and externals. Further, when weight was used as an independent variable, no significant main effect (ps>0.69) or interactions with Model size or WLOC were present (ps>0.08), again suggesting that weight does not drive model size effects. Perceived similarity (PS) was associated with selfreferencing (b=0.23, p<0.05). Further, we repeated the mediation analysis using.

 

 

Values shown are standardized coefficients. WLOC Weight locus of control n.s. Not significant (p>0.05) *p<0.05 **p<0.01 ***p<0.001

PS as a mediator rather than self-referencing. This indicated for internals that PS was associated with Aad (b=−0.52, p<0.01), Ab (b=−0.58, p<0.05) and PI (b=−0.59, p< 0.05), and that PS generally reduced the effect of model body size when included as a mediator (Aad: b=−0.44, p<0.05, Ab: b=−0.05, NS, and PI: b=−0.35, p<0.05). Yet the results for externals for all paths involving PS were nonsignificant (ps>0.48, results available by request from the authors) suggesting PS does not act as a mediator for externals.

Consistent with Smeesters and Mandel (2006), a significant main effect was evident for model size on appearance self-esteem (ASE) with participants reporting a lower ASE after viewing a SM (M=3.99) than after viewing a LM (M=4.75, F1,63= 8.67, p<0.01, 52 =0.10). Interestingly, a WLOC X Model size interaction was also present for ASE (F1,63=4.57, p<0.05, 52 =0.05). Planned contrasts revealed that internals reported a lower ASE after viewing the SM (M=3.83) than the LM (M= 5.13, p<0.01). No such effect was evident for externals (MSM=4.15, MLM=4.36, NS). A WLOC X Model size interaction was also present for perceived normalcy (F1,63=10.17, p<0.01, 52 =0.11). Planned contrasts showed that internals view the SM (M=5.46) as more normal looking than the LM (M=3.00, p=0.001). Externals view LMs and SMs as equally normal (MLM=3.75, MSM=3.70, NS).

4.3 Discussion

Study 2 replicates Study 1 using a non-fattening product with two additions. First, the WLOC X Model size interaction involved stronger effect sizes than Study 1 (Aad: 52 =0.23 versus 0.02, Ab: 52 =0.06 versus 0.02, PI: 52 =0.15 versus 0.05). Second, externals preferred LMs over SMs for Aad and PI. Self-referencing again had a mediating effect for internals, and this time also for externals (Table 2) Consistent with Study 1, differences in the weight of participants was not an influential variable. Perceived similarity did correlate with self-referencing and did reflect similar results for internals, yet this variable did not provide insight for externals. SMs did negatively influence ASE (Smeesters and Mandel 2006). However only internals demonstrate this negative effect. Externals appear unaffected. Further, internals regard SMs as normal looking, not LMs, whereas externals make no such distinction.

5 General discussion

The present research shows that considering a female consumer’s weight locus of control, and the extent to which they engage in self-referencing, offers insights into their model body size evaluations. Study 1 showed that internals prefer SMs advertising a fattening product, an effect mediated by self-referencing. In contrast, externals self-referenced LMs but exhibited a similar preference for SMs and LMs. Study 2 showed that these effects generalize to a non-fattening product with stronger effect sizes evident for internals. Externals again self-referenced LMs, but this time preferred LMs over SMs.

This research contributes to LM research in marketing by showing that the persuasive advantage of LMs (Peck and Loken 2004) is not generalizable to all female consumers. Whereas Peck and Loken (2004) document that the priming nontraditional beliefs about women can influence perceptions of attractiveness, our research shows that positive evaluations of LMs can occur without the need for such priming. However, positive evaluations of LMs are restricted to externals who view ads for a non-fattening product. Moreover, in contrast to Peck and Loken, who found females to engage in more self-referencing of LMs than males, we found that the extent of self-referencing engaged in by females depends on their WLOC beliefs. Externals self-reference LMs, internals self-reference SMs.

Indeed, internals judge LMs harshly for both fattening and non-fattening products. Internals believe they can control their own weight, do not regard LMs as normal looking, and yet suffer a decrease in their own appearance self-esteem (ASE) when exposed to a SM. In contrast, externals, who believe a person’s weight is due to fate, appear more accepting of body size and are unaffected in terms of their ASE when exposed to models. This difference in ASE and perceived normalcy of the models represents an intriguing avenue for future research. Indeed, if we assume internals are more likely to be chronic dieters (i.e., unlike externals who do not believe in the efficacy of diets), our findings in this regard contradict Mills et al. (2002) who found that female dieters display self-enhancement after exposure to SMs. Instead, our main effect for model size on ASE confirms the results of Smeesters and Mandel (2006), yet we add to this work by showing a WLOC X Model size interaction on ASE. We speculate that the decrease in ASE for internals after viewing a SM could result from internals seeking to achieve the slimness of the SM through their own efforts. Thus, internals could be motivated by the ad, resulting in favorable ad attitudes, and yet be disappointed that they are not as slim as the SM themselves, thereby resulting in lowered ASE. This interpretation could be explored in future research.

Regarding alternative explanations, our findings are not explained by differences in participant weight, and perceived similarity only offers complementary insights for internals. The positive correlation between self-referencing and perceived similarity concurs with Martin et al. (2004, p. 28) who suggest that self-referencing represents a cognitive process, which can result in a judgment of perceived similarity. However, why did externals prefer LMs advertising salads, but not burgers? Gender research suggests that a female’s visible choice of diet is used in judgment formation. Specifically, females form more favorable impressions of other females who eat non-fattening foods, such as salads (e.g., Mooney and Lorenz 1997). We speculate that this may have influenced evaluations in Study 2, resulting in a persuasive advantage that was not present for externals in Study 1.

A limitation of this research relates to waist-to-hip ratio (WHR) which has been identified as a key influence on perceptions of female attractiveness (Weeden and Sabini 2005). Specifically, the Pictorial Body Image Scale we used conflates WHR with BMI. Further, our LM (WHR=0.83) had a higher WHR than the SM (WHR=0.72), although both are within the typical range of 0.70 to 0.90 for young adult women (Weeden and Sabini 2005). Future research should consider using SMs and LMs that have identical WHRs and explore how different levels of model attractiveness influence consumer responses to SMs and LMs. Another area for future exploration involves examining the antecedents of WLOC beliefs which would offer useful insights to this line of research.

In terms of managerial implications, the current results contradict the view that all female consumers want to see larger-sized models in advertising. Our findings suggest that LMs are only preferred over SMs when advertising non-fattening products to externals. In contrast, where internals are the target market, slim models are more effective, irrespective of product type. Marketers seeking to determine WLOC could use questionnaires or an assessment of the media vehicle’s audience. For example, a fitness magazine being considered for advertising would presumably be read by internals.

Acknowledgements The authors thank Peter Danaher, Bruce Hardie, David Griffith, Cristel Russell, Avi Shankar, the Editors and the anonymous reviewers for helpful comments.

References

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182. Bower, A. B., & Landreth, S. (2001). Is beauty best? Highly versus normally attractive models in advertising. Journal of Advertising, 30(1), 1–12.Burnkrant R. E., & Unnava H. R. (1995). Effects of self-referencing on persuasion. Journal of Consumer Research, 22(June), 17–26. Crandall, C. S. (1994). Prejudice against fat people: Ideology and self-interest. Journal of Personality and Social Psychology, 66(5), 882–894. Furnham, A., & Nordling, R. (1998). Cross-cultural differences in preferences for specific male and female body shapes. Personality and Individual Differences, 25(4), 635–648. Groesz, L. M., Levine, M. P., & Murnen, S. K. (2002). The effect of experimental presentation of thin media images on body dissatisfaction: A meta-analytic review. International Journal of Eating Disorders, 31(1), 1–16. Grover, V. P., Keel, P. K., & Mitchell, J. P. (2003). Gender differences in implicit weight identity. International Journal of Eating Disorders, 34(1), 125–135. Heatherton, T. F., & Polivy, J. (1991). Development and validation of a scale for measuring state selfesteem. Journal of Personality and Social Psychology, 60(June), 895–910. Holt, C. L., Clark, E. M., & Kreuter, M. W. (2001). Weight locus of control and weight-related attitudes and behaviors in an overweight population. Addictive Behaviors, 26(3), 329–340. Martin, B. A. S., Lee, C. K.-C., & Yang, F. (2004). The influence of ad model ethnicity and selfreferencing upon attitudes: Evidence from New Zealand. Journal of Advertising, 33(4), 27–37. Mills, J. S., Polivy, J., Herman, C. P., & Tiggemann, M. (2002). Effects of exposure to thin media images: Evidence of self-enhancement among restrained eaters. Personality and Social Psychology Bulletin, 28(12), 1687–1699. Mooney, K. M., & Lorenz, E. (1997). The effects of food and gender on interpersonal perceptions. Sex Roles, 36(9/10), 639–653. Mussweiler, T. (2003). Comparison processes in social judgment: Mechanisms and consequences. Psychological Bulletin, 110(3), 472–489. National Center for Health Statistics. (2005). Prevalence of overweight and obesity among adults: United States, 1999–2000. http://www.cdc.gov/. National Heart, Lung, and Blood Institute (1998). Clinical guidelines on the identification, evaluation and treatment of overweight and obesity in adults: The evidence report. (NIH Publication No. 98-4083). Bethesda, MD: National Institute of Health. Netemeyer, R. G. (1997). Personal correspondence cited in William O. Bearden and Richard G. Netemeyer (eds.), Handbook of Marketing Scales: Multi-item Measures for Marketing and Consumer Behavior (425). Thousand Oaks, CA: Sage. Ohanian, R. (1990). Construction and validation of a scale to measure celebrity endorsers’ perceived expertise, trustworthiness, and attractive. Journal of Advertising, 19(3), 39–52. Peck, J., & Loken, B. (2004). When will larger-sized female models in advertisements be viewed positively? The moderating effects of instructional frame, gender, and need for cognition. Psychology & Marketing, 21(6), 425–442. Richins, M. L. (1991). Social comparison and the idealized images of advertising. Journal of Consumer Research, 18(June), 71–83. Rotter, J. B. (1966). Generalized expectancies for internal versus external control of reinforcement. Psychological Monographs, 80(1), 1–28. Saltzer, E. B. (1982). The weight locus of control (WLOC) scale: a specific measure for obesity research. Journal of Personality Assessment, 46(6), 620–628. Sawyer, A. G. (1975). Demand artifacts in laboratory experiments in consumer research. Journal of Consumer Research, 1(March), 20–30. Smeesters, D., & Mandel, N. (2006). Positive and negative media image effects on the self. Journal of Consumer Research, 32(March), 576–582. Stunkard, A. J., Sorensen, T., & Schulsinger. F. (1983). Use of the Danish adoption register for the study of obesity and thinness. In S. S. Kety, L. P. Rowland, R. L.Sidman, & S. W. Matthysse (Eds.), Genetics of neurological and psychiatric disorders, (115–120) New York: Raven. Tiggemann, M., & Anesbury, T. (2000). Negative stereotyping of obesity in children: The role of controllability beliefs. Journal of Applied Social Psychology, 30(9), 1977–1993. Tiggemann, M., & Rothblum, E. D. (1997). Gender differences in internal beliefs about weight and negative attitudes towards self and others. Psychology of Women Quarterly, 21(4), 581–593. Weeden, J., & Sabini, J. (2005). Physical attractiveness and health in western societies: A review. Psychological Bulletin, 131(5), 635–653.

 

Using Self-referencing to Explain the Effectiveness of Ethnic Minority Models in Advertising

Christina Kwai-Choi Lee,

Nalini Fernandez and Brett A.S. Martin

University of Auckland Business School, New Zealand

Advertising research has generally not gone beyond offering support for a positive effect where ethnic models in advertising are viewed by consumers of the same ethnicity. This study offers an explanation behind this phenomenon that can be useful to marketers using self-reference theory. Our experiment reveals a strong self-referencing effect for ethnic minority individuals. Specifically, Asian subjects (the ethnic minority group) self-referenced ads with Asian models more than white subjects (the ethnic majority group). However, this result was not evident for white subjects. Implications for academics and advertisers are discussed.

INTRODUCTION

Society is faced with two dramatic and opposing forces. First, there is the advancement in world communication and transportation which has given rise to a global economy, potentially moving people towards a homogenised identity. However, there is an opposing force to this one-world identity, as groups become more aware of their self or group identity on the basis of their ethnic background (Costa & Bamossy 1995). In terms of marketing, there is currently a pressing need in many industries to determine how best to advertise to ethnic minorities (e.g. Bernstel 2000; Liebeskind 2001; Schnuer 2001). Yet media advertising has traditionally assumed that a market with a white ethnic majority and other ethnic minorities can be reached simultaneously. Thus ads for mass audiences have tended to use white models exclusively (Kinra 1997) which is consistent with the melting-pot theory.

This theory suggests that a process of acculturation, resulting from racial and cultural contacts between ethnic minority groups and the host society, eventually results in ethnic minorities becoming more white-like, thereby melting into the larger host society (Kinra 1997). However, researchers such as Rossman (1994) have argued that the trend is towards greater ethnic and cultural diversity.

Culturally distinct segments cannot all be successfully targeted using the same marketing and advertising strategies which succeeded when society was a uniform, Anglo-dominated market (Rossman 1994; Berman 1997; Kim & Kang 2001). Given these opposing views, and the current relevance of ethnicity, the objective of this study is to examine the impact of using ethnic minority models in advertising on the evaluations of ethnic minority and ethnic majority groups.

While past studies have linked ethnicity to consumer behaviour (e.g. Deshpandé & Stayman 1994; Koslow et al. 1994; Lee & Tse 1994; Nwankwo & Lindridge 1998), they have generally confirmed that differences exist between ethnic groups, and that ethnic minority groups may have more positive reactions to ethnic endorsers. The aim of this study is to provide a theoretical basis to understanding this phenomenon that can be useful to marketers who are advertising to an ethnic minority market. We investigate the role of self-referencing as a mechanism for explaining effects which the ethnicity of the model in an advertisement has upon consumers.

SELF-REFERENCING AND ETHNICITY

Self-referencing occurs when a consumer processes information by relating it to some aspect of their self, such as their past experiences. For example, a consumer may see an ad which reminds them of a past holiday they took. Or, as in the case of this study, the ethnicity of the model may result in consumers of the same ethnicity self-referencing the ad information. This perspective assumes that the self is a highly organised, complex memory structure that contains both the semantic and episodic knowledge gained over a lifetime (Burnkrant & Unnava 1995). Thus, when an ethnic minority consumer is exposed to a message that involves a dimension that is central to the self – like a model of the same ethnicity – self-referencing is activated and influences how the message is processed (Rogers et al. 1977;

Krishnamurthy & Sujan 1999). The advantage for marketers is that research suggests that relating information to oneself heightens ad recall and can generate more favourable ad evaluations (Meyers-Levy & Peracchio 1996; Krishnamurthy & Sujan 1999). Consequently, ads are better remembered and better liked by consumers. The rationale for viewing ethnicity as a dimension of the self, that is central to ethnic minority consumers, is based on distinctiveness theory (McGuire & Padawer-Singer 1976; McGuire 1984).

Distinctiveness theory proposes that when a person perceives a complex stimulus, such as their self, they notice distinctive differences and characteristics which have greater informational value in discriminating themselves from others. Thus ethnicity is more salient to the self in an ethnically mixed society than in a uniform one. Further, in an integrated society, ethnicity is more salient to the self-concept of members of the minority group than of the majority group. When an individual’s self is salient, being exposed to information that is consistent with this dimension should result in spontaneous selfreferencing.

Likewise, one of the determinants of an effective advertisement is the source of the message, such as a model endorsing the product, and that source’s familiarity and similarity to the consumer (McGuire 1984; McCracken 1989). In particular, the similarity judgements an individual makes when exposed to an ad, and the ability to picture oneself relative to the model in the ad, results in cognitive activity in the form of spontaneous self-referencing (Debevec & Iyer 1988). Thus H1 suggests that ethnic minority individuals will perceive an ethnic similarity between themselves and the model portrayed in an ad (i.e. the message source), and that this perceived similarity will result in these individuals self-referencing the ad.

H1: Ethnic minority individuals will self-reference advertising portrayals of models of a similar ethnicity. Strength of ethnic identification Ethnicity reflects more than a demographic classification of consumers. Instead, ethnic minorities may differ in the strength of their affiliation with the minority group of their birth. This factor appears to have a significant impact on consumer behaviour (Deshpandé et al. 1986; Stayman & Deshpandé 1989; Deshpandé & Stayman 1994).

For example, ethnic minority individuals who strongly identified with their own ethnic groups had a stronger preference for an advertising spokesperson of their own ethnicity (Williams & Qualls, 1989; Whittler, 1991). It is hypothesised here that the strength of an individual’s identification with his or her own ethnic group moderates the extent to which he or she self-references an ad that portrays a model of his or her own ethnic group. Hence spontaneous selfreferencing of an ethnic role model in an ad should be greater for those individuals who strongly identify with their own ethnic groups, in contrast to those who have a weak sense of ethnic identification. H2: Ethnic minority individuals with a high degree of ethnic identification will express greater self-referencing of an ad containing an ethnic minority model than individuals with a low degree of ethnic identification.

Self-referencing and favourable thoughts

Previous studies suggest that one of the effects of self-referencing is the generation of favourable cognitive responses, or positive thoughts while the ad is being viewed (e.g. Debevec & Iyer 1988; Sujan et al. 1993; Burnkrant & Unnava 1995; Krishnamurthy & Sujan 1999). Cognitive responses reflect an association between the information in a message and information from memory (Cacioppo et al. 1982). Since self-structures are believed to represent a network of cognitive generalisations about the self (Markus 1977; Schouten 1991), activating the self-structures through self-referencing should affect cognitive responses. Thus self-referencing by a consumer of ad information should result in the positive feelings which the consumer associates with his or her self-structure being transferred to the ad. This, in turn, should result in more favourable thoughts about the ad (Sujan et al. 1993; Burnkrant & Unnava 1995). This leads to the following hypothesis:

H3: High self-referencing individuals will exhibit more favourable cognitive responses than low self-referencing individuals. When characters and/or situations portrayed in an ad serve as an explicit cue to stimulate self-referencing, the feelings associated with this self-referencing have been shown to enhance ad evaluations (Sujan et al. 1993). This suggests that individuals who strongly selfreference an ad will have more positive attitudes and purchase.

intentions than those who self-reference in a weaker fashion: H4: Individuals’ attitudes and intentions will be positively influenced by the extent to which they self-reference an ad. High self-referencing individuals should have more positive attitudes and intentions than low self-referencing individuals.

RESEARCH METHOD

Subjects, design and procedure One-hundred and seventy-eight business undergraduates were randomly assigned to a 2 (ad model ethnicity: Asian, white) × 2 (involvement: high, low) between-subjects factorial design with selfreferencing, and strength of ethnic identification used as measured independent variables after median-splitting (Batra & Stayman 1990). Subject ethnicity was also used as an independent variable. The experiment included an involvement manipulation as previous research suggests that self-referencing is more evident under high involvement (Burnkrant & Unnava 1995). Thus high and low involvement product categories were used. Based on the results of a pre-test, watches were chosen for the high involvement conditions and facial tissues for low involvement. However, since there were no differences in the self-referencing scores between the two products in the main study (F < 1), the data for these two products were collapsed and examined as one group. Further, while an overall ‘Asian’ category may comprise a variety of cultures, it is consistent with the Asian/ Anglo dichotomies of previous research and past definitions of Asian (e.g. Taylor & Stern 1997; Aaker 2000; Briley et al. 2000). Regarding procedure, subjects were given a folder consisting of the experimental stimuli with filler ads and the questionnaire. Subjects were instructed to take as long as they wished to examine the print ads. They were then asked to close the booklet of ads, and proceeded to fill in the questionnaire that contained the manipulation checks and dependent measures.

Ad design AD MODEL SELECTION

Previous research indicates that audiences self-reference more when models are attractive rather than average-looking (Debevec & Kernan 1987). Hence, to ensure that model attractiveness did not confound the results, six facial photographs of models (three Asian, three white) were selected from overseas magazines. Sixty students from the target population, but excluded from the main study, then rated model attractiveness. Consequently, a white model and an Asian model with similar high attractiveness scores were chosen. AD STIMULUS All ads used in the experiment were printed in black and white, on magazine-size glossy paper. The facial snapshots of the models were superimposed on to a common body to ensure that body shape and alignment were identical. The two filler ads used contained no models. One showed bottles of mineral water against a background of mountains; the other was an ad for a stereo system. Measures SELF-REFERENCING The extent to which a subject self-referenced an ad was measured using an average of six 5-point scales anchored by: strongly disagree– strongly agree, derived from previous studies (Debevev & Iyer 1988; Burnkrant & Unnava 1995; Meyers-Levy & Peracchio 1996; Krishnamurthy & Sujan 1999). These items included statements such as: ‘I can easily picture myself using the advertised product’, ‘I can easily form similarity judgements between myself and the advertising model’, ‘The ad seemed to be written for me’, and ‘The ad made me think about my own experiences with the product’ (Cronbach’s alpha = 0.84). STRENGTH OF ETHNIC IDENTIFICATION For how strongly a subject identified with their ethnic group, a 5-point scale anchored by very strongly–very weakly from past research (Stayman & Deshpandé 1989; Deshpandé & Stayman 1994; Nwankwo & Lindridge 1998) asked subjects to rate the strength of their ethnic identity.

ATTITUDINAL MEASURES Attitude towards the ad was measured on four 5-point scales anchored by good–bad, likeable–not likeable, unpleasant–pleasant and enjoyable–not enjoyable (α = 0.88). Brand attitudes used four 5-point scales with the anchors good–bad, like–dislike, unpleasant–pleasant and poor quality–good quality (α = 0.68). Attitude towards the model used four 5-point scales anchored by likeable–not likeable, not trustworthy–trustworthy, credible–not credible, high overall effectiveness–low overall effectiveness (α = 0.76). Purchase intentions were measured using two 5-point scales assessing likelihood of purchase, with the anchors very low, very high. COGNITIVE RESPONSES Subjects were instructed to list their thoughts immediately after they had read the ads. Two independent judges coded these thoughts as either positive, negative or neutral. Interjudge agreement was 96% with disagreements resolved through discussion. Following past research, an individual score was determined by subtracting the number of negative thoughts from positive thoughts (Breckler & Wiggins 1991).

RESULT

Self-referencing and ethnicity (H1) H1 predicts that minority group members self-reference more strongly when exposed to models in an ad that are of a similar ethnic minority than they do when exposed to models from the ethnic majority. In accordance with H1, Asian subjects exposed to the Asian model reported higher self-reference scores than white subjects (MAsian = 2.66, Mwhite = 1.87, F(1,89) = 9.18, p < 0.01). Further, there was no difference in self-referencing between Asian and white subjects when exposed to the white model (F(1,89) = 7.17, p > 0.10). Strength of ethnic identification (H2) H2 predicts that ethnic minority individuals who strongly identify with their ethnic group will self-reference more when exposed to an ad containing a model of similar ethnicity than subjects who are weak in ethnic identification. Contrary to expectations, there was no difference.

in the extent of self-referencing between Asians who identified with their ethnicity strongly and those Asians who did not identify with their ethnicity strongly (F(2,79) = 0.85, p > 0.10). Self-referencing effects (H3 and H4) H3 and H4 tested the effects of self-referencing. H3 predicts that high self-referencers have more favourable thoughts than low selfreferencers, while H4 predicts that high self-referencers have more positive attitudes and intentions than low self-referencers. A MANOVA revealed that consistent with H3, high selfreferencers had more favourable thoughts than low self-referencers (Mhigh SR = 1.13, Mlow SR = –0.02, F(1,168) = 8.56, p < 0.01). Further, consistent with H4, self-referencing does appear to influence attitude towards the ad (Mhigh SR = 4.67, Mlow SR = 3.91, F(1,176) = 26.77, p < 0.01), attitude towards the model (Mhigh SR = 4.53, Mlow SR = 4.08, F(1,168) = 10.95, p = 0.01), and purchase intentions (Mhigh SR = 4.54, Mlow SR = 3.80, F(1,168) = 43.77, p < 0.01). However, there was no difference in attitude towards the brand (F(1,168) = 1.28, p > 0.10). These results partially support H4. A similar pattern of results was found when examining the ethnic groups. The Asian subjects who displayed stronger self-referencing than the white sample also had more positive attitudes towards the ad, model in the ad and purchase intentions. Yet there was no difference between the ethnic groups for brand attitudes as displayed in Figure 1.

DISCUSSION AND CONCLUSION This study suggests that when consumers are exposed to advertising that is consistent with a salient dimension of their self, they spontaneously self-reference the ad. This leads to more favourable thoughts, attitudes and purchase intentions. We proposed that ethnicity is one such salient dimension for ethnic minority consumers. Our research shows that when ethnic minority individuals are exposed to an advertising model of the same ethnicity, they spontaneously self-reference the ad information. They also produce more favourable thoughts, more positive attitudes towards the ad, the advertising model and higher purchase intentions. However, while some researchers assert that where individuals strongly identify with their ethnic groups this may affect consumer.

behaviour (Deshpandé & Stayman 1994; Nwankwo & Lindridge 1998), we found a lack of support for the influence of ethnic identification. This lack of support may be due to the student sample which results in a young age demographic. An intriguing area of future research would be to consider the influence of ethnic identification and advertising across a variety of ages and ethnic groups. Interestingly, while it was found that an increase in self-referencing leads to enhanced attitudes and purchase intentions, this increase did not influence brand attitudes. In other words, Asians (who selfreferenced strongly) did not have higher brand attitudes than whites.

Theoretical reasons for these results may be found by studying past self-referencing research. For instance, Sujan et al. (1993) suggest that strong feelings are associated with self-related knowledge structures. Consequently, when an ad is associated with the self structure, these feelings should transfer to the ad, resulting in more favourable cognitive responses and attitudes. However, their study revealed that the carry-over of these feelings to brand judgements was dampened, similar to what was found in this study. They attributed this result to a state of discounting (Schwarz 1990). Discounting occurs when individuals discount or dismiss feeling states as a relevant cue for making unrelated judgements.

It is possible that while Asians did experience more positive states due to self-referencing, as evidenced by the ETHNIC MINORITY MODELS IN ADVERTISING 375 FIGURE 1 ATTITUDES AND PURCHASE INTENTIONS BY ETHNICITY Ethnicity Mean White Asian Ad attitude Model attitude Purchase intentions Brand attitude 4.6 4.5 4.4 4.3 4.2 4.1 4.0 3.9 Lee.qxd 19/07/02 12:11 Page 375 1 Footnote. favourability of cognitive responses, ad attitudes and intentions, when it came to making judgements relating to the advertised brand, they found this positive state to be an irrelevant cue for making brand judgements. This opens up an interesting avenue for future research. Finally, the results revealed that Asian subjects did not selfreference high involvement product categories to a greater extent than low involvement product categories. This may be because selfreferencing is so attention consuming that the increased elaboration, which normally would be prompted by other types of information such as involvement with product category, becomes diluted. Information that is related to the self is of high relevance and a salient distraction from other sources of information.

This suggests that relating information to one’s self is an extremely attention-consuming task that would affect other types of encoding (Baumgartner et al. 1992; Sujan et al. 1993). In summary, while the hypotheses formulated were largely supported, the theoretical reasons why others were not supported provide stimulating possibilities for future research. Advertising implications A primary reason that has been offered for a lack of ethnically diverse faces in marketing stimuli is a fear of negative attitudes from the majority white population or a ‘white backlash’.

So despite the growth of ethnic minority groups in many societies, most advertisers have failed to reflect today’s realities and have tended towards excluding or minimising ethnic minorities from their advertising mix (Dunn 1992; Marshall 1997). This study, however, with its cross-ethnic orientation, found that using ethnic minority models raised the attitudes and purchase intentions of audiences of the same ethnicity without decreasing the attitudes and purchase intentions of the majority ethnic group. Further, the ethnicity of the advertising model had no significant influence on the ad, brand and model attitudes and purchase intentions of the white majority group. While Asians showed more positive attitudes and purchase intentions towards ads that featured Asian models, the white majority’s attitudes and purchase intentions were not significantly influenced by the ethnicity of the advertising model. Thus marketers and media planners, by simply varying the ethnicity of the models featured in promotional materials, may improve their rapport with their target minority groups without endangering their position with the ethnic majority group.

REFERENCES

Aaker, J.L. (2000) Accessibility or diagnosticity? Disentangling the influence of culture on persuasion processes and attitudes. Journal of Consumer Research, 26(4), pp. 340–357. Baumgartner, H., Sujan, M. & Bettman, J.R. (1992) Autographical memories, affect and consumer information processing. Journal of Consumer Psychology, 1, pp. 53–82. Batra, Rajeev & Stayman D.M (1990), The role of mood in advertising effectiveness, Journal of Consumer Research, 17(2), pp. 203-215. Berman, G. (1997) Trend spotting: where the ethnic market is heading. Advertising Age, 68(46), p. S34. Bernstel, J.B. (2000) Courting culturally diverse customers. Bank Marketing, 32(10), pp. 32–36. Breckler, S.J. & Wiggins, E.C. (1991) Cognitive responses in persuasion: affective and evaluative determinants. Journal of Experimental Social Psychology, 27, pp. 180–200. Briley, D.A., Morris, M.W. & Simonson, I. (2000) Reasons as carriers of culture: dynamic versus dispositional models of cultural influence on decision making. Journal of Consumer Research, 27(2), pp. 157–178. Burnkrant, R. & Unnava, R.H. (1995) Effects of self-referencing on persuasion. Journal of Consumer Research, 22(1), pp. 17–26. Cacioppo, J.T., Petty, R.E. & Sidera, J.A. (1982) The effects of a salient selfschema on the evaluation of proattitudinal editorials: top-down versus bottomup message processing. Journal of Experimental Social Psychology, 18, pp. 324–338. Costa, J.A. & Bamossy, G.J. (1995) Marketing in a Multicultural World: Ethnicity, Nationalism and cultural Identity. London: Sage. Debevec, K. & Iyer, E. (1988) Self-referencing as a mediator of the effectiveness of sex-role portrayals in advertising. Psychology & Marketing, 5, pp. 71–84. Debevec, K. & Kernan, J. (1987) Self-referencing as a mediator of the physical attractiveness stereotype. Genetic, Social and General Psychology Monographs, 113(4), pp. 433–460. Deshpandé, R. & Stayman, D.M. (1994) A tale of two cities: distinctive theory and advertising effectiveness. Journal of Marketing Research, 31(1), pp. 57–64. Deshpandé, R., Hoyer, W.D. & Donthu, N. (1986) The intensity of ethnic affiliation: a study of the sociology of Hispanic consumption. Journal of Consumer Research, 13(2), pp. 214–220. Dunn, W. (1992) The move toward ethnic marketing. Nations Business, 80(7), pp. 39–41. Kim, Y.K. & Kang, J. (2001) The effects of ethnicity and product on purchase decision making. Journal of Advertising Research, 41(2), pp. 39–48. Kinra, N. (1997) The communication effectiveness of ethnically orientated advertising. International Journal of Advertising, 16, pp. 221–239. Koslow, S., Shamdasani, P. & Touchstone, E. (1994) Exploring language effects in ethnic advertising. Journal of Consumer Research, 20(4), pp. 575–585. ETHNIC MINORITY MODELS IN ADVERTISING 377 Lee.qxd 19/07/02 12:11 Page 377 1 Footnote. Krishnamurthy, P. & Sujan, M. (1999) Retrospection versus anticipation: the role of the ad under retrospective and anticipatory self-referencing. Journal of Consumer Research, 26(1), pp. 55–69. Lee, W.N. & Tse, D.K. (1994) Changing media consumption in a new home: acculturation patterns among Hong Kong immigrants to Canada. Journal of Advertising, 23(1), pp. 57–70. Liebeskind, K. (2001) Instinct often best multicultural tool. Advertising Age, 72(31), p. S36. McCracken, G. (1989) Who is the celebrity endorser? Cultural foundations of the endorsement process. Journal of Consumer Research, 16(3), pp. 310–321. McGuire, W.J. (1984) Search for the self: going beyond self-esteem and the reactive self. In R.A. Zucker, J. Aronoff & A.I. Rabin (eds) Personality and the Prediction of Behaviour. New York: Academic Press, pp. 73–120. McGuire, W.J. & Padawer-Singer, A. (1976) Trait salience in spontaneous selfconcept. Journal of Personality and Social Psychology, 33(6), pp. 743–754. Markus, H. (1977) Self-schemata and processing information about the self. Journal of Personality and Social Psychology, 35(2), pp. 63–78. Marshall, C. (1997) Adland’s true colours. Management Today, pp. 82–83. Meyers-Levy, J. & Peracchio, L. (1996) Moderators of the impact of self-reference on persuasion. Journal of Consumer Research, 22(4), pp. 408–423. Nwankwo, S. & Lindridge, A. (1998) Marketing to ethnic minorities in Britain. Journal of Marketing Practice: Applied Marketing Science, 4(7), pp. 200–216. Rogers, T.B., Kuiper, N.A. & Kirker, W.S. (1977) Self-reference and the encoding of personal information. Journal of Personality and Social Psychology, 25(9), pp. 677–688. Rossman, M. (1994) Multicultural Marketing: Selling to a Diverse America. New York: AMACOM. Schnuer, J. (2001) DTC marketers eye ethnic media. Advertising Age, 72(12), p. 39. Schouten, J.W. (1991) Selves in transition: symbolic consumption in personal rites of passage and identity reconstruction. Journal of Consumer Research, 17(4), pp. 412–425. Schwarz, N. (1990) Feelings as information: informational and motivational functions of affective states. In E.T. Higgins & R.M. Sorrentino (eds), Handbook of Motivation and Cognition: Foundations of Social Behaviour, Vol. 2, New York: Guilford Press, pp. 527–561. Stayman, D.M. & Deshpandé, R. (1989) Situational ethnicity and consumer behaviour. Journal of Consumer Research, 16(3), pp. 361–371. Sujan, M., Bettman, J.R. & Baumgartner, H. (1993) Influencing consumer judgments using autographical memories: a self-referencing perspective. Journal of Marketing Research, 30(4), pp. 422–436. Taylor, C.R. & Stern, B.B. (1997) Asian-Americans: television advertising and the ‘model minority’ stereotype. Journal of Advertising, 26(2), pp. 47–61. Whittler, T.E. (1991) The effects of actors’ race in commercial advertising: review and extension, Journal of Advertising, 20(1), pp. 54–60. Williams, J.D. & Qualls, W.J. (1989) Middle-class black consumers and intensity of ethnic identification. Psychology & Marketing, 6(4), pp. 263–286. 378 INTERNATIONAL JOURNAL OF ADVERTISING, 2002, 21(3) Lee.qxd 19/07/02 12:11 Page 378 1 Footnote. ABOUT THE AUTHORS Christina Lee is a senior lecturer at the University of Auckland, New Zealand, and is currently Visiting Assistant Professor at Hong Kong University of Science and Technology. Her current research interests include cross-cultural consumer behaviour, family decision-making and social marketing. Nalini Fernandez was a postgraduate with the Department of Marketing, School of Business and Economics, University of Auckland. Brett A.S. Martin is a senior lecturer in marketing at the University of Auckland. His research covers a variety of areas in advertising and consumer behaviour. His current research interests relate to advertising effectiveness, and he has research either published or forthcoming in peer-reviewed journals such as the Journal of Advertising Research and European Journal of Marketing.

How To Stop Binge Drinking and Speeding Motorists: Effects of Relational-Interdependent Self-Construal and Self-Referencing on Attitudes Toward Social Marketing

BRETT A. S. MARTIN1 *, CHRISTINA KWAI-CHOI LEE2 , CLINTON WEEKS1 and MARIA KAYA1 1 QUT Business School, Queensland University of Technology, 2 George Street, Brisbane QLD 4000, Australia 2 School of Business, Monash University, Sunway Campus, Jalan Lagoon Selatan, Bandar Sunway, 46150, Selangor Darul Ehsan Malaysia

 

ABSTRACT

 

How can marketers stop speeding motorists and binge drinking? Two experiments show that the beliefs consumers have about the degree to which they define themselves in terms of their close relationships (i.e., relational-interdependent self-construal (RISC)) offer useful insights into the effectiveness of communications for two key social marketing issues—road safety (Study 1, New Zealand sample) and alcohol consumption (Study 2, English sample). Further, self-referencing is a mechanism for these effects. Specifically, people who define themselves in terms of their close relationships (high-RISCs) respond most favorably to advertisements featuring a dyadic relationship (two people), and this favorable response is mediated by self-referencing. In contrast, people who do not include close relationships in their sense of self (low-RISCs) respond most favorably to self-reference advertisements featuring solitary models. Copyright © 2013 John Wiley & Sons, Ltd.

 

Binge drinking and excessive speeding have both been highlighted as issues that marketing, and specifically advertising, has had mixed or limited success in reducing (Lewis et al., 2007; Wechsler et al., 2003). Preventive health advocacies on these issues both fall within the domain of social marketing. Social marketing seeks to voluntarily influence the behavior of certain consumers for the benefit of society as a whole (Andreasen, 2002). Frequently, social marketing involves an advocacy, for example, calls to stop smoking. Social marketing is challenging because it can involve a short-term cost to consumers (e.g., not smoking) for an intangible long-term gain (e.g., a possibly increased life span).

 

Because of these difficulties, limited academic progress has been made to assist social marketers compared with research on products and services. To this end, this article responds to Andreasen’s (2002) call for investigating strategies to increase the effectiveness of social marketing communications. The study shows that considering a consumer’s relational-interdependent self-construal (RISC) and self-referencing can improve social marketing advertising effectiveness. The purposes of this article are to introduce RISC into the social marketing literature and to explore the role of RISC and self-referencing in terms of social marketing communication effectiveness. RISC relates to an individual’s beliefs about the degree to which they define themselves in terms of their close relationships (Cross et al., 2000).

 

Thus, the study offers three contributions. First, the study shows that consumer responses to social marketing messages are moderated by RISC. Although past consumer research has examined individualism–collectivism (e.g., Gürhan-Canli and Maheswaran, 2000), this research uses RISC, which relates to Western societies and which focuses on a person’s intimate relationships with close others rather than a collective relationship with generalized others (e.g., one’s ethnic group). Second, the study reveals how self-referencing acts as a mediator of these effects. Third, the study examines visual self-referencing manipulations in a social marketing context and introduces a new way of inducing self-referencing through the use of models in advertising, which are depicted in a manner consistent with the target market’s level of RISC.

 

BACKGROUND

Relational-interdependent self-construal Self-construal relates to a person’s self-concept or self-view (Guimond et al., 2006). For RISC, a person’s self-concept is defined in terms of close relationships (Cross et al., 2002). High-RISC individuals are more likely to regard close relationships as important for self-expression and their sense of self. In contrast, lowRISC individuals are less socialized to attend to personal relationships and to consider the needs and wishes of close others (Cross et al., 2003). RISC relates to a person’s close dyadic relationships (e.g., close friend and spouse) rather than a person’s relationship with generalized others (e.g., people of the same ethnicity). Consequently, Cross et al. (2002) suggest that RISC is suited to studies of individuals from Western, individualist cultures where people are more likely to include individual relationships (e.g., best friend) in their sense of self than more general in-groups.

Research indicates that high-RISCs give more importance than low-RISCs to the needs and wishes of signifi- cant others when making decisions (Cross et al., 2000). Further, high-RISCs have a richer cognitive network of relationship associations than low-RISCs. As these associations are part of the self for high-RISCs, they can be more easily primed and result in high-RISCs paying attention to relational stimuli (Cross et al., 2002). Specifically, high-RISCs are more likely to pay attention to information about others’ relationships. For example, the couple (e.g., a married couple) serves as useful memory tool for high-RISCs to implicitly organize information in memory around relationships (Cross et al., 2002). Thus, high-RISCs should be more likely than low-RISCs to pay attention to social relationship cues in advertising.

 

The print ad in Study 1 is about road safety and highlights the dangers of speeding. The ad features a picture of either a solitary male driver, or a driver and his terrified female passenger. Study 2 is about binge drinking featuring a picture of either a solitary male drinking alcohol, or a drinker and his male companion. For high-RISCs, the presence of a companion should remind them of the negative consequences of their actions in relation to people whom they are close to. As they value close relationships, they should be more responsive than low-RISCs to advertisements portraying a companion with the driver or drinker. On the other hand, close relationships are less important for low-RISCs. Thus, they should produce more favorable attitudinal responses toward advertisements portraying a solitary driver or drinker, as this type of ad should help them concentrate on the consequences of their own actions to themselves.

 

H1: RISC and model depiction interact to affect evaluations and behavioral intention. Specifically, high-RISCs have more favorable evaluations and behavioral intention than low-RISCs for exposure to ads showing a model in a dyadic relationship. Low-RISCs have more favorable evaluations and behavioral intention than high-RISCs for exposure to ads portraying a solitary model.

Self-referencing This research posits that the effects of H1 are mediated by self-referencing. Self-referencing is defined as a cognitive processing strategy where consumers relate message information to their individual self structures (Burnkrant and Unnava, 1989, 1995). For example, a consumer may relate ad information to his or her life, such as recalling a similar consumption experience. Self-referenced information is more easily associated with previously stored information because the self is a frequently accessed construct in memory. In marketing, research suggests that self-referencing can be induced by exposure to pictures of models.

 

For example, Martin et al. (2004) find that Asian consumers self-referenced ads showing an Asian model as compared with ads featuring a White model. Further, Martin et al. (2007) find people who believe they could control their own weight show more self-referencing in response to seeing ads featuring slim models, than people who do not believe their weight is within their control. This research suggests that high-RISCs should engage in self-referencing in response to the ad featuring two models. Low-RISCs, with their non-relationship view, should engage in self-referencing in response to advertisements featuring the solitary model.

 

This research uses path analysis to examine the mediation of RISC on ad model depiction effects by selfreferencing (Baron and Kenny, 1986). High-RISCs are expected to self-reference the dyad ad, resulting in a positive association between model depiction (dummy variable: solitary model = 0, dyad = 1), and self-referencing. As low-RISCs should self-reference the solitary model, they should show a negative model depiction-selfreferencing association.

 

H2: Self-referencing acts as a mediator between the effect of model depiction on evaluations and behavioral intention. Specifically,

H2a: Model depiction positively associates with selfreferencing for high-RISC individuals, which in turn associates positively with evaluations and behavioral intention.

H2b: Model depiction negatively associates with selfreferencing for low-RISC individuals, which in turn associates positively with evaluations and behavioral intention.

 

STUDY 1

Choice of social issue

Speeding in terms of road safety is chosen as the social issue for this study on the basis of two criteria. First, speeding is a serious social problem in New Zealand. In 2009, speeding contributed to 100 deaths and 1635 injuries. The social cost of speed-related crashes is estimated at $810 million (New Zealand Ministry of Transport, 2010). Second, speeding is a relevant issue for all road users. For every 100 drivers killed in road crashes from speeding, 59 passengers and 19 other road users die with them (New Zealand Ministry of Transport, 2010).

 

Method

Participants, design, and procedure Two hundred and eighty-five undergraduate business students from a business school in New Zealand were randomly assigned to the cells of a 2 (model depiction: solitary, two people)  2 (ad copy wording: second-person wording, third-person wording) between subjects design. RISC (high and low) is used as a measured independent variable, following a median split (median = 5.18, 7-point scale). Participants were informed that a study was being conducted on print advertisements. Next, they read a booklet containing an ad and the questionnaire. Participants were asked to read the materials at their own pace. The procedure lasted approximately 20 minutes.

 

Advertising stimuli

Model depiction is varied so that one group of participants viewed ads congruent to their RISC levels, whereas other participants viewed RISC-incongruent messages. Although this is the first time that RISC is employed to induce selfreferencing, this practice is consistent with prior research. Meyers-Levy and Peracchio (1996) manipulated audiences’ self-referencing levels by using photos from the perspective of an active participant or as a detached onlooker. In the study, the solitary depiction ad has the male driver traveling alone. In the dyad ads, the driver has a female passenger traveling with him.

The woman is a similar age, and her facial expression suggests that she is terrified. All ads show a young White male driver. An ethnic majority model is used to avoid differences in self-referencing caused by ethnic minority models (Lee et al., 2002; Martin et al., 2004). A male model is shown as the driver, as the majority of speeding drivers are male (85% in 2007 to 2009; New Zealand Ministry of Transport, 2010). A pretest (n = 21 undergraduates) was run to assess whether women would not self-reference a male model driver as highly as men. The results indicate no gender difference in self-referencing (p > 0.15). Ad copy wording is varied so that one group viewed the text worded in the second person and another group viewed the text worded in the third person.

The following text is used: “You are driving (He is driving) along an open road. You don’t (he doesn’t) notice you’ve (he’s) exceeded the speed limit. By the time you notice (he notices), it’s too late.” Ads featuring the passenger also include the sentence: “You don’t (He doesn’t) notice that your passenger is terrified.” The slogan “SPEED KILLS!” appeared in the bottom right-hand side of all ads. An example of the advertising stimuli is shown in Appendix A. As no significant RISC  model  ad wording interaction on self-referencing exists (p > 0.55), ad wording was not pursued any further in this research.

Measures

All measures use 7-point scales. Participants evaluated the ad on three items (bad–good, unconvincing–convincing, and uninteresting–interesting; Cronbach’s a = 0.90). Behavioral intention items are “I am now less likely to speed than I was before seeing the ad” and “I am now more interested in learning about the consequences of speeding than I was before seeing the ad” anchored by strongly disagree to strongly agree (r = 0.76). RISC was measured using the 11-item scale of Cross et al. (2000), which included statements such as “My close relationships are an important reflection of who I am” (a = 0.88, scale items displayed in Appendix B). Self-referencing was assessed on four items (“The ad made me think of my personal experiences in similar scenarios,” “The ad seemed to relate to myself,” “The ad seemed to relate to people who are close to me,” and “I can easily picture myself in the situation portrayed in the ad”; a = 0.85) anchored by strongly disagree to strongly agree. Factor analyses were performed on multi-item measures. As all groups load on single factors, mean scores are calculated and used in the analyses. Measures were also included for cognitive responses, involvement, driving habits, and demographics. However,as these are beyond the scope of this paper, they will not be discussed further.

 

Results

Manipulation checks and confound check First, the analyses support the assumption that model depiction affects self-referencing. An ANOVA on the selfreferencing index revealed a significant RISC  model depiction interaction (F(1, 267) = 7.97, p < 0.05). As expected, high-RISCs display higher self-referencing when viewing two models (M = 3.26) rather than a solitary model (M = 2.9, p = 0.05). In contrast, low-RISCs show higher selfreferencing for the solitary model (M = 3.4) than the ads featuring the dyadic relationship (M = 3.0, p < 0.05). Ad believability is measured using two scales (highly believable–not at all believable and totally acceptable–not at all acceptable; r = 0.68), adapted from Gürhan-Canli and Maheswaran (2000). These scales were averaged to create an ad believability index. ANOVA analyses on the confound check measure show no significant differences (ps > 0.17), suggesting that the advertising stimuli do not differ in believability.

 

Hypothesis testing

Hypothesis 1: A MANOVA revealed a significant interaction for 2 (RISC: high, low) 2 (model: solitary, two people) on evaluations (F(1, 267) = 9.74, p< 0.01) but not for behavioral intention (p> 0.22). Follow-up contrasts for this interaction are consistent with Hypothesis 1 (Table 1). High-RISCs have more favorable evaluations for dyad model ads (M = 3.6) than solitary model ads (M = 3.0, p< 0.01). Low-RISCs evaluate the solitary model ads (M = 3.8) more favorably than dyad ads (M = 3.3, p = 0.05).

Hypothesis 2: The mediating effect of self-referencing on evaluation and behavioral intention was tested using regressions for high-RISCs and low-RISCs (Baron and Kenny, 1986). First, evaluation was regressed on model depiction. Second, self-referencing was regressed on model depiction. Third, evaluation was regressed on model depiction and self-referencing. These analyses were repeated for behavioral intention.

Overall, the results support H2a and partially support H2b. For high-RISCs (H2a), a significant effect for model depiction is evident for evaluations (b = 0.23, p < 0.01) and behavioral intention (b = 0.17, p < 0.05). Model depiction also has a significant effect on self-referencing for high-RISCs (b = 0.17, p < 0.05). Further, the effect of model depiction is eliminated when self-referencing is included in the model for evaluations (b = 0.14, NS) and behavioral intention (b = 0.11, NS). These results support H2a (Table 2). For low-RISCs, the effect for model depiction is significant for evaluations (b = 0.21, p < 0.05), but is not significant for behavioral intention (b = 0.02, NS). Yet model depiction resulted in a negative association with self-referencing (b = 0.19, p < 0.05). The effect of model depiction is eliminated when self-referencing is included model for evaluations (b = 0.12, NS), but has no effect when included for behavioral intention. These results partially support H2b.

 

Discussion

Study 1 shows that RISC and model depiction interact to affect evaluations (H1). Further, self-referencing mediates the effects of RISC and model depiction on attitudes (H2). Study 2 replicates and extends Study 1 by using a different social issue to test the generalizability of the results of Study 1.

 

STUDY 2

Overview

Study 2 replicates and extends Study 1 by using a different social issue and a sample from a different country (England). This study also measure attitudes toward binge drinking as a covariate.

 

Choice of social issue

Alcohol consumption was chosen as a relevant topic for England. In 2009, alcohol was associated with 8664 deaths in the UK (Office for National Statistics, 2011). Further, in England, the social cost of alcohol has been estimated at £20 billion a year (USD$32.19 billion), with binge drinkingby people under the age of 25 years highlighted as a key concern (Cabinet Office Report, 2004). Binge drinking is defined as drinking to get drunk.

 

Method

Participants, design, procedure, and measures A total of 136 undergraduates participated. The design, procedure, and measures are identical to those of Study 1 (RISC: a = 0.91, evaluations: a = 0.81, behavioral intention: r = 0.63, self-referencing: a = 0.87, involvement: a = 0.86). Attitude toward binge drinking is measured on two 7-point scales (like–dislike and desirable–undesirable; r = 0.92).

 

Advertising stimuli

Model depiction was varied in a similar manner to Study 1. The solitary model ads show a male drinking. In the dyad ads, the drinker has a male friend lacking enthusiasm. A pretest (n = 31 undergraduates) found no gender differences in levels of self-referencing (p > 0.49), suggesting that the ad is equally applicable to both sexes. Ad copy wording was varied in a similar manner to Study 1 by using the following text: “You go out (He goes out) most nights. You (He) drink to get drunk. You (He) get so drunk you (he) can’t remember the night before. If you keep (he keeps) this up, you run (he runs) an increased risk of heart disease and brain damage.” Ads featuring the companion included the sentence: “Your (his) friend resents having to cancel their night short because of your (his) behavior.” The slogan “Regular Binge Drinking can Harm You” appeared in all ads (Appendix C). Similar to Study 1, no significant RISC  model  ad wording interaction on self-referencing is present (p > 0.32); therefore, ad wording is not pursued any further in this research.

 

Results

Manipulation checks and confound check An ANCOVA on the self-referencing index with attitude toward binge drinking as a covariate yielded a significant RISC  model interaction (F(1, 116) = 17.19, p < 0.001). Attitude toward binge drinking was used as a significant covariate for these analyses (F(1, 116) = 18.24, p < 0.001) but not for the hypothesis testing where it was not correlated with the dependent variables (p > 0.14). HighRISCs have higher self-referencing for the dyad (M = 4.2) than the solitary model (M = 3.1, p < 0.01). Low-RISCs have higher self-referencing for the solitary model (M = 3.6) rather than the dyad (M = 2.6, p < 0.01). ANOVA analyses on the ad believability confound check measure show no significant difference (ps 0.20), suggesting that the advertising stimuli do not differ in believability.

 

Hypothesis testing

Hypothesis 1: A MANOVA revealed a significant interaction for RISC model on evaluations (F(1, 117) = 6.43, p< 0.05) and behavioral intention (F(1, 117) = 13.51, p< 0.001). Follow-up contrasts generally support Hypothesis 1 (Table 1). High-RISCs have more favorable intentions after viewing the dyad ads (M = 3.4) than the solitary model ads (M = 2.5, p < 0.05). Yet no such difference is evident for evaluations (MDyad = 4.4, MSolitary = 4.0, p > 0.25). Low-RISCs have more favorable intentions after viewing the solitary model ads (M = 4.2) than the dyad ads (M = 3.6, p < 0.05). A similar pattern is evident for evaluations (MSolitary = 3.2, MDyad = 2.20, p < 0.01).

Hypothesis 2: Mediation analyses show for high-RISCs a significant effect for model depiction on behavioral intention (b = 0.31, p < 0.05), but not for evaluations (b = 0.16, p > 0.25). Model depiction affects self-referencing for high-RISCs (b = 0.40, p < 0.01). Further, the effect of model depiction is eliminated when self-referencing is included in the model for behavioral intention (b = 0.25, NS) and remain non-significant for evaluations (b = 0.03, p > 0.80). These results partially support H2a (Table 2). For low-RISCs, model depiction is associated with evaluations (b = 0.27, p < 0.05) and behavioral intention (b = 0.34, p < 0.01). Further, model depiction results in a negative association with self-referencing (b = 0.30, p < 0.05). The effect of model depiction is eliminated when self-referencing is included in the model for both evaluations (b = 0.20, NS) and reduced when included for behavioral intention (b = 0.30, p < 0.05). These results support H2b. Discussion Study 2 replicates Study 1 for alcohol consumption. The predicted RISC  model interaction is supported for both evaluations and behavioral intention. Simlar to Study 1, self-referencing mediates this interaction.

 

GENERAL DISCUSSION

This research examined how a consumer’s RISC and selfreferencing influence their evaluations and behavioral intention regarding social marketing advertisements. The results converged for Studies 1 and 2. High-RISC consumers are more persuaded by ads featuring two models (i.e., a dyadic relationship). Low-RISC participants prefer ads featuring solitary models. Mediation analyses reveal that evaluations for preferred ad types are mediated by self-referencing.

 

Theoretical contributions

 

This research makes several contributions. First, this research is the first to introduce RISC into the field of social marketing. The findings suggest that attitudes are most favorable when RISC and the models depicted in the social marketing ad are congruent, namely, a high-RISC person construing information in an ad featuring models in a dyad, and a low-RISC person construing information in an ad featuring a solitary model. This research also contributes to current social marketing research, which suggests that insights into effectiveness can be gained by considering personal aspects of consumers. For example, Cornelissen et al. (2008) suggested that social marketing messages can be enhanced by the use of cues that relate to personal norms rather than general social norms. This research builds on the research of Cornelissen et al. (2008) by suggesting that social marketing effectiveness can be enhanced by considering a consumer’s personal beliefs regarding close relationships rather than more general relationships. Indeed, RISC congruency is also relevant as research suggests that individual differences offer useful insights into social marketing effectiveness. For example, Christie et al. (1998) discussed the difficulty of having any impact on intentions to binge drink and speculated that individual differences offer a useful perspective for research on alcohol consumption. Likewise, this research shows that a key individual difference that influences evaluations of social marketing advertisements is RISC. Further, the intentions to binge drink can be influenced if RISC is taken into consideration. Thus, scholars researching social marketing should consider the role of RISC when developing social marketing communication. Another key contribution is revealing self-referencing as a mediator of attitudes for preferred ad types. Selfreferencing can favorably influence evaluations when a consumer’s RISC level and the models depicted in an ad are congruent. This research also builds on previous research (e.g., Martin et al., 2004; Meyers-Levy and Peracchio, 1996), suggesting the possibility of inducingself-referencing through the design of an ad. Thus, a self-referencing perspective represents a useful framework for future research in this area.

 

Managerial implications

How then do marketers stop speeding motorists and binge drinking? Our research suggests that the key is the RISC level of the target consumer. Instead of targeting a message only to the individual or to a wider group that has no relevance to the motorist or drinker (e.g., injuring unknown bystanders when speeding), the key is whether highlighting consequences to others who are socially close to the individuals (e.g., close friends) will be useful. Specifically, there are two strategies that social marketers could adopt. First, how people think about close relationships influences the way they react to social marketing messages. For highly relational individuals (high-RISCs), close others are an important part of their self-definition (Cross et al., 2003). Thus, social marketing communications are more effective if more emphasis is placed on interdependence with close others and how the consequences of actions can affect them (e.g., how passive smoking affects your children). On the other hand, lowRISCs should be more effectively targeted by messages focusing on self-relevant information. Second, the influence of self-referencing indicates that social marketers can improve the effectiveness of their advertising by increasing target consumer self-referencing levels. An approach to consider is RISC congruence, where the depiction of the model in the ad (solitary or dyadic) reflects the RISC level of the target consumer. Segmentation for such advertising can be achieved through studying the media or technology habits of consumers. A judgment can be made regarding the RISC level of the target consumer of a media vehicle or website where advertising is being considered on the basis of preferred content and featured articles. For instance, heavy users of social networking sites such as Facebook who keep in contact with close friends on the site are likely to be high-RISC and would suit dyad model ads.

Future research directions

The study examines relational self-construal as an individual difference in two Western cultures; future research may investigate the influence of the collective self and “other referencing” on evaluation and behavioral intention of social marketing communication. Collective self, a predominant self schema in collectivistic, Eastern cultures, extends beyond the individual’s significant other (as in RISC), to encompass the larger social group, for example, the family or the community (Markus and Kitayama, 1991; Triandis, 1989). In such collective cultures, previous research (Cornelissen et al., 2008) that suggest social marketing effectiveness can be enhanced by personal norms rather than social norms may not hold true. For example, Verplanken et al. (2009) find that social norms are a better predictor than personal values when the collective self is primed. Instead of self-referencing, it will be interesting to examine other referencing as the mediator of favorable attitudes. Aaker and Williams (1998) coined the term other referencing to indicate information processing with reference to significant others rather than to self. The authors suggest that the ability to induce other referencing in collectivistic cultures is more effective in evoking favorable attitudes than self-referencing. Social marketers will be interested to determine if ads targeted at collective cultures, which portray that an in-group (e.g. extended family) setting is more likely to induce other referencing and have more favorable attitudes than an ad which portrays only two people or one person. Further, we examined the role of models depicted in ads, but future research could also examine the influence of ad wording such as message framing, which has proven to be a popular technique in advertising (Martin and Marshall, 1999; Pervan and Vocino, 2008). Similarly, RISC could be investigated in relation to product placement where drama featuring relational or independent themes as a context for products could be examined (Pervan and Martin, 2002, 2006). If one assumes that some consumers may automatically adopt a negative view of preventive health advocacies suggesting a change in behavior, an interesting approach could be to examine ways to interrupt this negative view through strategies such as counterstereotypical mental imagery (Martin et al., 2011).

Future research could explore the inclusion of selfregulatory goals as another individual difference, which may affect an individual’s response to social marketing messages. Aaker and Lee (1998) suggested that individuals favor messages that are compatible with their self-regulatory goals, such that those who are interdependent respond more favorably to prevention-focused messages, whereas those who are independent respond more favorably to promotion-focused messages.

The research question could examine whether the ad visual and copy for high-RISC individuals should emphasize the negative consequences of a health warning to one’s family and for low-RISC individuals, whether the ad visual and copy should promote the positive effects of adhering to a health warning to oneself.

An understanding of these differences extends this research to include another element, self-regulatory goals, into consideration when designing social marketing messages. Likewise for research on speeding and binge drinking, future work should consider the role of a consumer’s view of time and the future, such as their temporal orientation as present- or future-oriented (Martin et al., 2009), and how this influences their appreciation of the health consequences of their actions.

 

ACKNOWLEDGEMENTS

The authors thank Echo Guo, Stephen Dann, Jiri Mocicka, and Frank Xu for helpful comments and assistance in data collection and ad design. Clinton Weeks and Maria Kaya contributed equally to the manuscript and should be regarded as equal co-authors.

 

BIOGRAPHICAL NOTES

Brett A. S. Martin is professor of marketing at QUT Business School, Queensland University of Technology, School of Advertising, Marketing and Public Relations. His research focus is on consumer research especially the study of individual differences, role of touch, and the role of fantasy in consumer behavior. His research work has been published in the journals such as Journal of Consumer Research, Marketing Letters, Psychology & Marketing, Journal of Advertising, and the Journal of Advertising Research. More information on his research is available at www.basmartin.com. Christina Kwai-Choi Lee is professor of Marketing at the School of Business, Monash University, Sunway Campus, Malaysia.

 

Her research interests include self and identity, family decision making, sustainable consumption, marketing strategy, and social marketing. She has contributed to conference proceedings and publications in Australian, European, and American journals across disciplines, for example, Journal of Advertising, Journal of Service Research, European Journal of Marketing, Journal of Property Research, Housing Studies, Australasian Marketing Journal, and Studies in Higher Education. Clinton Weeks is currently a lecturer in marketing at the QUT Business School, Queensland University of Technology, Australia. His research focuses on consumer memory and cognition. His work has been published in journals including the Journal of Consumer Research, Journal of Advertising, Psychology & Marketing, and Memory & Cognition.

 

REFERENCES

Aaker JL, Lee AY. 1998. “I” seek pleasures and “We” avoid pains: the role of self-regulatory goals in information processing and persuasion. Journal of Consumer Research 28: 33–49. Aaker JL, Williams P. 1998. Empathy versus pride: the influence of emotional appeals across cultures. Journal of Consumer Research 25: 241–61. Andreasen AR. 2002. Marketing social marketing in the social change marketplace. Journal of Public Policy & Marketing 21(1): 3–13. Baron RM, Kenny DA. 1986. The moderator–mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51(6): 1173–82. Burnkrant RE, Unnava HR. 1989. Self-referencing: a strategy for increasing processing of message content. Personality and Social Psychology Bulletin 15(4): 628–38. Burnkrant RE, Unnava HR. 1995. Effects of self-referencing on persuasion. Journal of Consumer Research 22: 17–26. Cabinet office report. 2004. Alcohol harm reduction strategy for England. Retrieved from http://www.newcastle-staffs.gov.uk Christie J, Fisher D, Kozup JC, Smith S, Burton S, Creyer E. 1998. The effects of bar-sponsored alcohol beverage promotions across binge and nonbinge drinkers. Journal of Public Policy & Marketing 20(2): 240–53. Cornelissen G, Pandelaere M, Warlop L, DeWitte S. 2008. Positive cueing: promoting sustainable consumer behavior by cueing common environmental behaviors as environmental. International Journal of Research in Marketing 25(1): 46–55. Cross SE, Bacon PL, Morris ML. 2000. The relational-interdependent self-construal and relationships. Journal of Personality and Social Psychology 78(4): 791–808. Cross SE, Morris ML, Gore JS. 2002. Thinking about oneself and others: the relational-interdependent self-construal and social cognition. Journal of Personality and Social Psychology 82(3): 399–418. Cross SE, Gore JS, Morris ML. 2003. The relational-interdependent self-construal, self-concept consistency, and well-being. Journal of Personality and Social Psychology 85(5): 933–44. Guimond S, Chatard A, Martinot D, Crisp RJ, Redersdorff S. 2006. Social comparison, self-stereotyping, and gender differences in self-construals. Journal of Personality and Social Psychology 90(2): 221–42. Gürhan-Canli Z, Maheswaran D. 2000. Determinants of country-oforigin evaluations. Journal of Consumer Research 27(1): 96–108. Lee CKC, Fernandez N, Martin BAS. 2002. Using self-referencing to explain the effectiveness of ethnic minority models in advertising. International Journal of Advertising 21: 363–375. Lewis IM, Watson BC, Tay RS, White KM. 2007. The role of fear appeals in improving driver safety: a review of the effectiveness of fear-arousing (threat) appeals in road safety advertising. International Journal of Behavioral and Consultation Therapy 3(2): 203–222. Markus H, Kitayama S. 1991. Culture and self: implications for cognition, emotion and motivation. Psychological Review 98 (2): 224–53. Martin BAS, Marshall R. 1999. The Interaction of message framing and felt involvement in the context of cell phone commercials. European Journal of Marketing 33: 206–218. Martin BAS, Lee CKC, Yang F. 2004. The influence of ad model ethnicity and self-referencing on attitudes. Journal of Advertising 33: 27–37. Martin BAS, Veer E, Pervan SJ. 2007. Self-referencing and consumer evaluations of larger-sized female models: a weight locus of control perspective. Marketing Letters 18(3): 197–209. Martin BAS, Gnoth J, Strong C. 2009. Temporal construal in advertising: the moderating role of temporal orientation and attribute importance upon consumer evaluations. Journal of Advertising 38: 5–19. Martin BAS, Lee MSW, Lacey C. 2011. Countering negative country of origin effects using imagery processing. Journal of Consumer Behaviour 10: 80–92. Meyers-Levy J, Peracchio LA. 1996. Moderators of the impact of self-reference on persuasion. Journal of Consumer Research 22: 408–23. New Zealand Ministry of Transport. 2010. Speed crash facts. Retrieved from http://www.transport.govt.nz/research/Pages/ Speed.aspx Office for National Statistics. Alcohol deaths. 2011. Retrieved from http://www.statistics.gov.uk Pervan SJ, Martin BAS. 2002. Product placement in US and New Zealand television soap operas: an exploratory study. Journal of Marketing Communications 8: 101–113. Pervan SJ, Martin BAS. 2006. Soap operas in New Zealand and the US: product placement strategy and consumption imagery. Advances in Consumer Research 33: 134–137. Pervan SJ, Vocino A. 2008. Message framing: keeping practitioners in the picture. Marketing Intelligence and Planning 26(6): 634–648. Triandis HD. 1989. The self and social behavior in differing cultural contexts. Psychological Review 96: 506–20. Verplanken B, Trafimow D, Khusid IK, Holland RW, Steentjes GM. 2009. Different selves, different values: effects of selfconstruals on value activation and use. European Journal of Social Psychology 39: 909–19. Wechsler H, Nelson TF, Lee JE, Seibring M, Lewis C, Keeling RP. 2003. Perception and reality: a national evaluation of social norms marketing interventions to reduce college students’ heavy alcohol use. Journal of American College Health 64(4): 484–494.

A Stranger’s Touch: Effects of Accidental Interpersonal Touch on Consumer Evaluations and Shopping Time

BRETT A. S. MARTIN

 

This article examines an unexplored area of consumer research—the effect of accidental interpersonal touch (AIT) from a stranger on consumer evaluations and shopping times. The research presents a field experiment in a retail setting. This study shows that men and women who have been touched by another consumer when examining products report more negative brand evaluations, negative product beliefs, less willingness to pay, and spend less time in-store than their control (notouch) counterparts. Our findings indicate that the AIT effect is especially negative for touch from a male stranger for both men (same-sex touch) and women (opposite-sex touch). Directions are provided for future study that highlight potential moderators and process explanations underlying the AIT effect.

 

During shopping, consumers are sometimes accidentally touched by other consumers. Indeed, retail anthropologist Paco Underhill (1999) highlighted an unusual finding based on years of observational data of shopper behavior in his bestselling book Why We Buy. When women are bumped from behind while shopping (accidental interpersonal touch), they are likely to move away from merchandise they are interested in and frequently leave the store. Within marketing, touch research has studied consumers touching products (Krishna and Morrin 2008; Peck and Shu 2009), products touched by other consumers (Argo, Dahl, and Morales 2006), products touching each other (Morales and Fitzsimons 2007), or the trait, need for touch (Peck and Childers 2003).

 

Yet no research has examined accidental interpersonal touch. This is surprising given the variety of contexts where consumers can be accidentally touched by strangers (e.g., attending a busy sale). Prior research suggests intentional touch frequently has Brett A. S. Martin is professor of marketing at the Queensland University of Technology, 2 George Street, Brisbane QLD 4000, Australia (brett.martin@qut.edu.au). The author thanks John Deighton, Debbie MacInnis, Dawn Iacobucci, Sekar Raju, Peter Nuttall, Maria Thompson, Amanda Wanstall, Joe Cannon, Judy Drennan, and the members of the Services Innovation Research Program. Thanks are also extended to the editor, associate editor, and reviewers for the constructive suggestions that enhanced the research. John Deighton and Ann McGill served as editors and Darren Dahl served as associate editor for this article. Electronically published September 12, 2011 a positive effect on consumers.

 

For instance, Hornik (1992b) found that female shoppers who were touched by a confederate, posing as an employee as they entered a store, spent more time in-store. Yet for accidental interpersonal touching, Underhill maintains that when women—and to a lesser extent men—are brushed by another consumer in a store, they are likely to cease considering the product and may leave. This is relevant to practitioners, since many retailers believe that time spent in-store and purchases are positively correlated (Inman, Winer, and Ferraro 2009). The purpose of this article is to demonstrate the accidental interpersonal touch effect (AIT) on consumer evaluations and shopping time using data from a field experiment.

 

We investigate the moderating effect of the gender of the stranger who touches the consumer, and consumer gender. We show that consumers who have been accidentally brushed in a store report less favorable brand evaluations and leave a store earlier than their no-touch counterparts. This AIT effect is more pronounced for consumers touched by male strangers. Psychological mechanisms underlying the AIT effect are not tested; however, a range of potential causes is discussed.

 

BACKGROUND

Touch Effects

Touch is integral to human social life. Touch is our most developed sensory modality at birth (Hertenstein et al. 2006), and prior research has demonstrated a variety of touch effects, ranging from influencing compliance with re-quests (Dolinksi 2010; Gue´gen, Jacob, and Boulbry 2007) to alcohol consumption (Kaufman and Mahoney 1999). One of the most frequently researched topics in touch is gender differences (Gallace and Spence 2008; Stier and Hall 1984). In this research, we propose that females and males respond more negatively to being touched by male strangers. However, since this involves females responding more negatively to opposite-sex touch, and males responding more negatively to same-sex touch, we present the logic for each gender separately.

 

Female Consumers. The present research suggests that female shoppers react negatively to being touched by a female or male stranger. Regarding female touch, prior observational research has documented that touch between females is more common than touch between males (Hall and Veccia 1990; Stier and Hall 1984). However, these studies reflect interactions between personal acquaintances rather than between strangers. In a review of touch research, Thayer (1986) asserts that uninvited touch from a stranger is frequently viewed as intrusive or threatening. We propose that females will respond negatively to touch from a female stranger. The assumption that women may view a stranger’s touch as inappropriate is consistent with research that suggests that the intimacy of a touch for women must match the perceived intimacy of the relationship to be viewed as appropriate (Hertenstein et al. 2006). Since it is likely that no intimacy will exist with a stranger, the implication of this finding is that touch from a female stranger should be viewed as inappropriate.

 

We expect AIT effects for females to be heightened when they are touched by a male stranger. A variety of studies suggest that females do not like being touched by male strangers (Hertenstein et al. 2006; Heslin, Nguyen, and Nguyen 1983). For example, Andersen, Andersen, and Lustig (1987), in a survey of 3,877 undergraduate students from 40 universities, found opposite-sex touch avoidance was higher for females than males. Thus, touch from a male stranger should be viewed negatively by females.

 

A question that naturally arises is how does this research that shows a female aversion to touch, particularly from male strangers, reconcile with touch research in marketing, which shows positive effects for touch, irrespective of touch confederate gender (see table 1). Prior research suggests that a person’s response to touch can be influenced by the context in which the touch occurs and the relationship between individuals (Hertenstein et al. 2006). Typically, the touch confederate in marketing studies performs legitimate service roles such as a store employee, and prior studies indicate that touch in a professional context may be viewed positively (Fisher, Rytting, and Heslin 1976; Hornik 1992a). In contrast, a stranger would be expected to follow societal norms for interpersonal distance and implicit rules of conduct in retailing settings (Grove and Fisk 1997). When these expectations are not met, a negative response is more likely (Hertenstein et al. 2006). In summary, females should regard touch from a stranger negatively, particularly touch from a male stranger.

 

Male Consumers. Prior research indicates that men prefer more interpersonal distance than women (Evans and Howard 1973), and males evaluate unjustified touch negatively (Sussman and Rosenfeld 1978). Thus, men should react unfavorably to interpersonal touch. Yet prior studies suggest that men can welcome touch from female strangers. For example, Heslin et al. (1983) found that men regarded touch from a female stranger to be as pleasant as from a close female friend, whereas women only found oppositesex touch pleasant from a close male friend. Similarly, prior research indicates that men are more likely than women to view social interactions in sexual terms (Levesque, Nave, and Lowe 2006). These studies offer the potential for males to respond favorably to touch from a female stranger.

In this research, we propose that males will respond more negatively to touch from a male stranger. Observational research of people meeting at an airport shows that male samesex pairs are less likely than female same-sex pairs to engage in touching, that male same-sex touching is for a shorter duration than for females, and such touching tends to involve handshakes (Greenbaum and Rosenfeld 1980). Men have also been found to evaluate same-sex touch more negatively than women (Hertenstein et al. 2006). A range of explanations have been offered in the literature for the male aversion to same-sex touch, including uninvited touch indicating higher status or dominance (Greenbaum and Rosenfeld 1980; Roese et al. 1992) and concerns of being perceived as homosexual (Dolinski 2010; Roese et al. 1992)

 

The implication of such findings is that men should respond negatively to being touched by a male stranger. Unlike prior marketing research where the touch confederate adopted a service role, and was proactive in the interaction toward a goal such as choice of meal (Gue´gen et al. 2007; Hornik 1992b), we propose that unexpected touch from a male stranger will be viewed negatively. Thus, like women, men should react more negatively to touch from a male stranger than touch from a female stranger.

H1: Consumer evaluations will be less favorable, and shopping times will be shorter, in response to accidental interpersonal touch than for the no-touch control group.

H2: Consumer evaluations will be less favorable, and shopping times will be shorter, in response to accidental interpersonal touch from a male stranger than a female stranger.

 

METHOD

Design and Overview

One hundred and forty-four participants (72 males, 72 females, age: M p 27.43 years, SD p 8.62 years) were recruited for a field experiment using a 2 (touch: touch, no touch) # 2 (confederate gender: male, female) # 2 (participant gender: male, female) between-subjects factorial de-sign. The experiment was conducted in a city in the southern part of England.

 

Procedure To recruit participants, two research assistants (male and female) approached members of the public in the main shopping district of the city outside the entrance of the store used for the experiment. The store sold a range of bags and suitcases. Participants were informed that a local university was conducting a project on how consumers viewed shopping. Participants were asked to enter the store, shop as they normally would, and give their view of a hip bag displayed in the middle of the store. Participants were told they were free to look at other parts of the store before and after viewing the bag if they wished, as the researchers were interested in their natural shopping behavior. They were offered £5 ($8) to participate.

 

Participants were tested individually and upon their return completed a questionnaire that included dependent measures and an open-ended suspicion probe of the study’s purpose. As the participant was told the study instructions, a confederate inside the store who conducted the touch condition unobtrusively viewed the discussion to recognize the participant.

 

Touch Manipulation

Two trained confederates (male and female, 32 and 34 years, respectively) were used for the touch/no-touch conditions. The touch condition involved a brief light touch, using the side of the confederate’s arm as they brushed past behind the participant as the participant viewed the target product. The confederate ostensibly browsed, looking at other products, before brushing past the participant as the confederate walked at normal pace out of the aisle. The duration of the touch lasted approximately half a second and targeted the middle of the right shoulder blade. Confederate training ensured that touch was applied consistently.

 

We chose the shoulder blade as the target area to build on recent touch research (Levav and Argo 2010) and to avoid intimate areas such as the lower waist or buttocks. We used the side of the confederate’s forearm for the touch (midway between wrist and elbow), rather than the hand. Touch with the hand can indicate a desire to communicate (e.g., poke, tap), as hands are purposive devices that are typically used on objects (Ackerman, Nocera, and Bargh 2010).

 

The no-touch condition was identical to the touch condition, except the confederate walked past in close proximity to participants (approximately 10 centimeters) rather than touching them. The bag was displayed on a wall of similar products, which allowed the confederate to look at products without raising suspicion. Research indicates that a distance of 10 centimeters is perceived as close by English consumers. Remland, Jones, and Brinkman (1995), in an observational study of naturally occurring interactions in shopping areas, train stations, and bus stations, found that Englishpeople maintain an average interpersonal distance of 38.5 centimeters. Thus, a 10 centimeter gap should be perceived as close by the participant. To verify this assumption, a pretest (n p 126) was conducted to measure whether a 10 centimeter gap between a participant and a stranger would be perceived as close by the participant compared with a distance of 40 centimeters.

 

Perceived closeness was tested using a mean of two scales (“not at all close to me/extremely close to me,” “not at all near to me/very near to me”; r p .76). Results showed that a confederate standing 10 centimeters from the participant’s right shoulder was regarded as significantly closer in personal distance (M p 5.30) than a confederate standing at a distance of 40 centimeters from the participant (M p 3.21; F(1, 118) p 83.18, h2 p .41). No differences for confederate or participant gender were found in perceived closeness (p 1 .30)

 

Dependent Variables

Brand evaluation was measured on four 7-point scales anchored by “positive/negative,” “favorable/unfavorable,” “high quality/low quality,” and “good/bad.” Items were averaged to form a brand evaluation index (a p .94). For willingness to pay (WTP), we used the Becker-DeGrootMarschak procedure (1964; see also Prelec and Simester 2001) where participants have the ability to buy the product and have no incentive to overstate or understate their true WTP. First, participants stated their WTP. Second, the price of the bag was decided by calculating a random number of reasonable WTPs. Third, a winner was selected.

 

Fourth, if the WTP stated by the participant was higher than the random number price, then the bag was sold to that participant (bag sold to participant for £10; retail price p £19.99). Product beliefs were measured on four 7-point scales (good material quality, easy to wear, zippered compartment, nice design). Items were averaged to form a product beliefs index (a p .85). Shopping time was measured in seconds from when the participant entered and exited the store. Store evaluation involved four 7-point scales (positive/negative, favorable/ unfavorable, high quality/low quality, and good/ bad). Items were averaged to form a store evaluation index (a p .93).

 

For cognitive responses, participants were asked to list all thoughts that crossed their minds while they were in the store. Two independent judges coded cognitive response data for the total number of thoughts, AIT-related thoughts, product- or brand-related thoughts, and other, irrelevant thoughts, as well as positive, negative, or neutral in valence (, , 0). Interjudge reliability was 95% with discrepancies resolved by discussion.

 

Manipulation and Assumption

Checks As a manipulation check, we used three measures. First, we used AIT-related thoughts. Second, a customer proximity measure (Argo, Dahl, and Manchanda 2005) of three 7-point scales was used, anchored by close/far, near/distant, and next to me/away from me. Items were averaged to form a customer proximity index (a p .90). Third, among a series of filler items of in-store characteristics (e.g., lighting, signs) was a 7-point scale measuring attention paid to other customers in the store, anchored at “paid a lot of attention/paid very little attention.” Spatial confinement was assessed on two reverse-scored 7-point scales anchored by “not wide/ wide” and “narrow/not narrow” from Levav and Zhu (2009). Items were averaged to form a spatial confinement index (r p .73).

 

To test whether familiarity with the brand and/or store may have influenced participant responses, brand familiarity and store familiarity were rated on separate single items (not at all familiar/very familiar 7-point scales). We also assessed confederate attractiveness on two 7-point scales anchored by “attractive/unattractive” and “good looking/not at all good looking.” Items were averaged to form a confederate attractiveness index (r p.79).

 

RESULTS

Manipulation and Assumption Checks

We performed a 2 (touch: touch, no touch) # 2 (confederate: male, female) # 2 (participant: male, female) between-subjects ANOVA on AIT-related thoughts, attention to other customers, customer proximity, spatial confinement, and confederate attractiveness. This revealed only a main effect for touch on AIT-related thoughts with more AITrelated thoughts in the touch relative to the no-touch condition (M p .94 vs. .00; F(1, 136) p 60.75, p ! .001, h2 p .31; see table 2 for means). Touched participants also paid more attention to other customers in the store (M p 4.26) than no-touch participants (M p 3.43; F(1, 38) p 10.14, p ! .01, h2 p .07). Further, no main effect for touch was evident for customer proximity between participants in the touch condition and the no-touch condition (p p .22). No significant differences were evident for spatial confinement (p 1 .19), suggesting that touch and gender differences did not induce differential aisle width perceptions. It is interesting to note that a significant main effect for touch was evident for confederate attractiveness (F(1, 136) p 7.70, p ! .01, h2 p .05) with touched participants rating confederate attractiveness lower (M p 3.70) than participants in the notouch condition (M p 4.19). Further, participant responses to the suspicion probe indicated that they were unaware of the study’s hypotheses.

 

Evaluations and In-Store Shopping Time

We performed a 2 (touch: touch, no touch) # 2 (confederate: male, female) # 2 (participant: male, female) between-subjects ANOVA. This revealed a main effect for touch on brand evaluations, product beliefs, WTP, and shopping time. Consistent with hypothesis 1, the main effect test showed that touched participants rated the brand lower (M p 3.32) than participants in the no-touch condition (M p 4.94; F(1, 136) p 63.41, p ! .001, h2 p .32). There wasalso a main effect for confederate gender on evaluations with male confederates (M p 3.83) resulting in lower evaluations than female confederates (M p 4.36; F(1, 136) p 8.85, p ! .01, h2 p .06). A similar pattern of results was evident for WTP (touch: M p £9.71 vs. £16.10; F(1, 136) p 61.85, p ! .001, h2 p .31; confederate gender: M p £12.10 vs. £13.71; F(1, 136) p 3.94, p ! .05, h2 p .03), product beliefs (touch: M p 3.89 vs. 4.90; F(1, 136) p 21.80, p ! .001, h2 p .14; confederate gender: M p 4.10 vs. 4.68; F(1, 136) p 7.28, p ! .01, h2 p .05), and shopping time (touch: M p 82.26 seconds vs. 157.67 seconds; F(1, 136) p 111.47, p ! .001, h2 p .45; confederate gender: M p 111.60 seconds vs. 128.33 seconds; F(1, 136) p 5.49, p ! .05, h2 p .04).

Consistent with hypothesis 2, this main effect for touch was qualified by a significant touch # confederate gender interaction on evaluations (F(1, 136) p 5.88, p ! .05, h2 p .05). No other interactions were significant (p 1 .17). The touch effect was significant for the male confederate conditions (F(1, 68) p 53.42, p ! .001, h2 p .44) and, to a weaker extent, in the female confederate conditions (F(1, 68) p 15.49, p ! .001, h2 p .19). Participants touched by a male confederate gave lower evaluations (M p 2.78) than those not touched (M p 4.88). Female confederate touch also resulted in lower evaluations (M p 4.99 vs. 3.87). A similar pattern of findings was evident for WTP, product beliefs, and in-store shopping time. A significant touch # confederate gender interaction was evident for WTP (F(1, 136) p 4.64, p ! .05, h2 p .03), product beliefs (F(1, 136) p 4.21, p p .05, h2 p .03), and shopping time (F(1, 136) p 4.10, p ! .05, h2 p .03). Touch had a significant effect for the male confederate conditions (WTP: F(1, 68) p 49.68, p ! .001, h2 p .42; beliefs: F(1, 68) p 22.88, p ! .001, h2 p .25; shopping time: F(1, 68) p 101.01, p ! .001, h2 p .59), where participants touched by males showed less favorable responses than those not touched (WTP: M p £8.03 vs. £16.17; product beliefs: 3.38 vs. 4.83; shopping time: 66.52 seconds vs. 156.53 seconds).

 

This touch effect was present to a lesser degree for female confederate touch for WTP (F(1, 68) p 16.47, p ! .001, h2 p .20) and shopping time (F(1, 68) p 29.93, p ! .001, h2 p .31), but not for product beliefs (F(1, 68) p 3.38, p p .07). As shown in figure 1, female stranger touch resulted in less favorable responses (WTP: M p £11.39 vs. £16.03; shopping time: 97.86 seconds vs. 158.81 seconds). Planned comparisons showed male touch resulted in more negative evaluations than female touch (M p 2.78 vs. 3.87, F(1, 70) p 12.86, p p .001): a pattern that was repeated for WTP (M p £8.03 vs. £11.39; F(1, 70) p 7.79, p ! .01), product beliefs (M p 3.38 vs. 4.40; F(1, 70) p 14.09, p ! .001), and shopping time (M p 66.67 seconds vs. 97.86 seconds; F(1, 70) p 10.17, p ! .01). Overall, touched participants reported less favorable evaluations, WTP, product beliefs, and stayed in the store for a shorter time than no-touch participants. These negative effects were stronger when participants were touched by a male stranger. No differential effects were detected with brand familiarity, store familiarity, and confederate attractiveness as covariates.

 

No significant main effects or interactions were evident for store evaluations (p 1 .22), suggesting the AIT effect is brand-specific and does not also affect global store evaluations. Regarding what may drive the AIT effect, table 3 displays correlations between the variables. This reveals that AIT thoughts (r p .23, p ! .01), negative thoughts (r p .37, p ! .01), and attention to other customers in the store (r p .17, p ! .05) are negatively associated with WTP. A similar pattern of results was evident for brand evaluations and shopping time. Further, touch was positively associated with more AIT thoughts (r p .61, p ! .01), more negative thoughts (r p .74, p ! .01), and negatively associated with perceptions of confederate attractiveness (r p .22, p ! .01). These analyses suggest that AIT results in consumers thinking about the AIT event and doing so in a negative manner.

 

DISCUSSION

This research examined the effect of AIT between consumers on evaluations and in-store shopping time. We investigated whether consumers who had been accidentally touched by a stranger while studying a product would give lower evaluations and leave a store earlier than those who were not touched. To the best of our knowledge, this is the first study to examine how AIT influences evaluations and in-store shopping time. We showed that people touched by a stranger spend less time in the store and evaluate brands more negatively than untouched people. This effect is very strong for both males and females when they are touched by a male stranger. Touch from females is also shown to be negative, including where female strangers touch males and females touch other females. No customer gender differences were identified. How can researchers explain the AIT effect? In the next section, we discuss potential avenues for  future research that may prove useful in better understanding the mechanisms that underlie this phenomenon.

 

Potential Topics for Future Research on Accidental Interpersonal Touch

 

Coping and Cognitive Appraisals. One explanation for the AIT effect is indicated by how touched consumers spend less time in-store than people who are not touched. It is plausible that consumers are distancing themselves from the stranger who touched them. Thus, the AIT effect may involve coping strategies, such as avoidant behaviors to alleviate negative affective states (Luce 1998). For example, Lazarus and Folkman’s (1984) appraisal-based model of coping suggests that cognitive appraisals (how a person construes an event) influence their emotional and coping responses to that event. Cognitive appraisals involve categorizing an event regarding its significance for well-being. Appraisals consist of primary appraisals (e.g., Am I in trouble?) and secondary appraisals (e.g., What can and might be done to manage the situation?). Primary and secondary appraisals influence the amount of stress an individual experiences and their emotional reaction. We speculate that AIT involves primary stress appraisals where the situation of being touched by a stranger is appraised as threatening.

 

Lazarus and Folkman (1984) suggest that threat appraisals are characterized by negative emotions such as anxiety, fear, or anger. Thus, an appraisal perspective could prove useful for future research on AIT. Assuming AIT results in negative emotions, a useful theoretical framework that could assist consumer researchers is the appraisal-tendency framework (Lerner and Keltner 2000). The appraisal-tendency framework examines how different emotions of the same valence (e.g., embarrassment and anxiety, both negative valence) can have different effects on judgments. This framework is consistent with research on emotion that suggests that emotions of the same valence differ in their antecedent appraisals (Lerner and Keltner 2000).

 

An emotion that appears useful for researching AIT is embarrassment. Embarrassment is a negative emotion that results from events that increase the threat of an unwanted evaluation from an audience (Dahl, Manchanda, and Argo 2001). Embarrassment can be generated by inappropriate public behavior by other people with whom one is interacting and can result in distancing behavior from the stressful situation rather than confrontation (Maltby and Day 2000). Hence, it is plausible that AIT results in embarrassment, which motivates consumers to distance themselves from the individual who has touched them. Accidental Interpersonal Touch and the Role of Disgust. Another negative emotion relevant to AIT is disgust, which involves revulsion in response to an offensive object (Rozin and Fallon 1987) or, for AIT, offensive person.

 

Disgust has been highlighted as having powerful effects on consumption (Ariely and Norton 2009). Tybur, Lieberman, and Griskevicius (2009) suggest disgust comprises three types: pathogendisgust (disgust elicited by objects likely to contain infectious agents), sexual disgust (motivating avoidance of potential sexual partners or threats to reproductive success), or moral disgust (motivating avoidance of social norm violators). We speculate that moral disgust (e.g., transgressing norms of interpersonal distance) and sexual disgust (e.g., interpreting AIT as sexually motivated) may drive AIT effects. Further, the disgust is elicited by the consumer being touched, and this affect may transfer to the product being evaluated. Thus, future research should investigate whether disgust plays a central role here.

 

Affect-as-Information. The current research shows that AIT results in more negative brand evaluations, but store evaluations remain unaffected. A possible explanation of these findings that warrants future research is whether consumers use AIT-evoked affect as information about the product. Affect-as-information theory (Schwarz and Clore 1983) suggests that consumers use their current affect as a source of information that directly influences judgments. For example, people can employ a “how do I feel about it?’’ heuristic to infer evaluations from the valence of their feelings (Pham 1998). Thus, the affect-as-information account suggests that negative affect from AIT is misattributed to products being evaluated when the consumer is touched, resulting in affect-congruent judgments.

 

Future research could also explore the potential for positive affect resulting from AIT that would provide information to the consumer. For example, consider hedonic contexts where consumers may expect and even seek AIT, such as at a rock concert or in a football crowd. In such instances, does AIT generate affect-congruent evaluations or is it discounted because of the expected close interpersonal context? Relatedly, we looked at evaluations of an unfamiliar product. Future research should examine the AIT effect where the consumer already holds a positive attitude toward the product. In such instances, consumers may discount the incidental affect, or the affect may transfer to their evaluations of the less familiar store instead of the familiar brand.

 

Arousal. Despite prior research on touch, the notion that interpersonal touch can generate arousal has not been investigated. Shiv (2007) suggests that arousal can provide the mobilizing energy for action tendencies that are recruited by emotion. Hence, building on the discussion of appraisals, arousal can influence the intensity of goals that a person activates from appraisal tendencies in response to a specific emotion. For example, high-arousal anger can result in an individual feeling a stronger urge to punish another individual (Shiv 2007). For AIT, high arousal (vs. baseline arousal) may drive a consumer’s reduced shopping time after AIT. With respect to cognitive processing, arousal disrupts the systematic processing of information by using cognitive resources for appraisal processes (Schachter and Singer 1962) or focusing attention on physical symptoms (Mandler 1975). Thus, if arousal is part of the emotional response to AIT, then systematic processing of informational stimuli would potentially be reduced.

 

Thayer (1989) suggests arousal comprises two factors: tense arousal and energetic arousal. Tense arousal involves a reaction to a real or imagined threat, and attention is focused on the threat stimulus, whereas for energetic arousal, attention is task directed. We speculate that AIT, especially from a male stranger, generates tense arousal, which motivates consumers to distance themselves from the stranger. Thus, future research should consider the role of arousal as a mechanism for AIT effects. Such research can adopt an arousal intensity (high arousal vs. baseline arousal) or arousal component (tense arousal vs. energetic arousal) perspective.

 

Male Same-Sex Touch, Perceived Homosexuality, and Status. A question that arises from this research is, “Why do men react so negatively to same-sex touch?” Research suggests that men are more concerned than women that same-sex touch could result in them being perceived as homosexuals, and hence avoid such behavior (Derlega et al. 1989; Dolinski 2010). Derlega et al. (1989) found that participants viewing photographs of same-sex and opposite-sex pairs (dyads) regarded male same-sex touch as abnormal and sexually motivated. Prior studies also indicate that men hold more negative attitudes toward homosexuals than women and that men with negative attitudes to homosexuals are less comfortable with same-sex touching and engage in less same-sex touching (Roese et al. 1992).

 

Similarly, research on masculinity indicates that men are subject to strong normative pressure to endorse traditional heterosexual norms (Connell 1995). In marketing, the in- fluence of masculine expectations on men has been recognized (Holt and Thompson 2004). Martin and Gnoth (2009) suggest that males engage in normative masculine behavior when their collective self is salient, based on concerns of how they will be viewed by other men. Further, they show that male expectations of how they will be classified by others in a collective-self (vs. private-self) context mediate how men respond to male models in advertising.

 

If we assume that shopping can occur in a collective-self context, then AIT research addressing self-construal, classification expectations, and attitudes toward homosexuality warrant further investigation. Further, a perspective useful for future research relates to status. Henley (1977) suggests that touch communicates status and dominance. Reciprocal touch communicates closeness, but nonreciprocal touch communicates status. Greenbaum and Rosenfeld (1980) argue that male same-sex interactions are strictly governed by concerns of status. Males are socialized to use touches that suggest equal status.

 

They suggest the handshake is symbolic of equal status and can allay dominance threats that can occur between males. This area represents an intriguing approach for consumer researchers to assess whether dominance threat appraisals drive male responses to same-sex AIT. The Role of Apology. An interesting avenue for future research relates to apology. Apologies can counter the perception that a transgression is linked to an underlying dis-position of the offender and can recognize that harm has been done to another person (Risen and Gilovich 2007). Effects of apologies include reduced avoidance behavior (McCullough, Worthington, and Rachal 1997), reduced negative affect toward a transgressor (Ohbuchi, Kameda, and Agarie 1989), and increased forgiveness (McCullough et al. 1997). However, prior research has often adopted a dichotomous perspective of apology (apology vs. no-apology conditions). An exception is Fehr and Gelfand (2010), who examined self-construal and apology type. They found that independent people prefer apologies that emphasize compensation (e.g., an offer to restore equity).

 

In contrast, relational people respond to empathy (e.g., demonstrating concern for one’s suffering), and people with a collective self-construal respond to apologies that acknowledge that the transgressor has violated group norms. Relatedly, work on apologies and forgiveness suggests people from collectivist societies view the behavior of others as being influenced more by cultural norms than by their own personal choice (Takaku, Weiner, and Ohbuchi 2001). Thus, considering self-construal and norm salience could offer useful insights to AIT research. In conclusion, the present research identifies a phenomenon that seeds a number of directions for future enquiry. Our research shows that the accidental interpersonal touch effect can have important consequences for brand evaluations, willingness to pay, product beliefs, and in-store shopping time. We hope that the present research motivates others to explore this intriguing area.

 

REFERENCES

Ackerman, Joshua M., Christopher C. Nocera, and John Bargh (2010), “Incidental Haptic Sensations Influence Social Judgments and Decisions,” Science, 328 (June), 1712–15. Andersen, Janis F., Peter A. Andersen, and Myron W. Lustig (1987), “Opposite Sex Touch Avoidance: A National Replication and Extension,” Journal of Nonverbal Behavior, 11 (2), 89–109. Argo, Jennifer J., Darren W. Dahl, and Rajesh V. Manchanda (2005), “The Influence of a Mere Social Presence in a Retail Context,” Journal of Consumer Research, 32 (September), 207–12. Argo, Jennifer J., Darren W. Dahl, and Andrea C. Morales (2006), “Consumer Contamination: How Consumers React to Products Touched by Others,” Journal of Marketing, 70 (April), 81–94. Ariely, Dan, and Michael I. Norton (2009), “Conceptual Consumption,” Annual Review of Psychology, 60, 475–99. Becker, Gary S., Morris H. DeGroot, and Jacob Marschak (1964), “Measuring Utility by a Single-Response Sequential Method,” Behavioral Science, 9 (3), 226–32. Connell, Raewyn W. (1995), Masculinities, Cambridge: Polity. Crusco, April H., and Christopher G. Wetzel (1984), “The Midas Touch: The Effects of Interpersonal Touch on Restaurant Tipping,” Personality and Social Psychology Bulletin, 10 (4), 512–17. Dahl, Darrenn W., Rajesh V. Manchanda, and Jennifer J. Argo (2001), “Embarrassment in Consumer Purchase: The Roles of Social Presence and Purchase Familiarity,” Journal of Consumer Research, 28 (December), 473–81. Derlega, Valerian J., Robin J. Lewis, Scott Harrison, Barbara A. Winstead, and Robert Costanza (1989), “Gender Differences in the Initiation and Attribution of Tactile Intimacy,” Journal of Nonverbal Behavior, 13 (2), 83–96. Dolinski, Dariusz (2010), “Touch, Compliance, and Homophobia,” Journal of Nonverbal Behavior, 34 (3), 179–92. Erceau, Damien, and Nicolas Gue´gen (2007), “Tactile Contact and Evaluation of the Toucher,” Journal of Social Psychology, 147 (4), 441–44. Evans, Gary W., and Roger B. Howard (1973), “Personal Space,” Psychological Bulletin, 80 (4), 334–44. Fehr, Ryan, and Michele J. Gelfand (2010), “When Apologies Work: How Matching Apology Components to Victims’ SelfConstruals Facilitates Forgiveness,” Organizational Behavior and Human Decision Processes, 113 (September), 37–50. Fisher, Jeffrey D., Marvin Rytting, and Richard Heslin (1976), “Hands Touching Hands: Affective and Evaluative Effects of an Interpersonal Touch,” Sociometry, 39 (4), 416–21. Gallace, Alberto, and Charles Spence (2008), “The Science of Interpersonal Touch: An Overview,” Neuroscience and Biobehavioral Reviews, 34 (2), 246–59. Greenbaum, Paul E., and Howard M. Rosenfeld (1980), “Varieties of Touching in Greetings: Sequential Structure and Sex-Related Differences,” Journal of Nonverbal Behavior, 5 (1), 13–25. Grove, Stephen J., and Raymond P. Fisk (1997), “The Impact of Other Customers on Service Experiences: A Critical Incident Examination of ‘Getting Along,’” Journal of Retailing, 73 (1), 63–85. Gue´gen, Nicolas, and Ce´line Jacob (2005), “The Effect of Touch on Tipping: An Evaluation in a French Bar,” Hospitality Management, 24 (2), 295–99. Gue´gen, Nicolas, Ce´line Jacob, and Gae¨lle Boulbry (2007), “The Effect of Touch on Compliance with a Restaurant’s Employee Suggestion,” Hospitality Management, 26 (4), 1019–23. Hall, Judith A., and Ellen M. Veccia (1990), “More ‘Touching’ Observations: New Insights on Men, Women, and Interpersonal Touch,” Journal of Personality and Social Psychology, 59 (6), 1155–62. Henley, Nancy M. (1977), Body Politics: Power, Sex, and Nonverbal Communication, Englewood Cliffs, NJ: Prentice Hall. Hertenstein, Matthew J., Julie M. Verkamp, Alyssa M. Kerestes, and Rachel M. Holmes (2006), “The Communicative Functions of Touch in Humans, Nonhuman Primates, and Rats: A Review and Synthesis of the Empirical Research,” Genetic, Social, and General Psychology Monographs, 132 (1), 5–94. Heslin, Richard, Tuan D. Nguyen, and Michele L. Nguyen (1983), “Meaning of Touch: The Case of Touch from a Stranger or Same Sex Person,” Journal of Nonverbal Behavior, 7 (3), 147–57. Holt, Douglas B., and Craig J. Thompson (2004), “Man-of-Action Heroes: The Pursuit of Heroic Masculinity in Everyday Consumption,” Journal of Consumer Research, 31 (September), 425–40. Hornik, Jacob (1992a), “Effects of Physical Contact on Customers’ Shopping Time and Behavior,” Marketing Letters, 3 (1), 49–55. ——— (1992b), “Tactile Stimulation and Consumer Response,” Journal of Consumer Research, 19 (December), 449–58. Inman, J. Jeffrey, Russell S. Winer, and Rosellina Ferraro (2009), “The Interplay among Category Characteristics, Customer Characteristics, and Customer Activities on In-Store Decision Making,” Journal of Marketing, 73 (September), 19–29.Kaufman, Douglas, and John M. Mahoney (1999), “The Effect of Waitresses’ Touch on Alcohol Consumption in Dyads,” Journal of Social Psychology, 139 (3), 261–67. Krishna, Aradhna, and Maureen Morrin (2008), “Does Touch Affect Taste? The Perceptual Transfer of Product Container Haptic Cues,” Journal of Consumer Research, 34 (April), 807– 18. Lazarus, Richard S., and Susan Folkman (1984), Stress, Appraisal, and Coping, New York: Springer. Lerner, Jennifer S., and Dacher Kletner (2000), “Beyond Valence: Toward a Model of Emotion-Specific Influences on Judgement and Choice,” Cognition and Emotion, 14 (4), 473–93. Levav, Jonathan, and Jennifer J. Argo (2010), “Physical Contact and Financial Risk-Taking,” Psychological Science, 21 (6), 804–10. Levav, Jonathan, and Rui (Juliet) Zhu (2009), “Seeking Freedom through Variety,” Journal of Consumer Research, 36 (December), 600–610. Levesque, Maurice J., Christopher S. Nave, and Charles A. Lowe (2006), “Toward an Understanding of Gender Differences in Inferring Sexual Interest,” Psychology of Women Quarterly, 30 (2), 150–58. Luce, Mary Frances (1998), “Choosing to Avoid: Coping with Negatively Emotion-Laden Consumer Decisions,” Journal of Consumer Research, 24 (March), 409–33. Lynn, Michael, Joseph-Mykal Le, and David Sherwyn (1998), “Reach Out and Touch Your Customer,” Cornell HRA Quarterly, 39 (3), 60–65. Maltby, John, and Liza Day (2000), “The Reliability and Validity of a Susceptibility to Embarrassment Scale among Adults,” Personality and Individual Differences, 29 (4), 749–56. Mandler, George (1975), Mind and Emotion, New York: Wiley. Martin, Brett A. S., and Juergen Gnoth (2009), “Is the Marlboro Man the Only Alternative? The Role of Gender Identity and Self-Construal Salience in Evaluations of Male Models,” Marketing Letters, 20 (4), 353–67. McCullough, Michael E., Everett L. Worthington, and Kenneth C. Rachal (1997), “Interpersonal Forgiving in Close Relationships,” Journal of Personality and Social Psychology, 73 (2), 321–36. Morales, Andrea C., and Gavan J. Fitzsimons (2007), “Product Contagion: Changing Consumer Evaluations through Physical Contact with ‘Disgusting’ Products,” Journal of Marketing Research, 44 (May), 272–83. Ohbuchi, Ken-ichi, Masuyo Kameda, and Nariyuki Agarie (1989), “Apology as Aggression Control: Its Role in Mediating Appraisal of and Response to Harm,” Journal of Personality and Social Psychology, 56 (2), 219–27. Peck, Joann, and Terry L. Childers (2003), “Individual Differences in Haptic Information Processing; The ‘Need for Touch’ Scale,” Journal of Consumer Research, 30 (December), 430–42. Peck, Joann, and Suzanne B. Shu (2009), “The Effect of Mere Touch on Perceived Ownership,” Journal of Consumer Research, 36 (October), 434–47. Pham, Michel Tuan (1998), “Representativeness, Relevance, and the Use of Feelings in Decision Making,” Journal of Consumer Research, 25 (September), 144–60. Prelec, Drazen, and Duncan Simester (2001), “Always Leave Home Without It: A Further Investigation of the Credit-Card Effect on Willingness to Pay,” Marketing Letters, 12 (1), 5–12. Remland, Martin S., Tricia S. Jones, and Heidi Brinkman (1995), “Interpersonal Distance, Body Orientation, and Touch: Effects of Culture, Gender, and Age,” Journal of Social Psychology, 135 (June), 281–97. Risen, Jane L., and Thomas Gilovich (2007), “Target and Observer Differences in the Acceptance of Questionable Apologies,” Journal of Personality and Social Psychology, 92 (3), 418– 33. Roese, Neal J., James M. Olson, Marianne N. Borenstein, Angela Martin, and Alison L. Shores (1992), “Same-Sex Touching Behavior: The Moderating Role of Homophobic Attitudes,” Journal of Nonverbal Behavior, 16 (4), 249–59. Rozin, Paul, and April E. Fallon (1987), “A Perspective on Disgust,” Psychological Review, 94 (1), 23–41. Schachter, Stanley, and Jerome E. Singer (1962), “Cognitive, Social, and Physiological Determinants of Emotional States,” Psychological Review, 69 (5), 379–99. Schwarz, Norbert, and Gerald L. Clore (1983), “Mood, Misattribution, and Judgments of Well-Being: Informative and Directive Functions of Affective States,” Journal of Personality and Social Psychology, 45 (3), 513–23. Shiv, Baba (2007), “Emotions, Decisions, and the Brain,” Journal of Consumer Psychology, 17 (3), 174–78. Smith, David E., Joseph A. Gier, and Frank N. Willis (1982), “Interpersonal Touch and Compliance with a Marketing Request,” Basic and Applied Social Psychology, 3 (1), 35–38. Stephen, Renee, and Richard L. Zweigenhaft (1986), “The Effect on Tipping of a Waitress Touching Male and Female Customers,” Journal of Social Psychology, 126 (1), 141–42. Stier, Deborah S., and Judith A. Hall (1984), “Gender Differences in Touch: An Empirical and Theoretical Review,” Journal of Personality and Social Psychology, 47 (2), 440–59. Sussman, Nan M., and Howard M. Rosenfeld (1978), “Touch, Justi- fication, and Sex: Influences of the Aversiveness of Spatial Violations,” Journal of Social Psychology, 106 (2), 215–25. Takaku, Seiji, Bernard Weiner, and Ken-ichi Ohbuchi (2001), “A Cross-Cultural Examination of the Effects of Apology and Perspective Taking on Forgiveness,” Journal of Language and Social Psychology, 20 (March–June), 144–66. Thayer, Robert E. (1989), The Biopsychology of Mood and Arousal, Oxford: Oxford University Press. Thayer, Stephen (1986), “History and Strategies of Research on Social Touch,” Journal of Nonverbal Behavior, 10 (1), 12–28. Tybur, Joshua M., Debra Lieberman, and Vladas Griskevicius (2009), “Microbes, Mating, and Morality: Individual Differences in Three Functional Domains of Disgust,” Journal of Personality and Social Psychology, 97 (1), 103–22. Underhill, Paco (1999), Why We Buy: The Science of Shopping, New York: Simon & Schuste

 

 

Planned or Impulse Purchases? How to Create Effective Infomercials

TOM AGEE

University of Auckland,

New Zealand t.agee@paradise.net.n2

BRETT A. S. MARTIN

University of Auckland,

New Zealand bas.martin@auckland.ac.nz

 

Conventional wisdom suggests that most purchases made from infomercials— 30-minute direct-response television advertisements—are made on impulse. However, this study of 878 infomerciai purchasers of six products from a major internationai infomerciai marketer indicates that the majority of purchase decisions invoived some degree of planning rather than simpiy being made on the spur of the moment. Factors infiuencing whether a purchase was an impulse or planned decision included: comments by experts, demonstrations, the levels of previous product interest, prepurchase thinking about the product, and prior exposure to the advertisement, as well as the number of infomercials viewed by consumers. Having children aged between 10 and 14 years old also had an influence.

 

INFOMERCIALS REPRESENT A FORM of advertising of considerable commercial significance. In 1996, advertisers were reported to have spent $800 million on infomerciai time (Lockard, 1997), with infomercials more recently generating global sales of $75 billion {Direct Marketing, 1999). Furthermore, a number of large companies are using or planning to use infomercials, including Cadillac, Philips, Ericsson, and Volvo (Guilford, 1999; Krol, 1998; Halliday, 1999; Wasserman, 1999).

 

Yet despite the commercial importance of this form of advertising, there is little past research in this area. This is surprising given that iitfomercials, as a form of direct-response advertising, have been highlighted as different from other types of television advertising. For instance, Andrews (1999) suggests that direct-response advertisen\ents provide consumers with enough information to make a purchase decision, as well as giving them a way to purchase the product immediately. This study addresses this gap in previous research by examining two research questions:

1. Are infomerciai purchases planned or impulse decisions?

2. What factors influence whether an infomerciai purchase is planned or impulse?

 

CONCEPTUAL BACKGROUND

Infomercials have been defined as a longer than average advertisement that ranges in duration from 3 to 60 minutes (Belch and Belch, 1993). They form a subset of the broader category of directresponse television advertising that can be split into three categories: infomercials, short-form commercials—usually two minutes or less—and home-shopping-channels devoted entirely to selling products via television on a 24-hour-a-day basis.

 

An intriguing feature of infomercials is that they may appear to the viewer initially as a program rather than a commercial. The long duration of infomercials also increases the chances of catching channel surfers (Duket, 1997). Indeed, the fact that viewers happen upon the infomerciai rather than actively seek it out is reflected in the typical construction of an infomerciai. They usually consist of segments containing demonstrations, with testimonials by experts and satisfied users separated by two internal commercials. In addition, they contain the offer, price, payment options, telephone number, and a repeated call for action. Each segment can stand on its own, thus viewers are able to tune in during any segment and receive a complete sales presentation.

 

Although there may be no other form of direct-response advertising that produces such measurable results so quickly, there ‘ “*’L t J. is little published research in the area. Respondent Demographjc Profile Elliott and Speck (1995) were the first to , , ,-• 11 • i: • 1 T-i • Demographics Frequency Percent* look specifically at mfomercials. Their ° ••• •. study was of viewer characteristics and Gender how these might relate to attitudes and ….1^.^.’.?. ^®.? ?:.?-.^ purchase intentions.

 

A mail survey by El- Female 716 81.5 liott, Speck, and Alpert (1995) indicated ^ g that viewers generally had negative be- <20 38 4 4 liefs about infomercials, which signifi- 20-29 257 29 6 cantly affected attitude and purchase in- . .• uu u ..1.J-.- 1 . 30-39 299 34.5 tentions, although additional exposure to infomercials did have a positive effect. ….^9!7!^^. ‘^^^ “‘”^’^ Donthu and Gilliland (1996) studied the 50-59 79 9.1 psychographics of infomerciai shoppers, 60-64 18 2 1 describing them as heavier television ^ 65+ 26 3.0 viewers who are more convenience seeking, variety seeking, innovative, and risk Household Income accepting.

 

Thus infomercials appear ‘. „…”.„ to have the potential to evoke impulse ….!.2.9.’.99?.7.^29’OO9 107 13.1 buying. $30,000-$39,000 134 16.4 Impulse buying ….?.’!’:9.’.99?.T.!5.?’.9.°.9 ^.^9 14.7 Impulse buying has been defined as a $50,000-$59,000 108 13.2 spontaneous, immediate purchase (Rook $60,000-$79,000 98 12.0 and Fisher, 1995). The consumer is not ac- ••••$8o;oo(^$99!oOO 63 ‘j.l tively looking for the product and has no prior plans to purchase (Beatty and Fer- ….?.1.9.°.’.99.9:^ ^.? l.^. rell, 1998; Weun, Jones, and Beatty, 1998). Education Rook and Hoch (1985) assert that people ….::.’Ji^^..^.’i=.’;’.°.°|.S”.^.^.^?^.^. 363 42.0 experience consuming impulses. Further- High school graduate 145 16.8 more. Rook (1987) identifies this buying Technical or trade qualification 97 11.2 impulse with descriptors such as a sponOther tertiary qualification 153 17.7 taneous, intense, exciting, urge to buy with the purchaser often ignoring the con- ….^°!^.^..^.9}}3^. ?^. .2.-9 sequences. While research in this area dis- College graduate 81 9.4 cusses impulse buying as a trait, rather Marital Status than as the classification of a purchase de- Single 168 19.4 cision, researchers agree that consumers “Married/Living together 589 Q8.1 vary in their impulse-buying tendency (Pud, 1996; Rook and Fisher, 1995). Re- …’^^’i”/’^^ . ^3 9.6 garding the impulse infomerciai decision, ….y:.’…?.^.^… _ .^..^ Stern (1962) has offered the suggestion ef- (continued) feet of unplanned purchase. Here, unaware of a new product, the consumer is TABLE 1 the TV purchase; previous interest in the Cont’d *^P^ °^ product purchased; previous ex- “~ ~~~~~’ posure to TV advertisements for the prodDemographics Frequency Percent* uct purchased as well as similar products; Ettinicity ^^’^ amount of thought given to the TV White 676 78,0 purchase, “”Maori descent 132 15,2 ^°’ example, whether the purchase was an impulse or planned decision was meaPacific isiand descent 19 2,2 j u ..u -^ /-^ u J U sured by the item To what degree would ,,,,?,’?,’,’?,?^^^,9,!,’^,^,^f^,^,’,^,?, ?:?, 3.:h you say your decision to purchase was Other 22 2,5 planned in advance?” (1 = Not at all—just Totai sampie size 878 100,0 ^ ^P”‘^ °^ * e moment impulse, 5 = Very much—had planned to buy the next time ‘Percentages based on totats of each characteristic. I saw the advertisement).

 

Likewise, predisposition toward purchasing a product exposed to stimuli that suggests a need is conducted in three waves over a three- similar to that advertised included quessatisfied through purchase, month period. The products were adver- tions such as: “I had seen TV ads for other tised during approximately 25 hours of products like this before,” and “I had been METHOD TV time purchased per week. We devel- looking around for a product like this, oped a database of 878 purchasers re- even before I saw the TV ad,” PreSample sponding to our mail survey, which rep- purchase thinking included questions The data used in this study was collected resents a response rate of 32,8 percent, A such as: “I thought a lot about the infowith the cooperation of the New Zealand profile of the sample is presented in Table mercial before 1 decided to buy,” and “Bedivision of a major international infomer- 1, This was a tradeoff between more ques- fore I decided I thought a lot about whethcial marketer.

 

New Zealand represents a tions and therefore more information with er I might benefit from the product,” microcosm of 3,8 million consumers who a slightly lower response rate; 30 percent Perceptions of advertising effectiveness are often used as a test market for launch- has been noted as a reasonable response related to seven items such as “I found the ing new products by global marketers rate to expect (Cooper and Fmory, 1995), infomercial interesting and informative,” (e,g,. Reader’s Digest). and “The demonstrations of how the Infomercials are programmed in off- product worked were very helpful in peak times, usually mid-morning and Survey instrument making my decision to buy,” Respondents after midnight.

 

Virtually all of these are The survey consisted of questions cover- were asked to indicate degree of agreeproduced for the American market with ing the type of purchase decision, as well ment with statements related to the above the only modification for the New as factors that influence this decision, measured by a 5-point scale with anchors Zealand market being price and ordering areas that we believe break new ground in 1 = agree strongly and 5 = disagree information, infomercial research. The survey package strongly.

 

We surveyed 2,670 people who had consisted of a cover letter, a prize draw purchased a product from an infomercial incentive, the questionnaire, and postagewithin the previous two weeks. Six prod- paid reply envelope, A follow-up mailing ucts were surveyed: three exercise devices with an additional survey was sent including a strider for aerobic fitness, a approximately two weeks later with rider providing resistance for major an additional prize draw opportunity to muscle groups, and an exerciser to reduce improve response.

 

The sections of the the buttock area; a facial cream; a chil- survey relevant to degree of planning dren’s reading program; and a memory- consisted of questions relating to the folimprovement course. The research was lowing: the degree of planning given to Anaiysis Two different analyses were conducted on the data. First, factor analysis was performed to assess the underlying dimensionality of scale items; this allowed indices to be created for constructs measured by multiple items.

 

Second, a multiple regression was used to determine what factors influenced whether an infomercial purchase was planned or made on impulse. The criterion variable here was the extent to which a decision was a planned or impulse decision. The results of these analyses and the independent variables used in the regression analysis will now be presented, RESULTS Factor anaiysis The principal components analysis generated six indices: advertising effectiveness (Cronbach’s alpha = ,79); comments and demonstrations (a = ,74); payment information (a = ,64); recognition, comparison, and extras (a = ,56); previous interest in the product index (a = ,63); and a prepurchase thinking index (a = ,83), These are displayed in Table 2, Overall, these indices demonstrate adequate reliability with only the “recognition, comparison, and extras” factor falling below a coefficient of ,60 (cf, Malhotra, 1993), yet it is still above the ,50 guideline of Guilford (1954),

Research Question 1: Infomercial purchases—planned or impulse? In response to being asked to what degree respondents felt their infomercial purchase was planned in advance, where 1 = impulse decision and 5 = very planned decision, the results indicated that purchases were somewhat planned (Mean = 3,66, standard deviation = 1,43), Only 13 percent of respondents considered their purchase a spur of the moment decision. On the other hand, 65 percent believed that some planning had gone into the purchase decision, with 38,9 percent of the total number of respondents rating their decision as very much planned.

Research Question 2: What factors influence whether a purchase is planned or impulse? The results for the multiple regression are displayed in Table 3, The regression model was significant (p < ,001) with 27 percent of the data explained by the model (Adjusted R-square = ,27), Two variables had a statistically significant positive association with the extent to which a decision was planned or impulse (1 = impulse decision, 5 = very planned decision).

 

The first was prior exposure to the advertisement (r = ,26, p < ,001), As prior exposure was a dummy variable for the question “Had you seen the infomercial before for the product you just bought?” (0 = no, 1 = yes), this result indicates that a “yes” response (i,e,, prior exposure) was more indicative of a planned decision rather than an impulse one. Presumably, impulse buyers are more likely to purchase upon first exposure, when they react to the “impulse” to purchase (Rook, 1987), Planned infomercial purchasers are more likely to have viewed the infomercial on previous occasions before committing themselves to the actual purchase of the advertised product.

 

To further investigate this result we reduced the dataset to those buyers who had answered “yes” for prior exposure (i,e,, they had seen the infomercial before). We then ran a correlation between their planned/impulse dependent variable scores (1 = impulse decision, 5 = very planned decision) and the number of times these consumers had seen the advertisement (1 = seen once, 4 = 4 times or more). Not surprisingly, this revealed a significant positive association (r = ,18, p < ,001) which suggests that planned decisions are associated with having seen the infomercial advertising that product numerous times.

 

In other words, the more often the infomercial is seen, the more planned the final decision purchase is likely to be. The second variable positively correlated with whether a decision was planned or impulse was children aged 10 to 14 years (r = ,10, p < ,05), This indicates that if infomercial buyers have children in their household aged between 10 and 14 years old the purchase decision is likely to be more planned. Buyers with children of this age are less likely to make an impulse purchase in response to seeing an infomercial.

 

Variables with significant negative associations with planned/impulse decisions were: Comments and demonstrations (p < ,01); previous interest in the product (p < ,05); prepurchase thinking (p < ,001); and amount of infomercial viewing (p < ,05), For comments and demonstrations this result shows that the more important that infomercial buyers rate customer testimonials, expert comments, and demonstrations shown in the advertising (1 = very important, 5 = not important at all), the more likely they are to make a planned purchase rather than an impulse one. For previous interest in the product, planned decisions were associated with higher levels of previous interest in the product advertised.

 

Buyers here are more likely to have been looking around for a product of this type before they saw the infomercial and thus are more planned and considered in their infomercial purchase. For prepurchase thinking, the more buyers think about the content of the informercial they have seen, the more likely they are to make a planned decision. This indicates that consumers who are stimulated to think by the infomercial are less likely to make impulse purchases. For the amount of infomercial viewing (0 = more than once a week, 1 = once a week or less), this negative correlation suggests that the fewer infomercials a consumer watches, the more likely that person is to make an impulse purchase.

Conversely, people who watch infomercials more than once a week, are more planned in their purchase decisions. . . . infomercial purchases are not always the impulse decisions that we might expect. Overall, purchases are rather planned.

DiSCUSSiON This study has found that infomercial purchases are not always the impulse deci- ^^^^^ ^f^^^ ^^^^^ ^^^ infomercial several sions that we might expect. Overall, pur- ^-^^^ Importantly, we uncovered some chases are rather planned. In this study, yggf^j findings for what factors influence pure impulse purchases represented the whether an infomercial purchase is a minority (or 13 percent) of total purchases, planned or an impulse decision.

 

Most were planned, at least to some ex- j ^ ^^^^^ ^j the factors that influence whether an infomercial purchase decision is planned or impulsive, the results suggest that consumers are more likely to “O^t ‘ ‘ make impulse purchases whether they are Multiple Regression Results infrequent viewers of infomercials. This suggests that it is the infrequent viewer of informercials who is most susceptible to making an impulse purchase. In contrast, Variabie Beta , . j ^ u more frequent viewers tend to be more Infomercial Elements likely to make planned decisions that ,,,,,9°^,’???i:’H,?^??,,R?,’?]?!?,?*!;?*!°’:i?,,,,:;,^?,’, are often associated with seeing the infoConsumer Ctiaracteristics mercial a number of times. Previous interest in product -,09” Interestingly, impulse purchasers often Prepurchase thinking -,24″^ have lower levels of previous interest in Prior exposure to ad’ “26^ * ^ P”°^”‘^* ^^ ^ ”” ^ *=^” P^^””^”^ P””- chasers.

 

This suggests that for impulse Amount of Infomercial viewing -,10^ , . , ,° buyers, infomercials succeed in making Demographics consumers aware of a product, convincing ,,,,,9!:’,’,’,^’?,”,,,l,!^^iy?,^’!, ;,^?.! them of their need for this product, and Diagnostics securing a purchase. Likewise, impulse R-Square 31 buyers, having seen the infomercial, are , ., ^ , ,, „-, less likely to engage in prepurchase Adjusted R-square ,27 / o o r r thinking whereby they contemplate the Standard error of estimate 1,21 -^ r ..u ,. ,. J • .u merits of the arguments presented in the ,’?!^,*i’!]:yy^H?r’,,^!^,?^i?!;’,*;^, h^ l infomercial.

 

Overall then, impulse buyers. Note: Variables include advertising effectiveness, payment relative to planned buyers, have seen the information, recognition, comparison and extras, prior pur- infomercial fewer times before purchase, chasin<^ usini^ infomercials, prior purchasing using regular TV ads, prior purchasing using mail-order catalogs^pay- ^^””^ ^^^^ previous interest in the product, ment method, product type, and the demographics—age, and think leSS about the merits of the adeducation, ethnicity, gender, income, marital status, chil- ^ertising before purchase. These findings dren in the household under 2 years, 2-5 years, 5-9 years, ^J i u and 15-20 years. The table includes only those variabtes would appear to implicitly Support the litxoith significant standardized beta coefficients. erature on impulse buying.

 

For instance, “Beta significant at p< .05 „ , .. , , ‘•Beta significant at p < ,01 ^eatty and Ferrell (1998) assert that im- ‘Beta significant at p< .001 pulse purchases tend to be spontaneous 4 0 JGDRRHL DF RGDERTISIRG RESERRCR November . December 200 1 and are acted upon without a lot of reflection or prepurchase intentions. Since planned buyers have a greater interest in the product before even seeing the infomercial, these consumers may have higher levels of intrinsic involvement with the product (Celsi and Olson, 1988), Owing to this higher natural interest, planned buyers may have more extensive evaluative criteria and information needs than impulse buyers.

 

The decision requires greater thought and evaluation than it does for an impulse buyer. Consequently, planned purchasers find demonstrations of product performance and expert comments more important, presumably as part of their more analytical assessment of the message’s argument quality (Laczniak and Muehling, 1993), This represents a useful avenue for future research. Furthermore, given that planned buyers have seen the infomercial more often it would be of interest to study how best to repeat infomercials to generate sales.

 

For example, do more purchases result from planned buyers viewing the same infomercial repetitively on one occasion (e,g, seeing the advertisement three times in a row) or from viewing that infomercial once each day over subsequent days? Consumers with children aged between 10 and 14 years of age also tended to make more planned decisions. One could argue that with a limited budget and less disposable income, or at least compromises that had to be reached in household expenditure, these consumers were more considered in their approach to purchasing.

 

Whether this was because of a sense of fiscal responsibility, or whether the views of family members were sought in approving the purchase (Beatty and Talpade, 1994; ChUders and Rao, 1992) is an interesting avenue for further research. Factors that had no influence on the extent of planning included: the type of product advertised, the payment information displayed, the effectiveness of the advertising, whether the consumer had purchased from an infomercial, TV advertisement, or mail-order catalog before, and the method by which consumers paid for their products. Likewise, demographic variables did not influence consumers, with the exception of the aforementioned age of children in the household. Of some surprise is that infomercial elements that appear to be employed often, such as endorsement by recognizable spokespeople (Ohanian, 1991), product comparisons (Neiman, 1987), and offering additional items in the infomercial, while involving the viewer or enhancing the selling proposition, did not seem to influence the nature of the decision,

 

CONCLUSION

This article suggests that it might be dangerous for marketers to overrate the persuasive power of the infomercial as a device that prompts mainly an impulse purchase. Our study shows that many purchases may be categorized as somewhat planned, especially if the consumer has viewed the advertising several times. Importantly, the present study shows that impulse purchasers are characterized by viewing infomercials less frequently than planned purchasers, having seen the infomercial for the product less often, and also thinking less about the reasons for purchase provided in the infomercial.

 

TOM AGEE is a senior iecturer of marketing at the University of Auckland, New Zealand, He established the country’s first degree in advertising in 1994 while Head of the School of Marketing and Advertising at the Auckland University of Technology. He has more than 25 years experience in virtually every facet of advertising, having worked with agencies both in New Zealand and the United States, where he was a founding partner of Finnegan & Agee, Richmond, Virginia, He actively consults on marketing communication issues with national advertisers and government departments,

BBETT A. S, MARTIN is a senior lecturer of marketing at the University of Auckland, New Zealand, He received his Ph,D, in marketing from the University of Otago, His teaching interests include consumer behavior, marketing strategy, and e-commerce. His research has been published in journals and books such as the European Journal of Marketing, Marketing Intelligence and Planning, and the Australasian Marketing Journal.

 

REFERENCES

ANDREWS, K, J, “Selling Short,” Target Marketing 22, 3 (1999): 56-57, BEATTY, S, E,, and M, E, FERRELL, “Impulse Buying: Modeling Its Precursors,” journal of Retailing 74, 2 (1998): 169-91, , and S, TALPADE. “Adolescent Influence in Family Decision Making: A Replication with Extension.” Journal of Consumer Research 21, 2 (1994): 332-41, BELCH, G, E,, and M, A. BELCH, Advertising and Promotion: An Integrated Marketing Communications Perspective. Burr Ridge, IL: Irwin, 1993, CELSI, R, L,, and J, C. OLSON, “The Role of Involvement in Attention and Comprehension Process,” Journal of Consumer Research 15, 2 (1988): 210-24, CHILDERS, T. L., and A. R. RAO. “The Influence of Familial and Peer-Based Reference Groups on Consumer Decision,” Journal of Consumer Research 19, 2 (1992): 198-211, COOPER, D, R,, and C, W, EMORY, Business Research Methods. New York, NY: Irwin, 1995, DIRECT MARKETING. “And Now a Word from Our Sponsor,” 62, 2 (1999): 48-50. DONTHU, N, , and D, GILLILAND, “The Infomercial Shopper,” Journal of Advertising Research 36, 2 (1996): 69-76, DuKET, J. V, “Infomercials and the Decline of Civilization,” Marketing News, 1997, ELLIOTT, M , T., and P, S, SPECK, “Antecedents and Consequences of Infomercial Viewership,” Journal of Direct Marketing 9, 2 (1995): 39-51, , M, T,, P, S, SPECK, and F, H, ALPERT, “The Effect of Prior Exposure Level on Consumer Perceptions of Infomercials,” Proceedings of the 1995 Conference of The American Academy of Advertising. Norfolk, Virginia, March 25-27, 1995, GUILFORD, D, “Cadillac Takes New Route for Seville STS: Infomercial,” Advertising Age, August 23, 1999, GUILFORD, ], P, Psychometric Methods, 2nd ed. New York, NY: McGraw-Hill, 1954, HALLIDAY, ], “Volvo Ready to Act on Leads after Infomercial Success,” Advertising Age, January 25, 1999, KROL, C, “Philips Uses Infomercial to Put Spotlight on New DVD Player.” Advertising Age, October 5, 1998. LACZNIAK, R, N. , and D, D, MUEHLING, “Toward a Better Understanding of the Role of Advertising Message Involvement in Ad Pro cessing.” Psychology and Marketing 10, 4 (1993): 301-19, LoCKARD, M. “Selling on the Television.” Target Marketing 20, 12 (1997): 54-55. MALHOTRA, N . K. Marketing Research: An Applied Orientation. London: Prentice-Hall, 1993. NEIMAN, J. “The Trouble with Comparative Ads.” Adweek, January 12, 1987. OHANIAN, R. “The Impact of Celebrity Spokespersons’ Image on Consumers’ Intention to Purchase.” journal of Advertising Research 31, 1 (1991): 46-54. PuRi, R. “Measuring and Modifying Consumer Impulsiveness: A Cost-Benefit Accessibility Framework.” journal of Consumer Psychology 5 (1996): 87-113. ROOK, D . W. “The Buying Impulse.” journal of Consumer Research 14, 2 (1987): 189-99. , and S. HOCH. “Consuming Impulses.” Advances in Consumer Research 12, 3 (1985): 23-27. -, and R. J. FISHER. “Normative Influence on Impulsive Buying Behavior.” journal of Consumer Research 22, 3 (1995): 305-13. STERN, H . “The Significance of Impulse Buying Today.” journal of Marketing 26, 2 (1962): 59-62. WASSERMAN, T. “Ericsson Releases Genie with Humor, $12.5M.” Brandweek, October 11, 1999. WEUN, S., M . A. JONES, and S. E. BEATTY. “The Development and Validation of the Impulse Buying Tendency Scale.” Psychological Reports 82, 3 Pt. 2 (1998): 1123-33.

The Role of Gender Identity and Self-Construal Salience In Evaluations of Male Models

Is the Marlboro man the only alternative?

The role of gender identity and self-construal salience in evaluations of male models

Brett A. S. Martin & Juergen Gnoth

Published online: 30 January 2009

Springer Science + Business Media, LLC 2009

Abstract This research examines how men react to male models in print advertisements. In two experiments, we show that the gender identity of men influences their responses to advertisements featuring a masculine, feminine, or androgynous male model. In addition, we explore the extent to which men feel they will be classified by others as similar to the model as a mechanism for these effects. Specifically, masculine men respond most favorably to masculine models and are negative toward feminine models. In contrast, feminine men prefer feminine models when their private self is salient. Yet in a collective context, they prefer masculine models.

These experiments shed light on how gender identity and self-construal influence male evaluations and illustrate the social pressure on men to endorse traditional masculine portrayals. We also present implications for advertising practice.

Keywords Advertising . Classification expectations . Gender identity . Self-construal . Evaluations

 

1 Introduction

An important decision for advertisers is the selection of an appropriate model to use in an advertisement. In reference to our title, the Marlboro man is an example of the traditional masculine male model in advertising. Content analyses suggest that advertisers tend to use traditional male stereotypes (Ganahl et al. 2003; Vigorito and Curry 1998). Yet whether the traditional male is the only depiction that would resonate with consumers today appears open to debate. In the social sciences, it is accepted that Market Lett (2009) 20:353–367 DOI 10.1007/s11002-009-9069-2

B. A. S. Martin (*)

Marketing Group School of Management, University of Bath, Bath BA2 7AY, UK e-mail: B.A.S.Martin@bath.ac.uk J. Gnoth Department of Marketing, University of Otago, P.O. Box 56, Dunedin, New Zealand e-mail: JGnoth@business.otago.ac.nz

 

a variety of masculinities now exist, such as jocks to sensitive new men (Smiler 2004). Likewise, marketing research is revealing consumer masculinities that differ from the traditional norm (Holt and Thompson 2004; Patterson and Elliott 2002). These emerging masculinities have been recognized in the popular press with the term “metrosexual” to represent a $1.3 billion market of heterosexual men who use traditionally feminine products, such as facial moisturizers (Prior 2004).

 

Given these changes, it appears useful to study whether alternative profiles to the masculine model (e.g., feminine male models) should be considered. The purpose of this article is to examine how men respond to print advertising featuring masculine, feminine, and androgynous male models. We show that gender identity affects evaluations, but that these effects must take into account consumer self-construal. We use the term “biological sex” to refer to the physical differences between males and females. In contrast, “gender identity” represents the psychological features often associated with these physical differences, which are socially constructed phenomena (Deaux 1985).

 

We contribute by showing how gender identity provides insights into how men react to advertising (studies 1 and 2). Furthermore, we explore the cognitive process that underlies male attitudes (study 2). We propose that expectations of being classified by other people as similar to an ad model (classification expectations) mediate the effect of model gender identity on male attitudes. Previous psychological research has focused on misclassification expectations where masculine men are concerned at being misclassified by strangers as feminine (Bosson et al. 2005).

 

We contribute by showing how the responses of feminine men to advertising are influenced by concerns of being correctly classified by others as feminine. From a managerial perspective, we also present important implications on how to advertise to men. 2 Background and hypotheses 2.1 Gender identity Prior to the 1970s, gender research assumed a unidimensional perspective where masculinity and femininity were opposite ends of a single continuum (Parsons and Bales 1955). This view embraced the notion that biological sex was the key determinant of sex-related behavior (Stern 1988). In a departure that was to have a profound impact on subsequent research, Bem (1974) asserted that the masculinity and femininity of an individual were independent dimensions influenced by socialization.

 

Here, masculinity and femininity represent socially desirable instrumental traits (e.g., independence) and expressive traits (e.g., sensitivity to others; Bem 1974). An individual can be high or low on each dimension. Gender identity is, therefore, independent of one’s sex and the potential exists for cross-sex-typed individuals (e.g., a feminine male), rather than just masculine males and feminine females. To this end, Bem offered the Bem Sex Role Inventory (BSRI; Bem 1974) which measures an individual’s masculinity and femininity and allows for their classification by median splitting into one of four groups: (1) masculine (i.e., high masculinity, low femininity), (2) feminine (low masculinity, high femininity), (3) androgynous (highmasculinity, high femininity), or (4) undifferentiated (low masculinity, low femininity).

 

The BSRI has received widespread use in both the social sciences (Beere 1990) and marketing (Stern 1988) and is used in the present study. In marketing, gender identity research has offered mixed results. Debevec and Iyer (1986) matched spokesperson sex with a product that was perceived as traditionally male, female, or neutral in terms of image. They posited that conformity to societal expectations would override gender identity differences. They found no significant main effect for gender identity on evaluations or usage intentions for products such as dishwashing liquid and beer. Nor did they find any significant interactions between gender identity, spokesperson sex, and product gender (masculine, feminine, or neutral). However, it is unspecified how data were classified for the gender identity variable.

 

Likewise, Stern (1988) suggests biological sex is at least as good as, if not better, than gender identity as an explanatory variable. Yet research by Jaffe (1994) and Jaffe and Berger 1988) suggests that gender identity can favorably influence attitudes where the gender identity of the model matches the consumer’s gender identity. Jaffe (1994), in a field experiment of 200 women, found women with higher masculinity evaluated ads that portrayed women in progressive careers vs traditional nurturing portrayals more favorably. She suggests that gender identity can be a useful predictor for responses to advertisements.

 

Much of this research treats gender identity as an independent variable. Yet logically, an individual’s responses may also be influenced by their social context. A more public social context could induce impression management (Aarts and Dijksterhuis 2003). It is our contention that men are influenced by social pressure from other men to conform to traditional masculine expectations when evaluating male models in print advertisements.

 

2.2 Self-construal salience and male pressure to conform

Self-construal relates to a person’s self-concept or self-view (Guimond et al. 2006). It is recognized in marketing that different self-construals can be salient in different situations, such as parent vs professor (Aaker 1999). In other words, self-construal represents more of a situational state variable as opposed to self-concept which is more of a trait.1 Similarly, Triandis (1989) identified three types of self: (1) the private self, which involves self-assessment (e.g., “I am extroverted”); (2) the collective self, which relates to in-group norms (e.g., “my friends think I listen to cool music”); and (3) the public self, which is the view held by general others (e.g., “people think professors are smart”).

 

For this research, we address the private self and the collective self. Research shows that private and collective self-construals can be primed and that priming the collective self results in people caring about what important others might think (Ybarra and Trafimow 1998). We contend that men experience social pressure to endorse traditional masculine stereotypes when judging male models. Here, the collective self involves other men and the impression they may form of the male judging the ad. We base our view on theoretical and empirical research.

 

It is widely theorized in masculinity research that men face substantial normative pressure to endorse traditional masculinity and exhibit an aversion to appearing feminine (Connell 1995; Smiler 2004). Traditional masculinity is an ideology where an individual internalizes cultural attitudes toward masculinity which informs their expectations of masculine behavior (Levant and Richmond 2007). Connell (1995) referred to this ideology as hegemonic masculinity which is a culturally dominant traditional masculinity that subordinates other male masculinities, such as homosexuality (Connell 1995).

 

In other words, there is a hierarchy of masculinities and traditional masculinity dominates them. From this perspective, traditional masculine stereotypes represent the normative standard for all men. In addition to expectations to act as a traditional masculine male, empirical research reveals how responses from other men encourage endorsement of traditional masculinity. For instance, masculine men regard feminine males as gay (Wade and Brittan-Powell 2001). This is relevant as masculine men are more likely to engage in antigay harassment than nontraditional men (Wade and Brittan-Powell 2001).

 

Furthermore, nontraditional men have been shown to be concerned of a social backlash from other men for violating traditional masculine stereotypes (Rudman and Fairchild 2004). Overall, this suggests that men are subject to normative pressure to endorse traditional masculinity which, in our research, means males being seen to endorse masculine models. The limited research on how men respond to male models in advertising supports this view. Elliott and Elliott (2005, p. 10) found men in a focus group setting uniformly condemned male models seen as feminine or “not manly enough.” Although they did not study participant gender identity, we posit that such a result reflects the influence of the collective self.

 

In a public setting, males endorse the traditional masculine norm. For masculine men who believe in traditional masculinity (Moore and Stuart 2004), such public endorsement reflects their personal views. However, for feminine men and androgynous men, there is a potential conflict between social pressures to be seen to endorse masculinity and their own feminine characteristics. We posit that how nontraditional men resolve this tension can be explained by considering self-construal salience. Next, we present our hypotheses.

 

2.2.1 Masculine men

Gender schema theory (Bem 1981) posits that masculine men have integrated cultural expectations of masculinity into their self-concept and that they process information on the basis of these expectations. Since masculine men (1) tend to support the dominance of traditional masculinity, (2) are more likely than other gender identities to be intolerant of those who deviate from gender norms, and (3) strongly avoid the negative associations of femininity in men (Garst and Bodenhausen 1997), they should prefer ads featuring the masculine model. Since this preference reflects their personal views, it should be exhibited in private self and collective self contexts.

 

H1: For masculine men, advertisements featuring a masculine model will generate more favorable evaluations than for advertisements featuring other model types, irrespective of their level of self-construal.

2.2.2 Feminine men

Feminine men, due to their lack of endorsement of masculine characteristics (Bem 1981), are anticipated to reject the appearance of masculinity, but only in a private self context where they are not subject to normative pressure. In private, feminine men will have more favorable attitudes to ads featuring the feminine model. Yet in a collective self context, feminine men should conform to the masculine norm and prefer the masculine model.

H2: For feminine men, advertisements featuring a feminine model will generate more favorable evaluations than for advertisements featuring other model types, when the private self is primed. However, advertisements featuring a masculine model will be preferred when the collective self is primed.

 

2.2.3 Androgynous men

Given that (1) androgynous men look to enact appropriate behavior in a given social context (Bem 1974; Ickes et al. 1979) and (2) androgynous people are more conformist than other gender identities (Anderson 1986), we posit that they should endorse the masculine model when the collective self is salient, but should endorse the androgynous model in a private self context. To this end, androgynous men have been found to mimic the behavior of masculine men when interacting with them, but to exhibit more expressive behavior when interacting with androgynous men (Ickes et al. 1979).

 

H3: For androgynous men, advertisements featuring an androgynous model will generate more favorable evaluations than for advertisements featuring other model types, when the private self is primed. However, advertisements featuring a masculine model will be preferred when the collective self is primed.

 

3 Study 1

3.1 Pretests

Forty students rated the masculinity and femininity of nine male models (three androgynous, three masculine, three feminine) on seven-point scales (not so masculine– highly masculine, not so feminine–highly feminine). As a result, a masculine model (Mmasculine=6.13, Mfeminine=2.60, p<0.001), feminine model (Mmasculine=1.63, Mfeminine=5.80, p<0.001), and androgynous model (Mmasculine=5.70, Mfeminine=5.68, NS) were selected. Based on a separate pretest (n=31) showing that mobile phones were familiar and gender neutral, mobiles were chosen for the main study.

 

3.2 Overview and data collection

The experimental design involved a within-subject variable (model gender identity: androgynous, masculine, or feminine) and a between-subjects variable (self-construal salience: private self, collective self). Participant gender identity (androgynous,masculine, and feminine) was measured. Thus, a 3×2×3 mixed design was used. Participant gender identity scores were collected 1 month before the main study (Aaker 1999). For the main study, 208 male undergraduates were given randomly distributed booklets. They read three print ads at their own pace and completed a questionnaire. The entire procedure took approximately 20 min.

 

3.3 Independent variables

Each print ad contained a male model. A pretested fictitious brand name was used to avoid the influence of brand inferences. For self-construal, we used the priming procedure of Reed (2004) where participants completed a hand writing study that asked them to write three sentences on any topic as a baseline measure and then to write five independent sentences on a topic which related to the prime. For the private self prime (collective self prime), participants read that these sentences should “describe your sense of independence as an individual young adult” (“your sense of connectedness with people you feel close to”). After completing the priming task, participants were asked to complete an unrelated study that looked at print ads which contained the experimental stimuli and dependent measures.

 

The ads were presented in random order (i.e., androgynous model first, masculine model first, or feminine model first). Analysis of variances (ANOVAs) showed no order effects on participant evaluations and cognitive responses (study 1: ps>0.14, study 1: ps>0.12). Thus, this is not discussed further. For consumer gender identity, participants completed the BSRI (Bem 1974) which involved rating 20 masculine, 20 feminine, and 20 gender neutral adjectives (not at all desirable–extremely desirable, seven-point scale). Participants were classified as androgynous, masculine, feminine, or undifferentiated on the basis of median splitting of the masculinity (α=0.91) and femininity scales (α=0.81).2 Following prior research (Jaffe and Berger 1988), undifferentiated people, who do not use gender expectations to process information (Bem 1981), were removed from further analysis.

 

3.4 Dependent variables

Evaluations used four seven-point scales (positive–negative, very favorable–not at all favorable, good–bad, would definitely consider buying it–would definitely not consider buying it) from Gürhan-Canli and Maheswaran (2000). Following Ahluwalia (2002), two judges coded the cognitive response data as impression management-related and nonimpression management-related (I, N) and as positive, negative, or neutral in valence (+, −, 0). Impression management-related thoughts were thoughts of the consequences of endorsing or using the brand publicly or influencing the impression that others formed of them (Ahluwalia 2002). Although our hypotheses relate to evaluations, this coding was performed to gain a fuller understanding of the effects of the independent variables. Examples include: “This 2 The distribution of participants by consumer gender identity for study 1 was as follows: 64 masculine males, 58 androgynous males, 45 feminine males, and 31 undifferentiated males.

For study 2, participants comprised 82 masculine males, 74 androgynous males, 60 feminine males, and 20 undifferentiated males.phone would make me hot” (I+), “Only (derogatory swear word) would use this” (I−), “Good colors” (N+), “There are better products on the market” (N−). These codings were used to create an index of valenced impression management thoughts (positive thoughts minus negative thoughts) and an index of valenced nonimpression management thoughts. Interjudge reliability was 90%

 

3.5 Results

3.5.1 Manipulation checks

A model gender identity manipulation check was performed using two seven-point scales (not so masculine–highly masculine, not so feminine–highly feminine). Analyses of these scales suggests that this manipulation was successful for masculinity (F2, 414= 323.29, p<0.001) and femininity (F2, 414=278.27, p<0.001). Specifically, appropriate profiles were presented for the masculine model (Mmasculine=5.49, Mfeminine=2.37, p< 0.001), androgynous model (Mmasculine=4.93, Mfeminine=4.78, p=0.31), and feminine model (Mmasculine=2.45, Mfeminine=5.35, p<0.001). For self-construal, we measured self thoughts on two seven-point items (While reading the ad, please describe the extent to which: you thought just about yourself, your thoughts were focused just on you) anchored by not at all–a lot, adapted from Aaker and Lee (2001). We averaged these items to form a self thoughts index (r= 0.91). We measured thoughts about others on two items (you thought about you and other people, your thoughts were focused on you and other people), adapted from Aaker and Lee (2001). We averaged these items to create an others thoughts index (r=0.83). ANOVA analysis revealed that exposure to the private self prime resulted in significantly more self thoughts (M=4.17) than the collective self prime (M=2.73, F1, 206=45.80, p<0.001). Furthermore, the collective self prime resulted in significantly more other-oriented thoughts (M=4.01) than the private self prime (M=2.73, F1, 206=51.01, p<0.001). These results suggest that the priming manipulation was successful.

 

3.5.2 Hypothesis testing

Masculine men A three-way multivariate analysis of variance (MANOVA) revealed a significant model gender identity×self-construal×consumer gender identity interaction for evaluations (F2, 138=7.46, p<0.001), valenced impression management thoughts (VIM, F2, 138=12.53, p<0.001) but not valenced nonimpression management thoughts (VNIM, p=0.07). To further investigate these results, a MANOVA was run for masculine men. A significant main effect for model type revealed that masculine models received the most favorable evaluations (M=4.54), followed by androgynous (M=3.96) and feminine models, respectively (M=3.10, F2, 56=34.80, p<0.001). As expected, self-construal priming did not influence the results (ps>0.14) nor did priming interact with model type (p>0.0.10). Thus, hypothesis 1 is supported. For cognitive responses, MANOVA revealed a significant difference for model type on VIM. Participants viewed the masculine model the most positively and the feminine model negatively (Mmasculine=0.41, Mandrogynous= 0.06, Mfeminine=−0.31, F2, 56=11.89, p<0.001). VNIM data yielded a similar pattern(Mmasculine=0.56, Mandrogynous=0.02, Mfeminine=−0.83, F2, 56=16.11, p<0.001). A review of the data suggested that the masculine model was viewed as an aspirational ideal, whereas the feminine model was frequently denigrated and presumed to be homosexual. Feminine men As expected, a significant model type×priming interaction was evident for evaluations (F2, 86=24.53, p<0.001). This interaction showed that feminine men prefer feminine models (M=5.02) over androgynous models (M= 3.86) and masculine models (M=3.73), when their private self is salient. However, when their collective self is salient, they claim a preference for the traditional masculine model (M=4.35) over the feminine model (M=2.81) and androgynous model (M=3.35; Table 1).

 

Thus, hypothesis 2 is supported. The cognitive responses revealed a significant model type×priming interaction for VIM (F2, 86=22.64, p<0.001) and VNIM (F2, 86=5.95, p<0.001). Under private self conditions, feminine men reported more positive VIM for feminine models (M= 0.55) than other model types (Ms<0.10). Yet when the collective self is activated, feminine men regard feminine models negatively (M=−0.78) as opposed to masculine models (M=0.72) and androgynous models (M=0.16). For VNIM, under the private self, feminine men dislike the masculine model (M=−0.19). For the collective self, feminine men report positive thoughts about the masculine model (M=0.72) as opposed to dislike of the feminine model (M=−0.41) and indifference toward the androgynous model (M=0.18). Androgynous men The model type×priming interaction was significant for evaluations (F2, 112=17.71, p<0.001).

 

As expected, when the private self is salient, androgynous men report more favorable evaluations after viewing an androgynous model (M=4.58) rather than a masculine model (M=3.36) or feminine model (M=3.09). Yet when the collective self is salient, they prefer masculine models (M=5.10) over androgynous models (M=3.88) and feminine models (M=3.82). Thus, there is support for hypothesis 3. The cognitive responses revealed a significant model type×priming interaction for VIM (F2, 112=19.05, p<0.001). The means generally converge with the attitudinal results. When the private self is salient, androgynous men report more favorable VIM in response to an androgynous model (M=1.00) rather than a masculine model (M=0.17) or feminine model (M= 0.07). When the collective self is salient, more favorable thoughts are reported for the masculine model (M=0.74) and the androgynous model (M=0.30) compared with the feminine model (M=−0.44). No such interaction was evident for VNIM (p>0.10).

 

3.6 Discussion

The findings suggest that men differ in their evaluations of male models when selfconstrual and gender identity are taken into account. Masculine men prefer masculine models and regard feminine models negatively. In contrast, feminine men conform to this pattern when their collective self is salient; but in private, their evaluations show the opposite pattern. The cognitive responses data generally.

 

Means with different letters are significantly different at p<0.05 Andro model androgynous model, VIM valenced impression management thoughts, VNIM valenced nonimpression management thoughts, Class. expect classification expectationsconverge with these findings and provide insight into the mechanism that accounts for the attitudinal effects.

 

Of note is the way that feminine and androgynous men endorse masculine models in a collective context but prefer models that resemble their own gender identity when their private self is salient. We posit that the reason for these results relates to male classification expectations. Specifically, how men feel they will be classified by others and the normative pressure men experience to endorse the traditional masculine male. Research suggests that heterosexual men seek to avoid misclassification as homosexual when they perform stereotypically feminine role behaviors, such as dancing in a ballet class (Bosson et al. 2005). Furthermore, Bosson et al. (2005) show that these misclassification expectations mediate feelings of self-conscious discomfort thereby promoting adherence to role norms.

Given that self-construal is context-dependent andthat, for the collective self, individuals emphasize their similarity to other in-group members (Guimond et al. 2006), such as male peers, we expect masculine men and androgynous men to endorse the masculine norm to avoid misclassification as feminine. However, while past research has examined misclassification (e.g., a masculine male regarded as feminine), we predict that in a collective self context, feminine men are concerned with being correctly classified (i.e., being revealed as feminine) because of social pressure and, hence, endorse the masculine norm.

 

This is because, for feminine men, there is a potential conflict between social pressures to be seen to endorse traditional masculinity—and hence the masculine model—and their own feminine characteristics. In other words, feminine males are likely to experience gender role dissonance because of the discrepancy between their private feminine views and the masculine stereotype which they are expected to support (Smiler 2004). To this end, recent research highlights the negative social reactions that nonconformists may suffer. Maas et al. (2003) show that masculine males are more likely to engage in harassment, particularly when exposed to feminine information which may threaten their self-identity.

 

Furthermore, Rudman and Fairchild (2004) show how people who violate traditional stereotypes can suffer a social backlash from other members of their in-group. They found that feminine men were more afraid of a backlash than traditional masculine men and that they actively engaged in strategies (e.g., feigning gender conformity) to avoid these negative reprisals. Overall, this suggests that feminine males hide their feminine views in order to avoid being classified by other men as feminine. Thus, their concern is that they will be correctly classified by other men as feminine. Thus, classification expectations should mediate male responses to male models in advertising

H4: Classification expectations mediate the effects of participant gender identity, model gender identity, and self-construal on evaluations.

 

4 Study

2 4.1 Overview, participants, and procedure

Study 2 replicates study 1 with a new product context and includes a measure for classification expectations. Two hundred and forty-three male undergraduates from the same subject pool as study 1 participated (BSRImasculinity: α=0.85, BSRIfemininity: α=0.77). 4.2 Independent variables and dependent variables The manipulation for self-construal and measures were identical to study 1. We selected watches as the product since a pretest showed they scored higher for men as a fashion item than mobiles. For classification expectations, participants rated the following seven-point item, “If this ad was used in the media and you bought the watch, how likely is it that a stranger would think you were like the model in the ad, if they saw you with the watch?” (not at all likely–very likely), adapted from Bosson et al. (2005).

 

4.3 Results

4.3.1 Manipulation checks

Analysis of the model gender identity manipulation check indicated that this manipulation was successful for masculinity (F2, 223=36.73, p<0.001) and femininity (F2, 223=58.54, p<0.001) with appropriate profiles for the masculine model (Mmasculine=4.93, Mfeminine=2.61, p=0.01), androgynous model (Mmasculine= 4.89, Mfeminine=4.71, NS), and feminine model (Mmasculine=3.22, Mfeminine=5.26, p< 0.001). For self-construal, the private self prime resulted in significantly more self thoughts (M=4.38) than the collective self prime (M=3.93, F1, 208=4.05, p<0.001). Furthermore, the collective self prime resulted in significantly more other-oriented thoughts (M=3.83) than the private self prime (M=2.65, F1, 208=26.15, p<0.001). These results suggest that the priming manipulation was successful.

 

4.3.2 Hypothesis testing

Masculine men A three-way MANOVA yielded a significant model gender identity×self-construal×consumer gender identity interaction for evaluations (F4, 196=6.44, p<0.001) and VIM (F4, 196=10.23, p<0.001). A MANOVA on masculine men data revealed similar significant main effects for model type on evaluations (F2, 76=9.02, p<0.001). Masculine models (M=4.47) are viewed more favorably than androgynous models (M=4.00) or feminine models (M=2.77). Thus, hypothesis 1 is supported. A model type×priming interaction was not significant (p=0.87). An ANOVA on cognitive response data yielded a significant main effect for VIM. Consistent with study 1, participants viewed the masculine model most positively (Mmasculine=0.53, Mfeminine=−0.21, Mandrogynous=−0.14, F2, 76=10.55, p<0.001) but no significant two-way interactions were evidentFeminine men A significant model type×priming interaction was evident for evaluations (F2, 54=23.70, p<0.001). This interaction showed that feminine men prefer feminine models (M=5.10) over androgynous models (M=4.40) and masculine models (M=2.10), when their private self is salient. Yet when their collective self is salient, they prefer the masculine model (M=4.33) over the feminine model (M=2.88) and androgynous model (M=3.20).

 

Thus, hypothesis 2 is supported.An analysis of cognitive responses revealed a significant model×prime interaction for VIM (F2, 54=14.39, p<0.001). When their private self is salient, feminine men think favorably about the feminine model (M=1.20) rather than the masculine model (M=−0.50) or androgynous model (M=0.60). Yet a salient collective self results in positive thoughts reported about the masculine model (M= 1.17) and negative thoughts about the feminine model (M=−1.00) and androgynous model (M=−0.80). Androgynous men A model type×priming interaction was evident for evaluations (F2, 66=8.78, p<0.001). When the private self is salient, androgynous men prefer androgynous models (M=6.00) rather than feminine (M=4.90) or masculine models (M=4.65). Yet when the collective self is salient, these men state a preference for masculine models (M=5.38) over androgynous models (M=3.63) or feminine models (M=3.69; Table 1).

 

Thus, hypothesis 3 is supported. For cognitive responses, androgynous men exhibit a significant model×prime interaction for VIM (F2, 68= 8.76, p<0.001) but not for VNIM (p>0.92). Specifically, in private, these men think more favorably of androgynous models (M=1.13) than masculine (M=0.20) or feminine models (M=−0.20). In contrast, under collective self conditions, these men report thinking more favorably of the masculine model (M=0.44) rather than the androgynous model (M=−0.17) or feminine model (M=−0.25).

 

4.3.3 Tests of mediation(H4)

To test for mediation, we followed three steps (Baron and Kenny 1986). First, as shown, model gender identity, self-construal, and participant gender identity should interact to affect evaluations. Second, the mediator, classification expectations (CE), is significantly affected by this interaction. Specifically, a significant model gender identity×self-construal×participant gender identity interaction was evident for CE (F4, 198=3.52, p<0.01). Third, including CE as a covariate in the MANOVAs from step 1 weakens the previously significant model type×priming×participant gender identity interaction for evaluations (F4, 197=4.78, p<0.01). The effect sizes for this interaction were reduced by 28.57% (i.e., ω2 =0.05 vs 0.07) suggesting partial mediation. Next, we repeated our two-way analyses. For masculine men, the model×prime interaction for evaluations remained nonsignificant (p=0.87). However, the main effect for model type on evaluations was reduced by 33.33% (F2, 75=6.35, p<0.001, ω2 =0.12 vs 0.18). For feminine men, including CE as a covariate resulted in a weaker model×prime interaction for evaluations (F2, 54=18.75, p<0.001) with the effect size reduced by 20.93% (ω2 =0.34 vs 0.43). For androgynous men, including CE as a covariate resulted in weaker model×prime interactions for evaluations (F2, 65=3.95, p<0.05) and a reduction in the effect size of 53.33% (ω2 =0.07 vs 0.15). Thus, hypothesis 4 is partially supported.

 

4.4 Discussion

Study 2 supports the hypotheses and converges with study 1 with a different product context. Furthermore, the mediation analyses suggests that the effect of model gender identity, self-construal, and participant gender identity is mediated by classification expectations. For example, when primed with their collective self, feminine men exhibit a concern they will be classified by others as similar to a feminine model (M=5.10; Table 1). Androgynous men exhibit a similar concern for classification by others in a collective self context (M=5.13; Table 1). 5 General discussion The present research shows that male responses to male models in advertising is influenced by their gender identity, self-construal, and the perceived gender identity of the model. The findings contribute to research on gender effects in marketing.

 

We show that considering model gender identity in isolation only applies to masculine men. In contrast, the responses of feminine men and androgynous men are influenced by self-construal salience. Our findings suggest that classification expectations provide insights for exploring the process that underlies gender identity differences in how men evaluate male models. Furthermore, we extend the findings of Bosson et al. (2005) by showing that the responses of feminine men to advertising in a collective self context is influenced by concerns of being correctly classified as feminine which results in them supporting traditional masculinity.

 

When the collective self is salient, the expectation that they will be revealed as feminine drives feminine men to endorse the masculine model and shun the feminine model. A useful avenue for future research to explore this result involves concealable stigma. Stigma involves some characteristic individuals possess, or are thought to possess, that conveys a social identity which is devalued in a particular social context (Smart and Wegner 1999). Frable et al. (1998) assert that concealable stigma effects may only have an effect when a person’s social group membership is salient. Our findings support this view, as feminine and androgynous men conceal their positive view of feminine models when their collective self is salient, and instead endorse the masculine model.

 

Thus, male femininity appears to represent a selfperceived undesirable characteristic to be hidden from discovery and the negative impressions or potential backlash from other men (Rudman and Fairchild 2004). Alternatively, the positive view that feminine men have in private of feminine models, along with their classification expectations of being recognized as masculine in private, may represent an evoked fantasy of social acceptance (Martin 2004). A limitation of this research is we did not measure sexual orientation which could be used as a covariate in future research. Similarly, although the evidence of the validity of the BSRI has been offered by researchers (e.g., Holt and Ellis 1998), Palan et al. (1999) suggest that changes in the sociocultural environment since the BSRI was developed in the 1970s may have influenced its validity.

 

To this end, researchers should consider developing new measures of gender identity which may offer additional insights into how men respond to male models in advertising. For advertising practice, our research suggests that masculine models like the Marlboro man are not the only alternative for advertisers. Instead, males can be segmented in terms of gender identity. Masculine men prefer masculine models. For these consumers, feminine models should be avoided. Thus, these men support the current widespread use of masculine models in advertising (Ganahl et al. 2003). Yet the opportunity exists for using other model gender identities to target segments beyond masculine men.

 

Specifically, advertisers can use gender identity congruency (i.e., feminine models targeting feminine men and androgynous models targeting androgynous men), but only for advertising appeals that emphasize the private self (e.g., a solitary consumption activity) and which are viewed in a social setting away from other males. Masculine models may appear to be a useful ingredient of advertising when targeting men in a collective self setting (e.g., advertising in a sports bar), but this overlooks how feminine and androgynous male consumers disguise their advertising preferences when the collective self is salient. Acknowledgements The authors thank the editors, the anonymous reviewers, Cristel Russell, David Griffith, and Simon Pervan for the helpful comments.

 

References

Aaker, J. L. (1999). The malleable self: the role of self-expression in persuasion. Journal of Marketing Research, 36, 45–57. (February) doi:10.2307/3151914. Aaker, J. L., & Lee, A. Y. (2001). ‘I’ seek pleasures and ‘we’ avoid pains: the role of self-regulatory goals in information processing and persuasion. The Journal of Consumer Research, 28(1), 33–49. doi:10.1086/321946. Aarts, H., & Dijksterhuis, A. (2003). The silence of the library: environment, situational norm, and social behavior. Journal of Personality and Social Psychology, 84(1), 18–28. doi:10.1037/0022-3514.84.1.18. Ahluwalia, R. (2002). How prevalent is the negativity effect in consumer environments. The Journal of Consumer Research, 29, 270–279. (September) doi:10.1086/341576. Anderson, K. L. (1986). Androgyny, flexibility, and individualism. Journal of Personality Assessment, 50 (2), 265–278. doi:10.1207/s15327752jpa5002_13. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Beere, C. A. (1990). Gender roles: a handbook of tests and measures. New York: Greenwood. Bem, S. L. (1974). The measurement of psychological androgyny. Journal of Consulting and Clinical Psychology, 42(2), 155–162. doi:10.1037/h0036215. Bem, S. L. (1981). Gender schema theory: a cognitive account of sex typing. Psychological Review, 88(4), 354–364. doi:10.1037/0033-295X.88.4.354. Bosson, J. K., Prewitt-Freilino, J. L., & Taylor, J. N. (2005). Role rigidity: a problem of identity misclassification. Journal of Personality and Social Psychology, 89(4), 552–565. doi:10.1037/0022- 3514.89.4.552. Connell, R. W. (1995). Masculinities. Cambridge: Polity. Deaux, K. (1985). Sex and gender. Annual Review of Psychology, 36, 49–81, (January) doi:10.1146/ annurev.ps.36.020185.000405. Debevec, K., & Iyer, E. (1986). The influence of spokespersons in altering a product’s gender image: implications for advertising effectiveness. Journal of Advertising, 15(4), 12–20. Elliott, R., & Elliott, C. (2005). Idealized images of the male body in advertising: a reader-response exploration. Journal of Marketing Communications, 11(3), 3–19. doi:10.1080/1352726042000263566. Frable, D. E. S., Platt, L., & Hoey, S. (1998). Concealable stigmas and positive self-perceptions: feeling better around similar others. Journal of Personality and Social Psychology, 74(4), 902–922. doi:10.1037/ 0022-3514.74.4.909. Ganahl, D. J., Prinsen, T. J., & Netzley, S. B. (2003). A content analysis of prime time commercials: a contextual framework of gender representations. Sex Roles, 49((9/10)), 545–551. doi:10.1023/ A:1025893025658. Garst, J., & Bodenhausen, G. V. (1997). Advertising’s effects on men’s gender role attitudes. Sex Roles, 36(9/10), 551–572. doi:10.1023/A:1025661806947. Guimond, S., Chatard, A., Martinot, D., Crisp, R. J., & Redersdorff, S. (2006). Social comparison, selfstereotyping, and gender differences in self-construals. Journal of Personality and Social Psychology, 90(2), 221–242. doi:10.1037/0022-3514.90.2.221. Gürhan-Canli, Z., & Maheswaran, D. (2000). Cultural variations in country of origin effects. JMR, Journal of Marketing Research, 37, 309–317, (August) doi:10.1509/jmkr.37.3.309.18778. Holt, C. L., & Ellis, J. B. (1998). Assessing the current validity of the BEM sex-role inventory. Sex Roles, 39(11/12), 929–941. doi:10.1023/A:1018836923919. Holt, D. B., & Thompson, C. J. (2004). Man-of-action heroes: the pursuit of heroic masculinity in everyday consumption. The Journal of Consumer Research, 31, 425–440. (September) doi:10.1086/422120. Ickes, W., Schermer, B., & Steeno, J. (1979). Sex and sex role influences in same-sex dyads. Social Psychology Quarterly, 42(4), 373–385. doi:10.2307/3033807. Jaffe, L. J. (1994). The unique predictive ability of sex-role identity in explaining women’s response to advertising. Psychology and Marketing, 11(5), 467–482. doi:10.1002/mar.4220110504. Jaffe, L. J., & Berger, P. D. (1988). Impact on purchase intent of sex-role identity and product positioning. Psychology and Marketing, 5(3), 259–271. Levant, R. F., & Richmond, K. (2007). A review of research on masculinity ideologies using the male role norms inventory. Journal of Men’s Studies, 15(2), 130–146. doi:10.3149/jms.1502.130. 366 Market Lett (2009) 20:353–367 Maas, A., Cadinu, M., Guarnieri, G., & Grasselli, A. (2003). Sexual harassment under social identity threat: the computer harassment paradigm. Journal of Personality and Social Psychology, 85(5), 853– 870. doi:10.1037/0022-3514.85.5.853. Martin, B. A. S. (2004). Using the imagination: consumer evoking and thematizing of the fantastic imaginary. The Journal of Consumer Research, 31, 136–149, (June) doi:10.1086/383430. Moore, T. M., & Stuart, G. L. (2004). Effects of masculine gender role stress on men’s cognitive, affective, physiological and aggressive responses to intimate conflict situations. Psychology of Men & Masculinity, 5(2), 132–142. doi:10.1037/1524-9220.5.2.132. Palan, K. M., Areni, C. S., & Kiecker, P. (1999). Reexamining masculinity, femininity, and gender identity scales. Marketing Letters, 10(4), 363–377. doi:10.1023/A:1008110204546. Parsons, T., & Bales, R. F. (1955). Family Structure and the Socialization of the Child. New York: Free Press. Patterson, M., & Elliott, R. (2002). Negotiating masculinities: advertising and the inversion of the male gaze. Consumption, Markets and Culture, 5(3), 231–246. doi:10.1080/10253860290031631. Prior, M. (2004). With metrosexuals in mid-America, merchants explore ‘masculine side’. Drug Store News, 26(8), 63–65. Reed II, A. (2004). Activating the self-importance of consumer selves: exploring identity salience effects on judgments. The Journal of Consumer Research, 31, 286–295, (September) doi:10.1086/422108. Rudman, L. A., & Fairchild, K. (2004). Reactions to counterstereotypic behavior: the role of backlash in cultural stereotype maintenance. Journal of Personality and Social Psychology, 87(2), 157–176. doi:10.1037/0022-3514.87.2.157. Smart, L., & Wegner, D. M. (1999). Covering up what can’t be seen: concealable stigma and mental control. Journal of Personality and Social Psychology, 77(3), 474–486. doi:10.1037/0022-3514.77.3.474. Smiler, A. P. (2004). Thirty years after the discovery of gender: psychological concepts and measures of masculinity. Sex Roles, 50(1/2), 15–26. doi:10.1023/B:SERS.0000011069.02279.4c. Stern, B. B. (1988). Sex-role self-concept measures and marketing: a research note. Psychology and Marketing, 5(1), 85–99. doi:10.1002/mar.4220050107. Triandis, H. C. (1989). The self and social behavior in differing cultural contexts. Psychological Review, 96(3), 506–520. doi:10.1037/0033-295X.96.3.506. Vigorito, A. J., & Curry, T. J. (1998). Marketing masculinity: gender identity and popular magazines. Sex Roles, 39(1/2), 135–152. doi:10.1023/A:1018838102112. Wade, J. C., & Brittan-Powell, C. (2001). Men’s attitudes toward race and gender equity: the importance of maculinity ideology, gender-related traits, and reference group identity dependence. Psychology of Men & Masculinity, 2(1), 42–50. doi:10.1037/1524-9220.2.1.42. Ybarra, O., & Trafimow, D. (1998). How priming the private self or collective self affects the relative weight of attitudes and subjective norms. Personality and Social Psychology Bulletin, 24(4), 362–370. doi:10.1177/0146167298244003

Email Advertising: Exploratory Insights from Finland

BRETT A. S. MARTIN University of Auckland Business School bas.martin@aucklancl.ac.nz

JOEL VAN DURME University of Auckland Business Schoolj.vandurme@auckiand.ac.nz

MIKA RAULAS Institute of Direct Marketing Excellence Helsinki School of Economics raulas@hkkk.fi

MARKO IVIERISAVO Institute of Direct Marketing Excellence Helsinki School of Economics merisavo@hkkk.fi

 

Since the advent of the internet, much speculation has ensued regarding its tangible benefits to business. This article looks at the effectiveness of email advertising to promote information to consumers. Within this email promotion context, and using data from a survey of 838 female Finnish consumers of a major international cosmetics brand, we investigate consumer perceptions of email advertising.

Specifically, within an exploratory research context we address two research questions: (1) What email advertising factors may influence visits to the company website? and (2) What email advertising factors may influence visits to a physical (i.e., bricks-and-mortar) company sales outlet? Results suggest that email advertisers should strive to generate emails that are perceived as useful. Useful emails appear to influence consumers to visit the store primarily to either buy the product or view the product firsthand, rather than visit the company website. However, as consumers could not buy the advertised products from the website, these findings should be regarded as preliminary.

Factors influencing perceptions of email advertising usefulness are explored along with limitations and future research directions.

EMAIL ADVERTISING,

where email is usvd as a vehicle for the distribution of promotional messages, is fast becoming an important advertising tool. Email advertising revenue totaled $948 million in 2001 and has been forecasted to increase by 32.91 percent to $1.26 billion in 2002 (Gartner, 2002) and to $7.3 billion by 2005 (Bcardi, 2001). indeed by 2004, marketers are predicted to send almost 210 billion email messages to consumers (Schwartz, 2000). Weil-known organizations currently using email to contact consumers include Barnes and Noble, Borders, Hcrshey Foods, and J.C. Penney {Landau, 2001; Schwartz, 2000; Weidlich, 2001).

 

Reasons suggested for the popularity of email advertising include, first, that email is cheaper than traditional direct mail with costs ranging from S5 to $7 per thousand consumer addresses, as opposed to $500 to $700 per thousand for direct mail (Gartner, 2002). Second, email advertising has been heralded as producing faster response times from consumers (Brown, 2002; Rickman, 2001). Gartner (2002) reports that consumers respond within 10 business days to an email campaign as opposed to four to six weeks for a direct mail campaign. Email advertising also allows for a rapid dissemLnation of an advertisement to a global target market. Third, email can encourage interactivity with consumers by including hyperlinks in the email (Brown, 2002; Garden, 2002).

 

These hyperlinks can invite consumers, for example, to visit the company’s website by clicking on the hyperlink in the email. Recent research undertaken by practitioners indicates that consumers are interested in email marketing. For instance, a survey by DoubleClick of 1,015 respondents reveals that 77 percent of consumers wish to receive promotional offers by email. Further, for 64 peirent of consumers, email is the most popular means to learn about new promotions, products, and services (DoubleChck, 2002). Although commercially important, however, email advertising has been relatively neglected by academic research.

 

In this article, we address this gap by exploring perceptions of email advertising using a sample of female consumers. Within this exploratory research context, we address two research questions: RQl. What email advertising factors may influence visits to the company website? RQ2. What email advertising factors may influence visits to a physical (i.e., bricks-and-mortar) company sales outlet? These research questions are examined using survey data from a sample of Finnish female coasumers. Finland is a Nordic country of 5.2 million consumers situated between Sweden and Russia. It is 130,559 square miles in size, making it similar in size to New Mexico (121,598 square miles). The rationale for studying Finnish consumers relates to their widespread use of the internet. Recent statistics reveal tliat Finland has one of the highest levels of internet penetration in the world with 43.93 percent of the population online (Nua, 2002).

 

Given the forecasts of increased email advertising by marketers, a study of Finnish consumers offers intriguing insights, especially since the international brand that provided the survey data for this study has been successfully engaged in email advertising in Finland since 2000. Thus, the insights we provide on perceptions of email advertising reflect what works for an experienced email advertiser for an international product (i.e., cosmetics), rather than the results of a novice, start-up strategy. Further, these research questions are explored in relation to permission-based email advertising that is most relevant to marketers today. Permission-based email is defined as email that has been requested by the consumer as part of an opt-in scheme (e.g., a consumer fills in their email address on a website aiid agrees to receive information of interest).

 

In effect, marketers are receiving the consumer’s permission to market to them. Permission-based emails are powerful because by signing up to an email list, the consumer is requesting the information from the advertiser rather than simply being exposed to it. Thus, advertisers can rate for permission-based emails is between five and eight percent (Gartner, 2002; Tchong, 2001). Moreover, the aforementioned Doubleclick survey suggests that over 88 percent of respondents have made a purchase as a result of receiving a permission-based email (DoubleClick, 2002). Hence, this study examines permission-based email advertising. In addition to permission-based email, there is also a growing recognition that appropriate email content plays a key role in advertising effectiveness (e.g., Carmichael, 2000; Waring and Martinez, 2002; Yager, 2001). Yet while email content as a whole is increasingly recognized as important, recommendations for what specific content advertisers should use tend Permission-based emails are powerful because .. . tbe consumer is requesting tbe information from tbe tiser ratber tban simply being exposed to it. gain greater effectiveness in the spending of their budgets as the message recipients have already indicated a level of interest in the messages.

 

Consequently, permission email advertising has been heralded as offering consumers rediiced search costs and advertisers an increased level of precision (Rowley and Slack, 2001). This form of email differs from unsolicited commercial email, also known as “spam,” which is an increasing problem for consumers accessing their email. Indeed hy 2006, the average email user is forecasted to receive 3,800 messages each year including 1,400 spam messages (Tchong, 2001). Researcli suggests that response rates for spam email stand at only 1 percent of the email sent out by advertisers, whereas the average clickthrough to be scarce and vague. For example, email content must be “targeted” (Waring and Martinez, 2002), “relevant and clear” (Carmichael, 2000), or “irresistible” (Yager, 2001).

 

An exception is Garden (2002) who suggests (1) providing relevant product information, (2) advertising special deals, and (3) offering invitations to company functions. One of the goals of this study is to explore consumer perceptions of email content to gain some preliminary insight into what specific email topics are regarded as useful. METHOD Sample The data used in this study were collected with the cooperation of the Finnish

division of a prominent cosmetics brand, iidvertising based on 200,000 mk 155 19.8 expectation-maximization metbod (Demp- Education ster. Laird, and Rubin, 1977). A posterior _Elennentary school 146 17.6_ test on tbe means and variances revealed , , , , High school graduate 129 15.6 no differences between tbe variables before and after imputation. Hence, this tech- ….[^ij^.P^pf^ssipnal studies 325 39.3_ nique did not distort tbe original data Higher professional studies 134 16.2 distribution. Finally, a iudgment decision ,, . . , ‘ ^ University degree 94 11.4 was made to remove the sole male respondent. As sucb, this study provides in- :^^?l ^?[^P’^..^i^^ 838 100.0 sights into female perceptions of email.

 

Perceptions of email usefulness were get more information about products, to amount of emails received {fi – -.862, measured by the item, “How useful do get personal assistance from a skillful sales- ;’ < .00], see Table 2). While only a preyou find the emnils received from (brand person, to buy products, and to visit a liminary finding, this suggests that the name)?” (1 = Not at all useful, 5 – Very (brand name) event. more useful an email message, and the useful). Likewise, email content interest greater the number of such emails rewas measured by the item, “What do you RESULTS ceived, the less likely consumers are to thinkabout the contents of the email mes- visit the company’s website.

Since consages?” (1 = Not at all interesting, 5 ^ Research Question 1: sumers are unable to purchase the prodVery interesting). Internet usefulness was Email advertising and website visits ucts via the company website but need to also measured by a 5-point item (1 = Not To address what email advertising factors visit a physical store, these results suggest at all useful, 5 = Very useful). Respon- may influence website visits, we per- that useful email advertising may repredents also indicated what types of email formed a hinary logistic regression that sent a reason for why people to go dimessages they regarded as very useful, analyzed factors affecting whether the com- rectly to the store and purchase. Message types included: (1) information pany website was ever visited.

Specifi- To further investigate this issue of webabout new products, (2) special sales of- cally, the dependent variable measured site visits, we performed a binary logistic fcrings, (3) information about beauty and whether respondents had visited the com- regression on the dependent variable of treatments, (4) information about interest- pany’s website via a hyperlink provided wehsite visits that were independent of email ing new make-up trends, (5) hyperlinks in the email advertising, with the alterna- advertising (i.e., website visits that were to interesting websites, (6) information tive responses of “never,” “once,” or “more not triggered by an email received by the about different upcoming events, and (7) than once.”

 

Independent variables were consumer) to provide insights of comparinformation about competitions. An open- email usefulness, amount of emails re- ison to the previous results. Thus responended question was also included for any ceived, interest generated by the email dents were categorized as visiting the category of importance that was not ad- advertising, usefulness of the internet, and website once a week oi- more, or as less dressed by this group. the importance of the company staying in than once a week.

 

These dichotomies were The amount of emails received from touch. Interestingly, this revealed signifi- chosen to distinguish between frequent the company was measured by the item, cant negative associations for email use- and infrequent visitors to the company’s “How many emails do you remember that fulness (jB = -.363, p < .05) and the website. As displayed in Table 2, signifiyou received from (brand name)?” (None, 1-4, 5-10, over 10). Perceptions of the importance of keeping in touch were mea- DI c o sured by the item, “How important is it that (brand name) is regularly in touch Binary LOgiStJC RegreSSJOn ReSUltS with you?” (1 ^ Not at all important, 5 = Independent Beta Exponential Very important).

Website visits were mea- ^^^^^^^^^ y^^.^,,,^ Variable (standard error) Value sured by the item, “Have you ever visited (brand name)’s internet pages?” (Yes, No). Website visit (emajO Email usefulriess -.363^ (.163) .696 A further variable asked how often re- Amount of emails -.862’^ (.152) .423 spondents had visited these pages (less ^^^^^^ ^.^.^ (general) Email usefulness -:719′:.(,-151) -487 than once or twice a week, once or twice , ^. ., , , ,, ^ Importance in touch -.276^ (.127) .759 a week, or more), amiilarty, wiiether store visits had been inspired by promotional Store visit Email usefulness .298^ (.103) email advertising was measured on a ^rnail interest .647″ (.113) ^P. 3-levcl item (never, once, more than once). , ^ ., n^ ^h , ^ ^ .,^ -, ^c-r Amount of emails .814”(.114) 2.257 Reasons for store visits inspired by email advertising were assessed by respondents ‘^”””- ‘^'”‘ *”””’ ‘”‘-‘^”^’•’^ “”‘.’/ -“”‘-iablct^ -with stnti^tienlli/ signifiiinii beUi eoeffidents. Vnrinbles iiietiiile internet usefuhieas. indicating how many of the followmg rea- ,,^^,^^ Mgnificaut .n p < ,05 sons were applicable: to see products, to ”BeUi significant at p < .001

 

cant negative associations were found for emails, consumers are likely to click on half of tbe respondents as being useful, email usefulness (^8 — -.719, p < ,001) the hyperlink, yet witb every following Namely, information about special sales and tbe importance of tbe company stay- email, they are less likely to visit tbe web- offerings (90.2 percent of respondents), ing in regular contact {(H =• -.276, p < site again. This could be because tbe web- new products (89 percent), competitions .05), Unsurprisingly, tbis indicates tbat if site does not cbange often, or if it does (69.2 percent), and information about an email is perceived as useful, respon- cbange, the website is not perceived to be beauty and treatments (68.7 percent).

 

Indents are less likely to find a need to visit more useful than it was tbe first time it terestingly, information on website bypertbe company website. Likewise, if a con- was accessed. We could imagine consum- links of interest were not seen as useful sumer regards it as important tbat a com- ers accessing tbe site out of initial curios- (43.7 percent). Tbese results suggest tbat pany stays in touch with tbem on a regular ity for instance, but after having visited it consumers are interested in new and timely basis, tben again general visits to tbe com- tjnce tbrougb tbe byperlink, tbey are im- information. pany website are decreased. These results likely to do so again wben additional tend to suggest tbat wben email advertis- emails are received.

 

Finally, tbe extent to Research Question 2: ing is seen as useful by consumers, tbey wbicb tbey find tbe internet in general as Email advertising and store visits do not fee! the need to visit the compa- useful has no significant effect on either For tbis researcb question, a binary logisny’s website, since tbe email is useful and visiting the firm’s website independently tic regression was performed using the provides sufficient information in tbe first (p = .39) or tbrougb a hyperlink (p — .21), dependent variable of store visits and the place.

 

Accordingly, it follows tbat consum- suggesting tbat consumers are driven by same independent variables as for reers wbo place a high importance on a tbe usefulness of tbe email message ratber search question 1, Specifically, the depencompany staying regularly in toucb witb tban a general perception of tbe useful- dent variable separated tbose consumers tbem do so because tbey find tbese emails ness of tbe internet medium as a whole, wbo bad never visited a company saies useful. This view is supported by the poly- Overall, these results suggest a furtber outlet from tbose who bad visited at least cboric correlation-coefficient between tbe question: If email advertisement useful- once.

 

As sbown in Table 2, tbis analysis importance of staying in toucb and email ness may negatively affect website visits, produced tbree significant positive assouscfulness, which is both higb and statis- wbat types of email content intiuence ciations for email usefulness {p = ,298, tically significant {r = .7S, p = .001). whetber an email is perceived as useful? p < ,05), email interest (^ = .647, p < Two other results are also of interest. As displayed in Table 3, four types of .001), and tbe amount of emails received First, wbetber email content was interest- email content are favored by more tban by tbe consumer (p = .814, p < .001). ing was not a significant predictor of Tbus, tbis preliminary finding suggests consumers visiting tbe website either in- tbat consumers may be more likely to dependently (/’ = .26) or by means of tbe TABI C O ‘^^^^^ ^ store if they perceive emails as hyperlink provided in tbe email (/’= ,63). useful and interesting, and if they have Tbis suggests tbat consumers may be goal- L.rildll O0riL6riL I ildL IVIdKeS received many email advertisements from driven and tbat tbey look for information a n Email uSefu l tbe company, tbat is useful to tbeir purposes, rather Tbis indicates that keeping in contact ,, 114 ^ c J ll Variable Percent .^, , ,, , than merely interestmg.

 

Second, tbe witb consumers by email may make conamount of emails received from the com- Special sales offerings 90,2 sumers more likely to visit tbe store dipany was also not significant (p = .16). information about new products 89.0 ^^^^ly ratber Uian visit tbe website. In Tbis suggests tbat sending out large num- otber words, tbey forego tbe website and Competitions 69.2 . , bers of emails to consumers does not make go straigbt to tbe store. Wby is this? An them any more or less likely to visit tbe Information about beauty ar\alysis of frequencies indicates tbat the company website independently. How- ….^.’^.’^..^-.”.^.^.^IT?,?.’?,!^ .??.”.7 reason wh y consumers visit a store is to ever, since tbe amount of emails received Information about different eitber buy the product (40.4 percent of from tbe company makes it less likely events 43.9 respondents) or to see the product firsttbat a consumer will visit the companv ,., ,..,,,. , ^^ -, band (40,1 percent). To a lesser degree, ^ – Website hyperlinks 43.7 ^ jr ‘ & – website via a hyperlink in an email, this consumers visit tbe store to gain addisuggests tbat after receiving relatively few i^.^yy,,[l^^.’^?.^.P..^,^^.”.^^ ^.^:?. tional product information (28.8 percent)

. . . keeping in contact with consumers by email may make consumers more iikely to visit the store directiy rather than visit the website. Consumers also visit the store for the personal assistance provided by sales representatives (19.1 percent), whereas attending in-store events (6,7 percent) do not appear to be a dominant reason for store visits. These exploratory results suggest that while much product information can be obtained by email or from the website, consumers presumably need to \’isit a store to experience other sensory aspects for an experiential product like cosmetics (e,g,, the smell of a new fragrance).

 

DISCUSSION

 

The purpose of this study was to explore consumer perceptions of email advertising. Within this exploratory context, we studied what aspects of email advertising may result in consumers visiting, first, a company website, and second, a physical (i.e., bricks-and-mortar) company sales outlet. We found that visits to the company website appeared to be less likely the more useful the email advertisement, and the more emails received by the consumer from the advertising company. Instead, consumers who viewed emails as useful were more likely to visit the physical store.

 

Our results suggest that the reason for a store visit is usually for consumers to either buy the product or to study it firsthand. As noted by Kover (2001), the web is ideally suited to products that do not involve human interaction with people or objects. In the case of cosmetics with fragrances or makeup products, such as lipstick, it is understandable that consumers visit the store to see if the product advertised by email suits them. Consumers who find emails useful appear to w^ant the company to stay in regular contact with them, suggesting that email offers advertisers the opportunity to become an important avenue for consumers to obtain information.

 

Likewise, consumers who received many email advertisements appear to be more likely to visit the store. We also found that the perceived usefulness of the internet medium as a whole had no effect on either website visits or store visits. This suggests that consumers may be goal-oriented, and that they value email ad\’ertisements that are useful, rather than merely interesting. Useful email content included special sales offerings, new products, competitions, and information about beauty and treatments. Interestingly, sending consumers hyperLinks in emails was not viewed as useful.

 

This is perhaps surprising given the suggested benefits of hyperlinks as, for example, allowing consumers to obtain more information (see, e,g,, Gallagher, Fosters, and Parsons, 2001). Our results therefore suggest a possible qualification to the benefits of offering consumers hyperlinks in the context of email advertising, Further, previous research in the field of print advertising offers theoretical support for tliis result. This research suggests that consumers consider purchase-specific advertising copy, such as information on the attributes of specific products, as more relevant than more general advertising claims, such as advertising the product class in general (Fernandez and Rosen, 2000). From this perspective, as was found in our results, email advertising copy regarding price and new product information should he viewed as more useful by consumers than general hyperlinks.

 

Limitations and future research A variety of limitations should be acknowledged. First, the sample was limited to females, which limits the generalizability of our findings. This issue is relevant given that gender differences have been found in how consumers react to advertising and process information (e.g,, Martin, 2003; Meyers-Levy and Sterntbal, 1991), Thus, our results should be regarded as an exploratory study into female perceptions of email advertising, rather than being generalizable to the wider population of internet users. Future research should examine email advertising using genderbalanced samples of males and females, and could study populations from other cultures. In addition, as suggested by a reviewer, data could be collected involving more diverse products and sampling frame. For this study we used data from a single site and single brand.

 

To improve predictive ability, researchers should employ two or three data sets from different e-commerce sites. Second, most of the items in our study were single item measures. However, multi-item measures offer the opportunity to tap differing aspects of a construct (Robinson, Shaver, and Wrightsman, 1991). Further, multi-item measures allow estimates of reliability to be calculated and the use of statistical techniques, such as structural equation modeling, to be considered (Wanous, Reichers, and Hudy, 1997). Therefore, future research should employ multi-item measures. Third, while we suggest that characteristics of email advertisements may influence consumers to visit physical stores, this finding requires qualification. It is important to note that the website in this study did not allow for on-line purchases.

 

Consumers had to visit a physical store if they wanted the product. Thus, our conclusion should be regarded as a preliminary finding, A stronger test would be provided by studying a site where consumers can choose to buy the advertised product from the company website or at a physical store and to then study what consumers choose to do.

 

This would offer interesting insights into why consumers may choose to visit a physical store even though they can buy the product on-line. The results ol this study suggest a number of intriguing avenues for future research. For example, given that the importance of the interactive capabilities of the internet appears widely accepted (e.g., Cho and Leckenby, 1999; Yoo and Stout, 2001; Yoon and Kim, 2001), a natural extension of this study would be to examine consumer email responses to email advertising. Two areas in particular are of interest. First, consumer responses to the advertiser.

 

Researchers have argued that in the digital domain, marketers and consumers can shape the content of promotional messages together (Rowley and Slack, 2001). Likewise, since liighly focused, customized communications can be beneficial to building long-term relationships (Arnold and Tapp, 2001), it would be useful to explore how an interactive email response to email advertisements aids the development of the relationship between marketers and their consumers. Second, since email offers the convenient function of forwarding messages received to other people, the forwarding of email advertisements to other consumers in terms of word-of-mouth influence and penetration should also be examined.

 

A further avenue tor future research involves the use of email advertising in conjunction with other media. Scholars have highlighted the tieed to explore the proper mix for marketers of online and traditional media (Kover, 1999; Sheehan and Doherty, 20U1). This is particularly relevant given predictions that the internet will become an important component of future Integrated Marketing Communications (Brackett and Carr, 2001). Further, research suggests that email use does not detract from the television viewing time of consumers (Coffey and Stipp, 1997), which offers the opportunity for synergistic mixes of email advertising and more traditional advertising media to be investigated, BRETT A. S. MARTIN is a senior lecturer of marketing at the University of Auckland, New Zealand, He received his Ph.D. in marketing from the University of Otago, New Zealand. His teaching interests include advertising, consumer behavior, and e^;ommerce.

 

His research has been published in journals such as Psyctiology & Marketing and ihe Journal of Advertising Research. JOEL VAN OURME is a lecturer of marketing at the University of Auckiand, New Zealand. His teaching interests include marketing strategy and marketing communications. His research has been published in journals sucli as Marketing Theory. MiKA RAULAS is a director of the Institute of Direct Marketing Excellence at the Helsinki Schooi of Economics. Finiand. His research interests include direct marketing, consumer relationship management, digital marketing channels, and electronic business. MARKO MERISAVO is a researcher at the Institute of Direct Marketing Excellence at Helsinki School of Economics, Finland.

 

His research interests include email marketing, consumer relationship management, and brand management. REFERENCES AcEE, T., tind B. A. S. MARTIN. “Pliinncd or Impulse Purchasfs? How to Create HfffCtive Lnfomerciiils.” loiirim! of Adivrtising Research 41, 6 (2001): 35^2. ARNOLD, M. ]., and S. R. TAIT. “The Effects of Direct Markfting Techniques on Performances: An Application to Arts Organizations.” journal of Interactive Markethig 15, 3 (2001): 41-52. BEAKDI, C. “This Party’s Just Started.” Adverti:}ing Age, 72, 6 (2001): 14. ETr, L. K., and B. N. CAiui. “Cyberspace Advertising vs. Other Media: Consumer vs. Mature Student Audiences.” journal of Advertising Research 41, 5 (200t): 23-32. BKOWN, W. J. “E-Mail Marketing: A Better Alternative?” Lodging Hospitatit}/ 58, 5 (2002): 20. CARMICHAEL, M. “HOW to Keep E-mail Recipients Happy.” Advertising Age’s Business Marketing 85, 1 (2000): 33. CHO, C , and }. D. LECKENBY. “Interactivity as a Measure of Advertising Effecti\’eness: Antecedents and Consequences of Interactivity in Web Advertising.” In Proceedings of the 1999 Conference of the American Acadenni of Advertising, Marilyn S. Roberts, ed. Gainesville, PL: American Academy of Advertising, 1999. COFFEY, S., and H. STLPP. “The Interactions between Computer and Television Usage.” journal of Advertising Ret^earch 37, 2 0997): 61-67. DEMPSTER, A., N. LAFIU), and D. RUHIN. “Maximum-Likelihood trom Incomplete Data via the EM Algorithm.” journni of the Roi/a! Statisiical Society, Series B 39 (1977): 1-38. DOUBLECLICK. “DoubleClick’s Dartmail Study Reveals 88″/, of Consumers Have M.ide Purchases as a Result of Permission-Based Hmail.” May 27, 2002: http://wvi’w.doi-ibleclick.com/ us/corporate/press kitFERNANDn:z, K. V., and D. L. ROSEN, “The Effectiveness of Information and Color in Yellow Pages Advertising.” jourtuil of Adivrtisinii 29, 2 (2000): 61-73, GALLAGHER, K., K. D. FOSTER, and J. PARSONS, “Tlie Medium is Not tiie Message: Advertising Effectiveness and Content Evaluation in Print and on the Web.” lourttul of Adivrtisitt^ Research 41, 4 (2001): 57-70. GARDEN, A. “Using Business Email Effectively.” Bu^iuL’s.’S to Bttsincss U, 3 (2002): S. GARTNER. “GartnerG2 Says E-Mail Marketing Campaigns Threaten Traditional Direct Mail Promotions.” March 19, 2002: http:// www3.gartner.com/5_about/press_releases KIM, Y K,, and J, KANG. “The Effects of Ethnicity and Product on Purchase Decision Making.” joiiriiiil of AdiK’rthiiig RefiCiirdi 41, 2 (2001): Kovi-K, A. ]. “Editorial: Anotlier Unsolved Problem.” jotirnal of Advertising Keseardi 39, 3 ): 5. • . “Editorial: A Wake-Up Call Repeated,” Jotiriia! pf Advertising Reseiirdi 41, 4 (2001): 5. LANDAU, P. “JCPeiiney.coni.” Aduvek 42, 46 (2001): IQ12. MAHTEN, B. A. S. “The Influence of Gender on Mood Effects in Advertising.” Psi/dtologi/ & Marketing 20, 3 (2003): 249-73, MEVERS-LEVY, J., ciT\6 B. SIERNIHAL. “Gender Differences in the Use of Message Cues and judgments.” lournal of Marketing Research 2H, 1 (1991): 84-96. NuA. “How Many Online?” May 28, 2002: http://www. nua.ie/sur\’eys/how_many_ online/index.html .Nj, C. “Harnessing the Killer App: E-mail,” February 28, 2001: http://www, digitrends,net/marketing/l 3640_14649.html ROBINSON, J. P., P, R, SHAVER, and L. S. WRIGHTSMAN. Measures of Personality and Social Psifchological Attitudes. Toronto: Academic Press, 1991. ROWLEY, J,, and F. SLACK. “Leveraging Consumer Knowledge—Profiling and Personalisation in E-Business.” liiternuliottid lonrital of Retail ntid Distribution Management 29, 9 (20Q1): 409-15. SCHWARTZ, J. “Marketers Turn to a Simple Tool: E-Mail.” New York Times, December 13, 2000: http://www.nytimes.com/library/tech/00/12/ biztech/teclTiiology/13schw.html SHEEHAN, K, B., and C. DOHERTV. “Re-weaving the Web: Integrating Print and Online Communications.” jottrnal of Interactiiv Marketing 13, 2 (2001): 47-59. TcHONC, M. “E-Mail Marketing Works.” No- \’ember 6, 2001: http://www.iconocast,com/ issLie/9001,5,1101,06,l,htTnl WANOUS, J, P., A. E. REICHERS, and M. J. HUDY. “Overall Job Satisfaction: How Good are SingleItem Measures,” jonrnal of Applied Psydiology 82, 2 (1997): 247-52. WARING, T., and A. MARTINEZ, “Ethical Customer Relationships: A Comparative Analysis of US and French Organizations using Permission-based E-mail Marketing.” jourttal of Dataimsc Marketing 10, 1 (2002): 53-69. W[-:iDLiCH, T. “Hershey Kisses.” Catalog Age 18, 7 (2001): E9-E11, YAGER, T, “Customer, May I? Getting the OK to Sell.” infoWorld 23, 4 (2001): 48, Yoo, C. Y, and P. A. STOUT. “Factors Affecting tJsers’ Interactivity with the Website and the Consequences of Users’ Interactivity.” In Proceedings of the 2001 Conference of the American Academy of Adz’ertising, Charles R. Taylor, ed. Villanova University: American Academy of Advertising, 2001, YooN, S,, and J. KIM. “IS the Internet More Effective than Traditional Media? Factors Affecting the Choice of Media,” lottrnai of Advertising Researdi 41, 6 (2001): 53-60.