Document View

Our research in the developing market economy of Romania employs in-depth interviews and an advertising-trial experiment to examine the effects of attractiveness in advertisements on product evaluations and self-judgments. Our qualitative data indicate that attractive (versus average-looking) models yield favorable ad and product evaluations, but generate both negative and positive self-judgments. Our post-ad exposure experimental data corroborate immediate positive product evaluations and negative self-judgments. The subsequent two-week trial equalizes product evaluations across model conditions (attractive versus average-looking versus no model) and mitigates the negative effect of attractiveness on women's satisfaction with their own attractiveness. Moreover, women who use products advertised by attractive models report more favorable self-attractiveness and self-confidence. [PUBLICATION ABSTRACT]

Much research has documented that consumers' reactions to advertisements are based not only on product or brand claims, but also on other aspects of the advertisement, including spokesperson characteristics and execution cues (MacKen- zie and Lutz 1989; Pham and Avnet 2004). With regard to spokesperson characteristics, advertisers have long been inter- ested in understanding the effects of attractiveness on product evaluations and self-judgments. It is interesting to note that research brought to bear on the effects typically has focused either on advertising/product evaluations or on self-judgments (see Bower 2001 for an exception). With regard to attitude toward the ad, product, and brand, as well as purchase inten- tion, researchers have argued that the favorable impressions generated by an attractive model are transferred to ad and product evaluations via affect transfer (Lutz 1985) or because their attractiveness characteristic may provide relevant infor- mation for attractiveness-relevant products, such as clothing and cosmetics (Kahle and Homer 1985; Kamins 1990). With regard to self-judgments, however, social comparison theory presents a more complex effect of attractive models (Mussweiler 2003; Wood 1989). Specifically, depending on the comparison goal, the comparison process may serve self-evaluation, self- improvement, and self-enhancing goals (Martin and Gentry Romanian women conducted to understand their perspectives on advertising related to cosmetic products. We then discuss our ad-trial field experiment in which we assessed post-ad exposure and post-trial evaluations. The post- ad exposure judgments help determine whether extant Western-based post-ad exposure findings related to attractive models (versus average-looking models and no model) are replicable in this emerging market. They also served as a base of comparison with post-trial product evaluations to enable assessment of the two-week product trial. Thus, our collective work contributes to the ad-trial literature, providing a more realistic consumption context for trying symbolic products (Hoch and Ha 1986; Kempf and Smith 1998; Smith 1993), as well as to a growing dialogue about the impact of exporting idealized media images and consumer culture to the transitional economies of eastern and central Europe.

THEORETICAL BACKGROUND

Advertising and Trial Effects on Ad and Product Evaluations

For decades, marketers have used attractive women to draw attention to and advertise a wide range of brands, products, firms, and industries. From an affect transfer perspective, research suggests that these attractive spokespersons generate positive affect (Kallen and Doughty 1984) that would be transferred to consumers' attitudes toward the brand or product, and result in greater purchase intentions (Lutz 1985). Highly attractive models, however, do not always generate more favorable reactions than normally attractive or unattractive models (Bower and Landreth 2001 ; Caballero and Solomon 1984). Rather, as argued by the match-up hypothesis, they should be most effective when matched with attractiveness-relevant products (e.g., Kahle and Homer 1985; Kamins 1990). Bower and Landreth (2001), however, document that highly attractive models are more effective for beauty-enhancing products (e.g., lipstick), but are not more effective than normally attractive models for problem-solving products (e.g., dandruff shampoo) (see Bloch and Richins [1992] for a classification of attractivenessrelated products). Similarly, if viewers believe a model in an ad possesses a physical characteristic (e.g., muscularity) that indicates the model has improved the appearance because of product use (e.g., exercise equipment), viewers may believe the associated product is responsible for the model's appearance (Lynch and Schüler 1994). Kamins and Gupta (1994) also show that congruence between the spokesperson and the product advertised enhances product evaluations.

Research also suggests that attractive models can affect consumers' price expectations about products. Tsao, Pitt, and Caruana (2005) report that when consumers lack direct experience with the product, they rely on advertising to form inferential price-quality beliefs. Thus, consistent with Bower and Landreth (2001), on exposure to an advertisement, we anticipate that a beautiful model endorsing a beauty-enhancing product signals a higher quality product, which carries a higher price (Rao and Monroe 1989)· Hence, we expect:

H1: For beauty -enhancing products, women viewing ads with highly attractive models versus women viewing ads with average-looking models will have more positive ad attitudes (H1a), higher quality assessments (H1b), higher price judgments (H1c), and more favorable buying intentions of the advertised product (H1d).

H2: For beauty-enhancing products, women viewing ads with highly attractive models versus women viewing a control ad (with no model) will have: more positive ad attitudes (H 2a), higher quality assessments (H2b), higher price judgments (H 2c), and more favorable buying intentions of the advertised product (H2d).

An important stream of work has investigated the combined effects of advertising and trial. One significant finding is that once people have tried highly diagnostic products (i.e., products for which trial offers tangible, credible evidence of the product and its attributes), pretrial advertising has little or no effect on their product evaluations (Hoch and Ha 1986; Kempf and Smith 1998). Similarly, Vakratsas and Ambler (1999) argue that in the context of unambiguous trial, post-ad exposure product beliefs are updated after trial. This resolves uncertainty, and confirms or disconfirms the pretrial expectations about product performance. "Since the validity of one's own senses is rarely questioned, these [trial] beliefs are, at least initially, held with maximal certainty" (Fishbein and Ajzen 1975, p. 132). Furthermore, if the salient attributes presented in the pretrial advertising are mostly experiential (i.e., can be assessed only by directly experiencing the product), advertising does not have a significant effect on post-trial product evaluations (see Kempf and Smith 1998; Micu and Coulter 2007). Rather, trial assists consumers to form brand beliefs. Kempf and Smith (1998) also note that during trial, consumers make inferences about nonadvertised product attributes for which they do not have information. In other words, for example, consumers may infer prices based on the quality of the brand's trial performance. Thus, we expect that consumers' post-trial product evaluations will reflect their assessment of the highly diagnostic product. Hence, we hypothesize that

H3: Product trial will moderate the effects of model attractiveness on post-trial evaluations, such that, after product trial, there will be no difference across model conditions in consumers' post-trial product quality assessments (H3a), price judgments (H3b), and purchase intentions (H3c).

Advertising and Trial Effects on Satisfaction with One's Own Attractiveness

Social comparison theory addresses how people develop selfknowledge and make choices based on their comparisons with others. Festinger (1954) hypothesized that humans have a drive to compare their own abilities and opinions with others for the purpose of self-evaluation, and others have argued that comparison may also serve self-enhancement and selfimprovement goals (see Wood 1989 for a review). Upward comparison (generally prompted by self-improvement and self-evaluation goals) is typically associated with attractive models, or those individuals who are "better off." Martin and Gentry (1997) document that upward comparison generated by self-improvement goals can lead to inspiration, and more favorable perceptions of one's attractiveness. However, research also suggests upward comparison is more likely to result in self-evaluation and deflation when the referent is "extreme," because it is difficult to reach the standard of the referent (Mussweiler 2003; Mussweiler, Rüter, and Epstude 2004; Stapel and Suis 2004). Indeed, an abundance of research has demonstrated that young American women are less satisfied with their appearance after viewing highly attractive models in ads (Martin and Kennedy 1993; Richins 1991).

Several researchers speculate that negative effects of exposure to idealized images have an immediate, but temporary effect on one's self-perceptions (Martin and Gentry 1997; Richins 1991). Some research suggests that this temporary effect is a function of individuals using multiple standards of comparison, and that momentary changes in social settings can make one standard more cognitively accessible than others (Higgins, Strauman, and Klein 1986). Thus, recent advertising exposure can cause attractive models to become temporarily more salient, changing the comparison level for attractiveness judgments for a short time (Richins 1991). As such, women's satisfaction with their own attractiveness may initially be affected by exposure to a particular ad featuring attractive models, but as time elapses (e.g., our two-week trial period), the salience dissipates, and satisfaction with personal appearance is no longer affected by that particular ad. Our research is the first to empirically investigate the timing and nature of these expected temporary effects. Based on these theoretical perspectives, we expect:

H4: Immediately after viewing ads with highly attractive models, women will report lower satisfaction with their own appearance than women viewing ads with average-looking models (H4a), or women viewing a controlad (with no model) (H4b). After the extended (i.e., two-week) trial period, however, the immediate effects of model attractiveness on satisfaction with one's own attractiveness will dissipate, such that there will be no difference across model conditions in consumers' post -trial attractiveness satisfaction (H4c).

Advertising and Trial Effects on Self- Attractiveness and Self-Confidence

McCracken's meaning transfer model (1989) suggests that beautiful models promoting cosmetic products will produce an "attractiveness enhancer" meaning for the endorsed product and the product user. In the context of beauty-enhancing products, McCracken's model suggests that an attractive model in advertising would lead consumers to believe that, through product use, they will be more attractive and will more closely approach that ideal image in the advertisement. Similarly, social comparison theory suggests that attractive models can serve goals of self-improvement, and provide sources of inspiration (Collins 1996; Etcoff 1999; Martin and Gentry 1997). As Peiss suggests, cosmetics advertising offers "the mutual transformation of external appearance and inner well-being" (1998, p. 144). Orth and De Marchi (2007) show that symbolic beliefs (that usually correspond to nonproduct-related attributes and relate to needs for social approval and personal expression) are more likely to be influenced by advertising than experiential beliefs (i.e., beliefs that usually correspond to experiential attributes) that are mostly influenced by trial. Furthermore, functional beliefs (e.g., product quality) are more influenced by physical attributes tested during trial than symbolic beliefs. In conclusion, while attractiveness in advertising can temporarily lower one's satisfaction with one's own appearance (as hypothesized in H4a and H4b) and result in no difference across model conditions in product evaluations (as hypothesized in H3), the theoretical perspectives discussed above argue that exposure to attractive models (as compared to more average-looking models or no models) will bolster consumers' perceptions of their own attractiveness and self-confidence when consumers use the product. Hence, we expect:

H5: Women who viewed the advertisement with the highly attractive model will report feeling more attractive and more self-confident when using the advertised product than those who viewed the average-looking model (H 5a) and those viewing the control (no model) ad (H5b), controlling for satisfaction with one's own attractiveness.

RESEARCH CONTEXT AND METHOD

We tested our hypotheses in the emerging market of Romania. This venue provided a context in which international firms (e.g., Avon, Procter & Gamble, Colgate Palmolive, and L'Oréal) were beginning to enter the market, introducing Western cosmetic brands (PricewaterhouseCoopers 2004). Consistent with an adapted etic approach, we engaged in a preliminary data collection (discussed subsequently) to better understand young Romanian women's perspectives with regard to attractive models in cosmetics advertising. In our experiment, we used the cover story that an international firm planned to launch several cosmetics products in the Romanian market, and that the firm was interested in getting reactions about advertising strategies and product offerings.

Preliminary Data CoUection

We conducted in-depth interviews with eight women age 18-23, four from a major city and four from smaller cities in Romania. This age cohort (sensitized to global brands and with access to a global flow of communications and ideologies) is a particularly relevant target for multinational companies (Kjeldgaard and Askegaard 2006). Our interviews, conducted in Romanian, and later translated into English, followed a structured set of questions, and ranged from 50 to 90 minutes in length. We focused on informants' general perspectives related to advertising and cosmetics, as well as their attitudes and opinions of six advertisements - three cosmetics magazine ads featuring a beautiful model, Josie Maran, and three lipstick ads (one with an attractive model, one with an average-looking woman, and a control ad [no model], described subsequently in the "Development of Ad Stimuli" section) that were used in our experiment (see the Appendix). The authors conducted an across-case analysis and the findings are structured around the research questions and emergent themes (Spiggle 1994).

The Effects of Model Attractiveness on Product Evaluations

Consistent with Western findings (Petty, Cacioppo, and Schumann 1983), the young Romanian women positively evaluate cosmetics advertised by attractive models. Specifically, attractive models serve as signals about product quality and price. Ana, one of the interviewees, staunchly believes that "the product has some advantages if applied on a beautiful woman." Raluca discusses both price and quality as she examines two ads, one with a beautiful model and one with a more average-looking woman: "I would pay a lot more for the lipstick [in an ad with an attractive model] than [in an ad with an average-looking woman]. It also looks of better quality, and this is because of the woman [in the ad]. The beauty is always associated with quality." In reaction to viewing several Maybelline cosmetics ads with model Josie Maran, Cristina comments: "the products are good. If they weren't, the model wouldn't be so attractive." In addition, the young Romanian women appear more likely to contemplate purchasing cosmetics on viewing an ad with an attractive model. For example, Diana reflects, "this is what incites you to buy the lipstick," and Elena enthusiastically endorses the lipstick, "I could buy the product; I like the model and [the way the lipstick] is emphasized." Bianca and Raluca, respectively, echo these sentiments, "if you see a beautiful woman and the product suits her so well, you may want to try the product, too. So, it is an advantage to use a beautiful woman," and "if you see a super beautiful woman on TV, you want to buy the product; beauty is always associated with good things and a good product."

The Effects of Attractive Models on Self-Judgments

Our informants provided a range of reactions to attractive models in terms of social comparison. First and consistent with the negative effects associated with self-evaluation goals, some informants agreed that a comparison with these idealized advertising images can result in disaffection (Coulter, Zaltman, and Coulter 2001; Lasch 1978). Diana comments that there "may be a segment of women who are threatened by these images and will think, 'Oh God, look how beautiful she is.' There is no way that I'll ever buy this product." Raluca agrees, "some people who know they don't look very good could be affected by the [model's] beauty . . . because you think the model looks so good, what's the point to buy the lipstick." Anca thoughtfully comments that typical Romanian women cannot reach these ideal images; they "cannot win." In addition, some informants report not engaging in comparison with beautiful models. Ana, for example, who thinks cosmetics ads must include beautiful models says, "no, no" when asked if she compares herself to them, saying, "I think that would be fruitless." Mihaela also conjectures that "some people consider [attractive models] too perfect, and consequently do not contrast themselves to the models."

Second, our informants engage in realistic-tempered comparison with idealized images in cosmetics ads - they are acutely aware of the discrepancy between their own appearance and the beauty of the models (Scott 2005). They make careful comparisons, and then judge their interest in the product. Cristina reports: "I compare the way they look, because they are beautiful and have a great complexion, and their makeup is beautiful and they look good. I wish I looked like them, of course, but that is a long way off." Similarly, Elena uses Pantene, which "is advertised so heavily, but I think that my hair cannot look like hers. ... I pretend to look like her, but who knows how many stylists arrange her hair before she appears in the ad." Bianca provides a personal perspective of comparing herself to attractive models, yet gauging her own interest in the advertised products: "It is a comparison all the time. If you see an ad, you think all the time how good the person in the ad looked. But I am not stressed about it. ... I think about how I would look wearing that product, and if I think that I won't look good, then I won't buy it."

Finally, as Wood (1989) suggests, some informants compare themselves to these idealized images to serve self-improvement goals, to inspire and inform their cosmetic choices. These informants believe that an attractive model is an integral part of a cosmetics advertisement and provides useful information about how the lipstick looks on someone, and how their own looks might be transformed by product use (Collins 1996; Lockwood and Kunda 1997; Martin and Gentry 1997). Cristina works part-time as a sales clerk and realizes, "I need to take care of myself ... I need to look good," and "if I look better, then I feel better." She learns from models in advertisements and makes cosmetic choices with an eye to what is appropriate for daily wear, as well as for holidays and special activities. When asked what she would include in a magazine ad for a lipstick, Cristina was adamant that "you must show a beautiful woman, to show that cosmetics] can make you beautiful, can make you look good." Similarly, Diana insists, "when you see a beautiful woman, you say, 'the product can make me as beautiful as she is,'" and goes on to discuss the transformative abilities of wearing cosmetics, stating, "the way you look has an impact on the way you feel." Bianca also comments about the attractive model's ability to make her believe that she can be more attractive: "If she looks so good with this lipstick, I should try it, too, maybe I can look as good as she does." Within the past year, Bianca has put more emphasis on her appearance: "I saw in magazines, on TV, and I was attracted by this. I like to be groomed and to wear makeup, to arrange myself. ... I really liked how I look." Mihaela also wears cosmetics to "look better" and have "more confidence in myself."

Summary

Our in-depth interviews offer preliminary evidence that young Romanian women react similarly to women in more developed economies. They associate attractive models with higher quality and higher priced products, and are more inclined to purchase higher quality products. In addition, our informants react both negatively and positively to attractive models in ads, reporting that they may denigrate satisfaction with one's own attractiveness, but coupled with product use, such models may have transformative abilities to shape the informants' own self-attractiveness and self-confidence.

The Ad-Trial Experiment

Experimental Procedure

Romanian women age 18 to 24 participated in the two-session mixed between- within-subjects experiment. In Session 1, participants were randomly assigned to one of three conditions in which they were exposed to an ad booklet that included one target ad for lipstick and two filler ads (one for face powder and one for mascara) advertised by either highly attractive models (n = 69), average-looking models (n = 57), or no models (control group; ? = 68). The ads containing a model

featured only the face of the model. The product image and message content for each product were consistent across the three model conditions. To avoid a confound effect, we used a fictitious brand of cosmetics, Elisa.

We chose lipstick as the focal product (target ads are included in the Appendix) because of its attractiveness-enhancing qualities. Lipstick is "the one cosmetic [women] can't live without. ... It has the highest usage of any cosmetic product" (Ragas and Kozlowski 1998, p. 7). The lipstick ad appeared first among the three cosmetics ads in each booklet (Kahle and Homer 1985), and included highly diagnostic attribute information (e.g., "lasts long on your lips, moisturizing them and leaving them smooth and soft"), which could easily be tested during product trial. The ad copy, the information sheet about the experiment, and the questionnaire were first written in English, then translated into Romanian and retranslated by a native-speaking Romanian. Modifications were made to assure the translations mirrored the English version.

Development of Ad Stimuli

To identify models that shared both local and global standards of attractiveness, a native Romanian female prescreened Internet and magazine images of women with varying levels of attractiveness. She identified seven images consistent with beautiful models used in Romanian cosmetics ads and 10 average-looking women. Then, in a pilot test, 21 young adult Romanian females viewed the women (10 seconds each) in a PowerPoint presentation. The average-looking women were presented first, because Richins (1991) found that participants rated average-looking models lower when they were first exposed to beautiful models. After viewing each image, participants rated the model's attractiveness (1 = "not at all attractive"; 9 = "very attractive"). Based on the results, we selected six women to use as experimental manipulations. Analysis of variance (ANOVA) results indicate a significant difference among the six models, F(5, 120) = 41.93, /> < .001, and the post hoc contrast indicated that the three beautiful women (?? = 8.71, 7.71, and 7.81) were rated significantly more attractive than the three average-looking women, ?? = 4.14, 4.95, and 5.95; /(20) = 14.14,/» < .001.

Session 1

As we have noted, participants in the experiment were told that an international firm planned to launch several of its products in the Romanian market, and that the firm was interested in understanding more about Romanian women and their reactions to ads. Participants first answered seven-point Likert items about the importance of appearance and the effects of cosmetics usage. Next, the participants examined the ad booklet and then responded to a series of questions. The manipulation check, model attractiveness with regard to the lipstick ads, was measured using two, seven-point (1 = strongly disagree; 7 = strongly agree) Likert items, "The model from the ad is attractive," and "The model from the ad is beautiful" (r = .77). Consistent with the manipulation, the highly attractive model (M = 6.25) was perceived as being significantly more attractive than the average-looking model, M = 3.80; t(124) = 12.98,t < .001.

The dependent variables were assessed using seven-point items: attitude toward the ad (very negative/very positive, very bad/very good, very unfavorable/very favorable; Cronbach's α = .90; MacKenzie and Lutz 1989), product quality (very poor/excellent quality), price (very low/very high priced), and purchase intention (not at all likely/very likely). To assess satisfaction with one's own appearance, we used five, sevenpoint items (e.g., "I am pleased with my appearance right now," "I think people find me attractive," and "I feel satisfied with the way my body looks right now"; Cronbach's α = .91) (Heatherton and Polivy 1991; Richins 1991). Past research indicates that exposure to facial pictures of attractive models can influence women's overall satisfaction with appearance (Richins 1991). To reduce potential demand effects related to satisfaction with one's own attractiveness, the questions were embedded in a section preceded by the statement consistent with the cover story, "The company is interested in learning more about you, your tastes, and how you feel about yourself so that they can better understand you as a consumer."

We assessed four potential confounds: product involvement, product knowledge, social comparison tendency, and effect of cosmetics use on attractiveness (using seven-point Likert scale items; 1 = strongly disagree, 7 = strongly agree). Because involvement with cosmetics and product knowledge could differentially impact their attention to and processing of advertising information and product trial experience (Bower and Landreth 2001; Petty, Cacioppo, and Schumann 1983), participants responded to six Likert items (e.g., "Cosmetics are important to me"; Coulter, Price, and Feick 2003) (Cronbach's α = .84); we found no difference across model conditions, M = 4.97; F(2, 187) = .07, p > .05. To assess product knowledge, participants reported their knowledge (1 = not very much and 7 = a great amount) related to four items (e.g., "How much do you know about cosmetic products?"; Cronbach's a = .84; Park, Mothersbaugh, and Feick 1994); again, we found no difference across model conditions, M = 3.96; F(2, 183) = .87, p > .05. To assess social comparison tendency (Bower 2001), participants responded to five Likert items (e.g., "When I see models in cosmetics ads, I think about how well or how badly I look compared to the models"; Cronbach's α = .69; Richins 1991); we found no difference across model conditions, M = 3.75; F(2, 189) = 1.50, p > .05. Finally, to assess perceptions of the impact of cosmetics use on attractiveness, participants responded to one Likert item, "Using cosmetics makes a woman look more attractive"; we found no difference across conditions, ?? = 5.98; F(2, 191) = 1.34, p > .05.

After completing the questionnaire, participants were given the advertised lipstick, which was brought from the United States and was not available in the Romanian market. The label was removed so that no identifier of the manufacturer was evident on the product. Participants were asked to return to Session 2 in two weeks.

Session 2

During Session 2, participants completed a follow-up questionnaire to assess their post-trial assessment of perceived product quality, price, purchase intention, and satisfaction with one's own attractiveness (Cronbach's α = .91). In addition, we also assessed consumers' post-trial evaluations of the advertised lipstick's experiential attributes (i.e., moisturizing and long-lasting), because these attributes can be important determinants of post-trial quality evaluations (Kempf and Smith 1998). Participants also completed seven-point semantic differential items to assess self-attractiveness ("Using this lipstick made me feel less/more attractive") and self-confidence ("Using the lipstick made me less/more confident about myself") (Richins 1991). To determine whether the extent of trial could account for post-trial product evaluations and self-judgments, participants reported the number of days they used the lipstick; we found no difference across experimental conditions, M = 6.80; F(2, 191) = 1.06, p > .05. When debriefed, participants' comments reflected their belief in the cover story.

FINDINGS

Ad and Product Evaluations

To examine HIa through H2d, we conducted a multivariate analysis of variance (MANOVA) using ad attitude, product evaluations (i.e., quality and price), and purchase intention as dependent variables, and model condition as the independent variable. Our results demonstrate a significant model attractiveness effect, Wilks's λ = .86, F(S, 366) = 3.61, p < .001, partial η^sup 2^ = .07.

We used univariate ANOVAs to examine our individual hypotheses (see Table 1 for means, F-values, and Bonferroni post hoc comparison tests). Specifically, model attractiveness has a significant effect on attitude toward the ad, and consistent with HIa, the post hoc comparison indicates that participants viewing the ad with the highly attractive model (?? = 5.79) have significantly more positive ad attitudes than those viewing the average-looking model (M = 4.99). In addition, our post- ad exposure measures indicate significant effects of attractiveness on product quality perceptions and price perceptions. Consistent with HIb and HIc, respectively, Bonferroni post hoc comparisons indicate that participants viewing the ad with the highly attractive model, compared to participants from the average-looking model condition, rated the lipstick of higher quality (M = 5.81 versus M = 5.12) and of higher price (M = 5.25 versus M = 4.56). Furthermore, participants in the highly attractive condition (M = 4.78) indicated a slightly greater purchase intention than those in the average-looking model condition (M = 4.26). However, the difference was not statistically significant (p > .05); therefore, H1d was not supported.

Hypothesis 2 posited that attractive models compared with no models would result in more positive attitudes toward the ad (H2a), more favorable quality evaluations (H2b), higher price perceptions (H2c), and stronger purchase intentions (H2d). We found support only for H2c, as participants viewing the ad with the highly attractive versus average-looking model believed the lipstick was more expensive (M = 5.25 versus M = 4.79). It is interesting to note that we found no significant differences on the other dependent measures between participants in the attractive model and control conditions, which is consistent with findings reported by Richins (1991).

Hypothesis 3 posited that product trial would moderate the effect of attractiveness on post-trial evaluations, resulting in participants reporting the same level of post-trial product-related evaluations. To examine this hypothesis, we conducted a mixed between-within subjects ANOVA with Session 1 model condition as the manipulated variable and the dependent variable as the repeated measure. First, with regard to product quality, we found no difference in the average post-ad and average post-trial product quality evaluations, Wilks's λ = 1.00, F(I, 19D = .03, p > .05, but a significant product quality X ad condition interaction, Wilks's ? = .93, F(2, 191) = 7.53, p < .001; ?2 = .07. Specifically, in contrast to the attractive model post-ad effect and consistent with H3a, we found no difference in the post-trial product quality judgment across model conditions. Furthermore, we examined participants' post-trial beliefs about the experiential attributes contained in the ad, and also found no difference across the experimental conditions in assessment of the attributes: moisturizing, M = 5.68; F(2, 191) = .85, p > .05, and long-lasting, M = 5.05; F(2, 191) = .27, p > .05. These means, significantly higher than the scale midpoint (p < .001), support the lipstick product quality judgments. To understand the shift in product quality judgments, we conducted paired sample (post- ad exposure versus post-trial) /-tests. The results indicate that the ad with the highly attractive model generated significantly less favorable quality ratings post-trial, M^sub trial^ = 5.38 versus M^sub ad^ = 5.81;t(68) = 3.03, /> < .01, whereas participants viewing the average-looking model reported significantly more favorable quality ratings post-trial, M^sub trial^ = 5.56 versus M^sub ad^ = 5.12; t(56) = 2.61, ? < .05.

Our results with regard to price perceptions are similar to the product quality evaluations; we found no difference in the average post-ad exposure and post-trial price perceptions across the three model conditions, Wilks's λ = .98, F(I, 191) = 3.11,/> > .05, but a significant price perception X ad condition interaction, Wilks's λ = .97, F(2, 191) = 3.14, p < .05, η^sup 2^ = .03. Again, in contrast to the attractive model effect after ad exposure, participants' post-trial price perceptions were not different across the three model conditions, supporting H3b. The paired (post-ad exposure versus posttrial) /-tests indicate that, after using the lipstick, participants viewing the highly attractive model report significantly lower price perceptions post-trial, M^sub trial^ = 4.80; M^sub ad^ = 5.25; /(68) = 2.89, p < 01. There were no differences in the posttrial versus post- ad exposure price perceptions for participants viewing the average-looking models, M^sub trial^ =4.67; M^sub ad^ = 5.56; t(56) = .69,p > .05.

Finally, we found a significant difference between the average post- ad exposure and post-trial purchase intention, Wilks's λ = .80, F(1, 191) = 47.90, p < .001, η^sup 2^ = .20, as well as a significant purchase intention X ad condition interaction, Wilks's λ = .94, F(2, 191) = 6.07, p < .01, η^sup 2^ = .06. Our results indicate no difference in post-trial purchase intention across ad conditions, supporting H3c. It is interesting to note that the paired (post-ad exposure versus post-trial) t-tests indicate that trial experience generated significantly greater intention among participants in both the highly attractive, M^sub trial^ = 5.45; M^sub ad^ = 4.78; t(68) = 3.37, p < .001, and averagelooking, M^sub trial^ = 5.67; M^sub ad^ = 4.26; /(56) = 7.71, p < .001, model conditions. Thus, consumers' ability to try the product enhanced the possibility of purchase likelihood.

Assessment of Satisfaction with One's Own Attractiveness

We hypothesized that women viewing attractive models would report less favorable satisfaction with their own attractiveness than those viewing average-looking models and those in the control group. An ANOVA with model condition as the independent variable and satisfaction with one's own attractiveness as the dependent variable and Bonferroni post hoc comparisons indicate that immediately after ad exposure, participants in the attractive model condition rated their satisfaction with their attractiveness significantly lower (?? = 4.86) than those in the average-looking model (M = 5.33) and the control (?? = 5.31) conditions, supporting H4a and H4b, respectively.

H4c posited that the two-week product trial would eliminate the post- ad exposure effects of model attractiveness on satisfaction with one's own attractiveness. A mixed betweenwithin subjects ANOVA with Session 1 model condition as the manipulated variable and satisfaction with one's own attractiveness as the repeated measure revealed that after the two- week period, model attractiveness no longer affected satisfaction with one's own attractiveness, Wilks's ? = .99, F(I, 190) = .55, p > .05; the interaction of satisfaction with appearance and ad condition was also nonsignificant, Wilks's λ = .99, F(2, 190) = .48, p > .05. As expected, we observed no differences in satisfaction with appearance across the model conditions after trial.

Self-Perceived Attractiveness and Self-Confidence

H5 posited that participants who used the lipstick advertised by the highly attractive models (versus average-looking model or no models) would report more favorable self-perceived attractiveness and self-confidence. We included satisfaction with one's own attractiveness as a covariate because it was statistically significant in adjusting each of the dependent variables (Tabachnick and Fidell 1989, pp. 345-346). The correlations between the items measuring satisfaction with one's own attractiveness (rs between .67 and .74) are much higher than the correlations between those items and (1) self-perceived attractiveness (rs between .08 and .21), as well as (2) selfconfidence (rs between .12 and .27), indicating discriminant validity (Churchill 1979). Consistent with our hypotheses, we found a significant difference in post-trial self-perceived attractiveness and self-confidence across model conditions (satisfaction with appearance was a significant covariate in the respective analyses, F[I, 190] = 24.13, p < .001; F[I, 190] = 24.31, p < .001). Bonferroni post hoc analyses indicate that women in the highly attractive model condition report feeling more attractive and more self-confident wearing the lipstick (M = 5.50; M = 5.43) than participants in the averagelooking model condition (M = 5.06; M = 5.02) and those in the control condition (M = 5.11; M = 5.03).

GENERAL DISCUSSION

Our research brings together multiple theoretical perspectives to hypothesize the effects of attractive models on product evaluations and self-judgments, as well as the extent to which product trial and the time associated with trial alter these effects. We focused on cosmetics (lipstick, in particular), a product category for which the attractiveness of the model is an important signal (Bower and Landreth 2001; Kahle and Homer 1985; Kamins 1990), and the venue for our research was the emerging market of Romania.

Post-Ad Exposure Effects

Our research considered the immediate post- ad exposure effects of an attractive model on product evaluations and self-judgments. It is interesting to note that with regard to attitude toward the ad, product quality, and purchase intention, our experiment participants responded equally favorably to ads with attractive models and ads with no models. These findings, considered on their own merits, argue for promoting cosmetics without using an attractive model - thereby taking advantage of the immediate positive reactions to the ad and the product, and avoiding the immediate negative impact of the beautiful models on women's satisfaction with their own attractiveness. However, our qualitative study informants show an awareness and expectation of attractive models in cosmetics ads - as Mihaela comments, "I think the ad should have a beautiful woman, because in Romania, since I first looked in a magazine, [beautiful models] are all I saw." Our experimental participants in the no-model condition reported a similar expectation of attractive models (6.34 on a seven-point scale) if a model was included in the ad.

The power of attractive models is evident in both our qualitative and experimental work. Specifically, our qualitative data document the importance of an attractive model in drawing attention to advertised cosmetics. Anca believes "the advantages are that she calls attention . . . everyone looks at a beautiful model," and Bianca is "fascinated" by beautiful models. Moreover, both data collections indicate that attractive models serve as signals about product quality and price, resulting in positive product evaluations for beauty-enhancing products. With regard to one's own attractiveness, we demonstrate that beautiful models can result in lower satisfaction (Martin and Kennedy 1993; Richins 1991). Thus, these post-ad exposure findings in Romania are consistent with prior research conducted in developed markets.

How Product Trial Alters the Effects of Model Attractiveness

Our research makes an important contribution by looking beyond the immediate reactions to advertisements to examine how product trial alters the attractiveness effects on product evaluations and self-judgments. Our experimental findings indicate that after a diagnostic product trial, there was no difference across model conditions in participants' product evaluations. It is interesting to note that participants exposed to the attractive model lowered their product evaluations after trial, whereas participants in the average-model condition increased their evaluations after trial. We suggest that although attractive models created high post- ad exposure expectations, these product quality expectations (and hence price judgments) were not met during trial, leading to lower post-trial evaluations. Because trial is maximally trustworthy for experiential attributes, the product information from a less credible source (e.g., advertising), cannot significantly compensate for product assessment based on trial. In contrast, in the average-looking model condition, product trial engendered more improved product evaluations (relative to post- ad exposure); hence, one could argue that the average-looking model does not do justice to promotion of the lipstick. Furthermore, it is interesting that despite lowered product evaluations for participants who had viewed the ad with the attractive model, they, along with participants in the other conditions, reported higher purchase intentions post-trial than post- ad exposure. These findings suggest that trial, because of its highly diagnostic and tangible information, creates stronger beliefs about the product than does advertising, and if positive, can result in brand commitment (Smith and Swinyard 1982).

Our research shows that trial affects self-judgments in two disparate ways. First, although attractive models spawn initial dissatisfaction with one's appearance, these effects dissipate over time. Second, our in-depth interview informants note that comparison to beautiful models can serve as information and inspiration (Lockwood and Kunda 1997; Martin and Gentry 1997), and our experimental findings indicate that trial preceded by ads with beautiful models provides the opportunity to envision the possibility of transformation, provoking hopeful anticipation (Peiss 1998). Furthermore, not evident from prior research, but consistent with McCracken's meaning transfer model (1989), our results suggest that the attractive qualities of the model have transferred to our participants, making them feel more attractive and confident while using the product.

Collectively, our results document the necessity for advertising, coupled with product trial, to fully understand and appreciate the positive effects of attractive models endorsing beauty-enhancing products. Scott posits that the indictment of advertising for beauty-enhancing products is "alarmingly casual with statistics and offers no substantiation - other than the ads shown - for any of the sweeping claims made" (2005, p. 316). Our results argue that these indictments ignore how women integrate advertising with their use of beautyenhancing products (Peiss 1 998; Thompson and Haytko 1 997 ; Thompson and Hirschman 1995).

IMPLICATIONS AND DIRECTIONS FOR FUTURE RESEARCH

Marketers are interested in ads that generate attention, create positive attitudes among consumers, and increase product or brand sales. Attractive models contribute to the branding of the product, and offer the best pretrial opportunity to persuade women that the advertised product has beauty-enhancing capabilities. Moreover, our research indicates that the initial negative effect of model attractiveness on satisfaction with one's own attractiveness is temporary - women report feeling more attractive and more self-confident when using the product. Thus, given that the cosmetics industry is about advertising an image that promises "hope in a jar" and justifies premium prices, it behooves advertisers to use attractive models in ads, particularly given that diagnostic ads have a greater impact, with no long-lasting negative implications.

Previous research indicates that advertising is less likely to influence post-trial product evaluations when the information provided by advertising and trial, as in our case, is positive (Smith 1993). When advertising is followed by a negative trial, however, favorable pretrial information can lead to a more positive evaluation of the unsatisfactory experience than trial alone (Smith 1993). Future research efforts might investigate the effect of idealized images of advertising on post-trial product evaluations when the product performs poorly, as well as examine the relationship between self-judgments and product evaluations more systematically (Bower 2001). Research might also examine how manipulation of comparison motives affects self-judgments and product evaluations in the context of attractiveness in advertising and product trial. A comparison with attractive models in advertising motivated by self-evaluation may temporarily lower women's self-judgments, whereas selfimprovement motivation may increase self-judgments. Martin and Gentry (1997) suggest that the extent to which women believe that they can improve their appearance may prompt a self-improvement motivation, but actual product usage is needed to make women believe that improvement is possible. Furthermore, Bower (2001) suggests that self-improvement motivation is more likely to be prompted by exposure to a facial picture rather than a full-body picture of a highly attractive model because of perceptions that it is easier to improve facial attractiveness than body shape.

Furthermore, research can investigate the effects of attractiveness in advertising and trial on women's evaluations related to social comparison feedback. Weary, Elbin, and Hill (1987) document that feedback indicating similarity with the comparison target affects people's feelings about themselves more than feedback indicating dissimilarity. Thus, future work might consider examining how product use affects people's belief about similarity or other characteristics such as friendliness or attractiveness with the comparison target, as well as how this feedback affects their feelings, and subsequent product evaluations and self-judgments.

Recently, Burgess and Steenkamp argued that "it is paramount for the future of marketing science and practice that we conduct more research in so-called emerging markets" (2006, p. 338). They also argued that "if at least weak generalizability holds for both high income countries and emerging markets, we can start to have confidence in the universality of the theory" (p. 349). Our findings demonstrate that attractive endorsers of beauty-enhancing products initially generate positive product evaluations and negative satisfaction with one's own attractiveness among our Romanian participants, replicating work in Western markets. We speculate that our observed trial effects (i.e., diagnostic trial experience allowing for calibration of product evaluations and enhancement of self-perceived attractiveness and self-confidence) would generalize to highly industrialized countries. Future research will need to examine contextual effects related to post-trial responses, however.

Our research venue, Romania, is a transitional country whose population has very little discretionary income - basic living necessities account for three-fourths of household consumption (Euromonitor International 2004). Despite these troubling times, the "new" media and availability of many new cosmetics fuels consumers' desires (Coulter, Price, and Feick 2003; Gal and Kligman 2000), and our qualitative study informants are attentive to changing appearance ideologies. Mihaela elaborates: "now women emphasize more their physical appearance. . . . There are lots more cosmetics . . . women use [cosmetics] more . . . you see that people around you pay attention to appearance and clothes; you feel strange if you don't change your old habits." The reality of purchase choices is made vividly clear by Raluca: "Without an appropriate financial situation, you can't arrange yourself. ... I am a simple worker in a store, and even if I like [some cosmetic product] and I might look good wearing it, I cannot afford it. . . . You can't afford to buy something for yourself, because you need to buy food first." The emerging markets offer multiple opportunities for furthering our understanding of consumers' product and brand-level choices, and the impact of advertising and product trial on those choices.

Pham, Michel Tuan, and Tamar Avnet (2004), "Ideals and Oughts and the Reliance on Affect Versus Substance in Persuasion," Journal of Consumer Research, 30(4), 503-518.

PricewaterhouseCoopers (2004), "2004/2005 Global Retail & Consumer Study from Beijing to Budapest," available at www.pwc.com/gx/eng/about/ind/retail/growth/romania.pdf (accessed March 27, 2009).

Ragas Cohen, Meg, and Karen Kozlowski (1998), Read My Lips, San Francisco: Chronicle Books.

Rao, Akshay R., and Kent B. Monroe (1989), "The Effect of Price, Brand Name, and Store Name on Buyers' Perceptions of Product Quality: An Integrative Review," Journal of Marketing Research, 26 (August), 351-357.

Richins, L. Marsha (1991), "Social Comparison and Idealized Images of Advertising," Journal of Consumer Research, 18 (June), 71-83.

Scott, Linda M. (2005), Fresh Lipstick: Redressing Fashion and Feminism, New York: Palgrave MacMillan.

Smith, Robert E. (1993), "Integrating Information from Advertising and Trial: Processes and Effects on Consumer Response to Product Information," Journal of Marketing Research, 30 (May), 204-219.

_____, and William R. Swinyard (1982), "Information Response Models: An Integrated Approach," Journal of Marketing, 46 (Winter), 81-92.

Spiggle, Susan (1994), "Analysis and Interpretation of Qualitative Data in Consumer Research," Journal of Consumer Research, 21 (December), 491-503

Stapel, Diederik ?., and Jerry Suis (2004), "Method Matters: Effects of Explicit Versus Implicit Social Comparisons on Activation, Behavior, and Self- Views," Journal of Personality and Social Psychology, 87 (6), 860-875.

Tabachnick, G. Barbara, and Linda S. Fidell (1989), Using Multivariate Statistics, New York: Harper & Row.

Thompson, Craig J., and Diana L. Haytko (1997), "Speaking of Fashion: Consumers' Uses of Fashion Discourses and the Appropriation of Countervailing Cultural Meanings," Journal of Consumer Research, 24 (June), 15-42.

_____, and Elizabeth C. Hirschman (1995), "Understanding the Socialized Body: A Poststructuralist Analysis of Consumers' Self-Conceptions, Body Images, and Self-Care Practices," Journal of Consumer Research, 22 (September), 139-153.

Tsao, Hsiu-Yuan, Leyland F. Pitt, and Albert Caruana (2005), "The Formation of a 'High Price- High Quality' Inferential Belief: A Study Among Young Buyers of Mobile Phones and Notebooks," Asia Pacific Journal of Marketing and Logistics, 17 (1), 50-60.

Vakratsas, Demetrios, and Tim Ambler (1999), "How Advertising Works: What Do We Really Know?" Journal of Marketing, 63 (1), 26-43.

Weary, Gifford, Susan Elbin, and Martha G. Hill (1987), "Attributional and Social Comparison Processes in Depression," Journal of Personality and Social Psychology, 52 (3), 605-610.

Wood, Joanne V (1989), "Theory and Research Concerning Social Comparisons of Personal Attributes," Psychological Bulletin, 106(2), 231-248.