The effect of thin versus round advertising models on female
self-evaluations, product evaluations and food preferences
The effect of thin versus round advertising models on female
self-evaluations, product evaluations and food preferences
Master Thesis
Author: Lidija Damjanovic Student Number: 1836080 Date: November 2009
Faculty: Faculty of Economics and Business
Specialization: Business Administration, MSc Marketing Management Supervisor 1: Dr. Debra Trampe
Supervisor 2: Dr. Liane Voerman
Author’s Address: Caeciliastraat 41a, 2312 XB Leiden Phone: + 31 681504102
Management summary
The main purpose of this paper was to examine the effect that thin vs. round models engaged in advertising have on female self-evaluations and evaluations of the product related to the advertisement. The main objective was to prove that, depending on the different conditions, thin and round models in ads can cause different self- and product evaluations. Also, as the check-up measure for product evaluations, willingness to pay among participants was assessed. Another goal of this study was to examine whether thin vs. round models from advertisements can have an impact on shaping food preferences among women.
In order to test the mentioned links, this study applied 2x3 between participants design. Data was collected via distribution of the questionnaire in the main shopping street in the city of Leiden, The Netherlands. During one week of the field experiment 125 Dutch women, aged 16 to 26 were questioned.
This study has replicated findings of Kallgren et al. (2000) that the general concept of self-awareness increases after seeing oneself in the mirror, compared to the condition when mirror is not present. Moreover, one of the main findings reported in this research is that if women see themselves in a (full-body) mirror which is positioned in a public place (in this study outside of the dressing room), public self-awareness increases. This is, to my knowing, a new finding and can be considered as a contribution to the existing knowledge about self-awareness.
With regard to the test about self-evaluations, the results are contradicting to accepted findings from previous researches (Smeesters and Mandel, 2006; Trampe et al. 2007). The participants in this study did not show different levels of self-evaluations after seeing one self in a mirror, or after being exposed to different body-sizes of models in the ads. For possible explanations for these contradicting finding please refer to the limitations section of this paper.
In testing the effect of advertising on product evaluations, product relevance was proven not to be a significant predictor. However, in assessing female’s willingness to pay product relevance was confirmed as one of the main predictors. Even though exposure to the thin model did not affect women’s willingness to pay, exposure to the round model did cause different levels of willingness to pay, depending on the product relevance. Specifically, women who exert product relevance are ready to pay more for AVAN after being exposed to the round model. Women with no product relevance are willing to pay less for the product after being exposed to the round model.
Preface
“Life is full of choices. I just hope, at some point of life, I will have the courage to step out of the safe zone, the cozy little place that we all have that prevents us from taking opportunities and experiencing different things in life. After all, dream is a creative vision for life in the future. Quite possibly, it is all about the unfamiliar and the unknown.”
I have always believed that once I will take that big step. My time spent in Netherlands is definitely one of those new, different things which created a long path of opportunities for me. Before I started with Master Studies in Groningen I had never been aware of the power of advertising. What was on TV – that was “in”. However, in just a year my view on advertising world has changed. Writing this thesis has led me to my future interests, and moreover has helped me realize that all the way long I have been the subject of my own research.
Hereby, I would like express my sincere gratitude to Dr. Debra Trampe for her endless support and immense patience. Without her guidance the road from advertising to consumer minds would seem like the “rocky mountain”. I would like to thank Dr. Liane Voerman, my second supervisor, for her flexibility and valuable comments at the final stage of my thesis. My appreciation also goes to my friends Pepi, Adela and Alan for sharing precious “international” moments with me.
I devote this thesis to my parents and my brother. To my parents: who have always believed in me. I thank them for their ceaseless love and support, and I am grateful to them for every choice I made. To my brother: who is my leader, inspiration, and my best friend to whom I have always looked up.
Finally, my gratefulness goes to Pim, for being my encouragement for the past year. I thank him for his unselfish, enduring love. After all:
“The good life is inspired by love and guided by knowledge”- Bertrand Russel.
Contents
1.Introduction ...6
1.1. Background ...6
1.2. Problem statement and research questions ...7
1.3. Structure of the report...8
2. Theoretical framework ...9
2.1. Introduction...9
2.2. Social comparison ...10
2.3. Public and private self awareness ...12
2.4. Relation between private and public self awareness and mirrors ...13
2.5. Body dissatisfaction and self-evaluations...14
2.6. Product evaluations ...16
2.7. Willingness to pay...18
2.8. Food preferences ...19
3. Methodology...21
3.1. Procedure and sample...21
3.2. Questionnaire design ...22
3.2.1. Self-evaluation items ...23
3.2.2. Product evaluation items...24
3.2.3. Public and private self awareness items ...24
3.2.4. Willingness to pay ...25 3.2.5. Food preferences ...25 4. Results ...26 4.1. Introduction...26 4.2. Descriptive analysis...26 4.2.1. Self-evaluation scale...27
4.2.2. Product evaluation scale ...27
4.2.3. Public and private self-awareness scale...28
4.2.4. Other ...28
4.3. Test of hypotheses...30
4.3.1. Hypotheses H1a, H1b and H1c ...30
4.3.2. Hypotheses H2a and H2b...32
4.3.3. Hypotheses H3a, H3b and H3c ...33
4.3.4. Hypotheses H4a, H4b and H4c ...33
4.3.5. Hypotheses H5a and H5b...35
4.3.6. Summary of the results ...37
5. General discussions...39
5.1. Introduction...39
5.2. Self-awareness and the mirror ...39
1.Introduction
1.1. Background
Picture 1. Fillipa Hamilton for RL On 16th of October, Dutch newspaper NRC-next has
reported in its international row a fashion story which has shocked the public. Paul Steenhuis, the columnist, has officially exposed the extreme demands that the fashion industry puts on their models.
A 23-year old photo model, that has been photoshoped into a caricature extremely skinny model by the American fashion house Ralph Lauren in an official campaign, has lost her job. “I was fired because I am too fat.”, said Filippa Hamilton in the Today Show of American TV-channel NBC. By doing so she has become the centre of attention in a growing international row about the extreme demands the fashion-industry puts on its models. Hamilton, who is 1.78 meters long, weighs 54 kilo’s and claims to be an American size 4 (European size 34), told that she was fired by Ralph Lauren in April.
She had been working as a model for the company from the age of 15, when she was discovered on the streets of Paris. “My agent was told that the reason for my discharge was that I was too fat and did no longer fit in the clothes they wanted to advertise.”
Ralph Lauren denies. In a statement the company declares that their relationship with Hamilton ended ‘as a result of her inability to meet the obligations under her contract’. “I have always been loyal and on time. I have done everything I could. I considered the company as my second family”, reacted the Swedish-French model Wednesday on TV. “They mean I was too fat for their clothing.”
Initially Hamilton did not want to give any publicity to her dismissal, but went public because she became the centre of controversy on the internet. She did so because she thinks it is ‘unhealthy’ that photo models in the fashion-industry should be so thin. The photo's which depicted her with hips about as narrow as her head, could make young women “think that it is normal to look like that - and it's not. I saw my face on this super-extremely skinny girl, which is not me; it's not healthy, it's not right”, she said.
Through the fashion industries and their related advertising consumer culture portrays individuals with images that contain “lifestyle and identity instructions that convey unadulterated marketplace ideologies (i.e. look like this, aspire to this kind of lifestyle)” (Arnould and Thompson, 2005). Appearance, beauty, and the perfect body have been central to women’s identity for a very long time, and it is very hard to remain untouched with normative influences portrayed in mass media as being “normal”, and desirable. Even during the history research has focus on a beauty as a central aspect of one’s life, stating that beautiful attracts all of the world’s material benefits and happiness (Dion et al., 1972). “There are specific ways in which consumer culture has a detrimental impact on many individuals. Without a doubt, the expansion of advertising is particularly significant in this aspect. The argument to be made here is that consumer culture perpetuates a number of myths, particularly through advertising, with the idealized images of the “body perfect” as a central aspect. The impact of this myth on children’s, adolescents’ and adults’ identity and well being is negative. Therefore, instead of encouraging people to be liberated, have bigger choices and greater self-expressions through consumption, advertising is entrapping people who are being unaware of its persuasive influence. While people believe that they are experiencing themselves and attaining happiness, they are in fact, developing, monitoring, and molding their identities with respect to unrealistic ideals promoted by consumer culture through advertising”, (Ditmar, 2007).
1.2. Problem statement and research questions
A thin body is nowadays considered as an ideal which every woman should admire and aspire to achieve (Smeesters and Mandel, 2006). According to this global trend, a lot of past research has focused on the detrimental influence of advertising on body dissatisfaction and decrease of self esteem. Richins (1991) has demonstrated in her research that women, after being exposed to advertisements featuring thin, attractive models experienced less satisfaction with their own physical appearance. However, a number of other researchers have provided the evidence that exposure to thin models in advertisements can lead to self-enhancement and more positive self-ratings, than exposure to round, next door models (Smeesters and Mandel, 2006).
Thus, the main problem within previous findings and the starting point for this research is the uncertainty under which conditions exposure to idealized images of women (thin models) versus every-day full figured women (round models) in ads exerts positive or negative effects.
Concerned with the pressure of advertising on fortifying thin body ideals this research also focuses on the food consumption. Setting the thin body standards can be seen as one of the main cause of the unhealthy food consumption. Eating disorders are currently one of the major public health issues, especially in industrialized countries (McFerran et al., 2010). Given this public concern and increase in advertising, it is surprising that little literature has devoted attention to women’s eating patterns. What has been lacking in the literature to date is an examination of how the food choices that women make are influenced by the body types (thin vs. round) of models featured in advertising. Therefore, the side research question this paper is:
Research question 2: Influence of advertising models (thin vs. round) on forming food
preferences among women.
1.3. Structure of the report
2. Theoretical framework
2.1. Introduction
In order to gain deep understanding of the effect that thin vs. round models engaged in advertising can exert on women, this section gives an overview of the relevant literature. Understanding the process of how women engage in social comparison represents the basis of establishing a potential relation between advertising and self-evaluations. How women react after being exposed to different body types, and in which way they process information is important for understanding how self-evaluations is formed. Therefore to begin with, the concept of social comparison and the direction of it’s arise are introduced.
2.2. Social comparison
Beauty might lie in the eye of the beholder, but research has provided consistent evidence that perceptions of physical attractiveness are subject to a perceptual contrast effect (Thornton and Maurice, 1999). If viewed in combination with highly attractive others, a person of average attractiveness may be judged less attractive than if that person had been evaluated without a comparison group. It has been commonly accepted among various authors that with exposure to others who are more physically attractive, self ratings of attractiveness generally become more negative. Also, it is interesting to note that with exposure to less attractive models, opposite may occur which can result in more positive evaluations of one’s attractiveness. Besides the self-perceptions of attractiveness, the contrast effect for social comparison is proved to influence also other aspects of the social self. For instance, in addition to diminished perceptions of attractiveness, the common occurrences are decreased self-esteem, increased public and private self-awareness, and heightened social anxiety.
Recent research suggests that depending on culture and context, people may represent themselves in relationship with others (social self-construals) or as individuals (personal self-construals). Wiekens and Stapel (2008) suggest that personal self-construals are more motivated to be independent/different and to be alone, and less motivated to be accepted/to conform and to be together. Conversely social self-construals are more motivated to be accepted/to conform and to be together, and less motivated to be independent/different and alone. Specifically, contrast occurs when the personal self is activated, but this effect is especially strong when its impact on self-evaluation is positive (i.e., downward comparison information). Similarly, assimilation occurs when the social self is activated, but this effect is especially strong when its impact on self-evaluation is positive (i.e., upward comparison information).
According to Ditmar and Howard (2004) social comparison information may affect self evaluations in circumstances associated with uncertainty. Uncertainty can be defined as the lack of self-clarity. Campbell et al (1996) define self-clarity as the extent to which the contents of an individual’s self concept (e.g. personal attributes) are clearly and confidently defined, internally consistent and temporally stable. However, in their research Stapel and Tesser (2001) have proved that self-activation is sufficient to promote interest in social comparison, even in the absence of uncertainty.
2.3. Public and private self awareness
Self-awareness is a central concept in several divergent approaches to behavior and life (Fenigstein, Scheier and Bus, 1975). Self-examination helps a person to examine his/her thoughts, motives and defenses. The consistent tendency of persons to distract attention inward and outward is the main trait of self-awareness. Over the years, self-awareness is proved to have two separate concepts: one deals with the cognitive, private mulling over the self, and the other emphasizes an awareness and concern over the self as a social stimulus (Fenigstein, Scheier and Bus, 1975).
According to Buss (1980) public self-awareness refers to characteristic concern with one’s self as a social object of other’s attention and is associated with greater attention and responsiveness to standards or expectations by which one’s behavior and personal attributes may be evaluated by others. According to Thornton and Maurice (1999) public self-awareness reflects a transient, induced state of self-focused attention brought about by situational stimuli (e.g. full length mirror, camera and similar), which results in heightened awareness of the self as an object subject to the focus of other’s attention. This contributes to an evaluation of the self which typically reflects “standards” for appropriate or desirable attributes or behavior which are relevant to the situation. Standards are usually presented in the form of ideal. When societal standards for appearance are made salient, heightened awareness as an object of others’ attention (situationally-induced state) contributes to the experience of a negative contrast effect.
On the other hand private self-awareness reflects a disposition to attend to one’s personal thoughts, feelings and motives. The introvert is generally oriented towards the internal world of ideas and concepts. In this way, women are turned more towards themselves, and avoid evaluation of others. Moreover these women are less concerned with other people’s opinions. This eventually leads to the behavior that reflects one’s own believes.
2.4. Relation between private and public self awareness and mirrors
Self-awareness as a general concept is of immense importance for establishing the relationship between advertising and female self-evaluations. According to the past research self-evaluative comparison effects are more likely to occur when self-related cognitions are made cognitively accessible (Stapel and Koomen, 2001). As mentioned in the previous sections, in order to activate tendency to compare, relevant standards for women need to be made salient. In the present research, the relevant standards are defined as a societal ideal regarding physical appearance and are expressed in the form of the ad with an attractive, thin female model. Also, this research focuses on the opposite state which is expressed in the form of the ad with round, next-door female model. Both induced state of self-awareness and low self-awareness were included and different effects on self-evaluations were expected. Namely seeing oneself in the mirror can indicate induced state of self-awareness. Specifically, seeing one’s self in the mirror, as an induced state of self-awareness, and being exposed to the ad with the attractive model is expected to result in perceived discrepancy between the self and an ideal state. This discrepancy is likely to intensify the tendency to compare according to the condition when such state is not present.
On the other hand, women with low self-awareness, who do not focus on themselves as social objects, are less appearance schematic, have not created any appearance or behavioral “standards” and therefore do not have the same need as their counterparts to evaluate themselves. Therefore, low self-awareness women will have lower tendency to compare. Past research has focused solely on the presence of the mirror, but not on the place where mirror is positioned. This research aims at investigating whether the position of the mirror can be considered as having effect on the direction in which self-awareness will become salient, meaning whether public or private-self awareness will arise. The direction in which self-awareness develops will serve as a springboard for further analysis.
According to all previously mentioned, the following hypothesis can be made:
H1a: When seeing one self in the mirror inside the dressing room, private self-awareness
increases.
H1b: When seeing one self in the mirror outside of the dressing room, public self-awareness
increases.
H1c: If one does not see her self in the mirror, self-awareness (both public and private) will
2.5. Body dissatisfaction and self-evaluations
The cultural ideal of female beauty portrayed in media is of particular psychological significance because appearance is central to female self-evaluation and sometimes even determinant of women’s social success (Ditmar and Howard, 2004). Therefore it is not surprising that the gap between women’s actual body sizes and media’s unrealistically thin ideals has given rise to public concern about “unhealthy” the media messages making women feel dissatisfied with their bodies and putting pressure on them to lose weight. Yet, despite the fact that women are surrounded by this type of advertising, their levels of body dissatisfaction and negative self-evaluations vary.
During the history of the research there has been a resurgence of interest in examining whether aspects of the self-evaluation are stable or adaptable. According to many authors cited in the research of Heatherton and Polivy (1991) there is considerable evidence that the self-evaluation is a stable concept and that people will actively seek information that confirms their self-concept and reject information that threatens their general view of the self (Greenwald, 1980; Swann 1985; Wylie 1979). On the other hand it is proved that momentary changes in self-evaluation can occur due to the influence of situational factors. It appears that self-evaluation is a relatively enduring state from which there can be some degree of divergence. Given this, substantial research has been done with the aim of manipulating self-evaluation, and there is agreement that self-evaluations do vary across situations (Heatherton and Polivy, 1991). For instance, exposure to physically attractive targets, such as advertised models, can negatively alter self-perceptions by lowering individuals’ satisfaction with their own bodies (Marsha L. Richins, 1991). One of the purposes of this study is to examine the cause of such fluctuations in self-evaluations. Whether viewing a highly attractive model will have an effect on self-evaluations depends both on who that target is and by whom that target is perceived (Trampe et al., 2007). According to Trampe et al (2007) models are generally treated as less relevant comparison standards than non-models. Thus when confronted with a physically attractive target, self-evaluations are more negative when the target is presented as a non-model than when the target is presented as a professional model. In addition, Brown et al. (1992) have proven that women’s self-evaluation is negatively affected after being exposed to attractive women, but not after being exposed to attractive men. In this situation, women are presented as relevant comparison targets and men as less relevant comparison targets. This is the underlying reason for choosing women as reference targets in the present research.
how they would really like to be (ideal state) are brought salient. Therefore, for those women who have internalized consumer culture ideals, exposure to idealized models should make prominent actual-ideal identity deficits, because their bodies are almost invariably inferior to those presented in advertising.
Although a lot of theories about body-perfect internalization exist, none of them focuses on the way and the extent to which people bring ideal standards salient. According to Kallgren et al (2000) internal or external focus of people’s attention importantly moderates the degree to which the personal norm (internal standard) is likely to become salient. While seeing themselves in the mirror, individuals should experience increased self awareness. If so, it can be suggested that by focusing greater attention to one’s inner state and traits, thin-ideal internalization is more probable to become salient. When ideal standards are present, it can be expected that women will tend to compare themselves to the greater extent, and vulnerability to the body image will probably become bigger. Exposure to the thin model can finally cause greater body dissatisfaction and negative self-evaluations.
According to the previous argument the following hypotheses can be suggested:
H2a: Seeing one self in the mirror and being exposed to the thin model causes decreased
self-evaluations, compared to the state when mirror is not present.
On the other hand, exposure to the round model under the same conditions can cause more positive evaluations among women, and therefore, the following can be suggested:
H2b: Seeing one self in the mirror and being exposed to the round model causes increased
2.6. Product evaluations
It is a common wisdom that “beauty sells” meaning that if you stick a beautiful skinny girl on the cover of the magazine you will sell more copies. In general, certain features of the model, such as model’s beauty, directly spread to product evaluations. But according to the global concern of women’s “health issues” caused by engaging ultra thin models in advertising, public has becoming more interested in round and average size models. Compared to extremely thin models these models are believed to be less threatening and may therefore have less negative consequences for the perceiver.
Moreover, according to Häfner and Trampe (2009) thin models can, under some circumstances and for some individuals, actually harm product evaluations. Apparently this strategy also seems to pay-off in terms of selling the advertised products. In particular, whether ultra thin or round models will result in positive product evaluations depends on one of the two routes that the person follows in determining behaviors and evaluations. Consumers can actively seek and process information, or they can be more superficial in their analysis of the ads. According to Petty et al.’s (1983) Elaboration Likelihood Model (ELM) of attitude change different methods of inducing persuasion may work best depending on whether elaboration likelihood of the communication situation (i.e., the probability of message, or issue, relevant thought occurring) is high or low. Under high elaboration likelihood the central route to persuasion should be particularly effective, but when the elaboration likelihood is low, the peripheral route should act as a better one. Petty et al. (1983) explain that the ELM contends that as an issue or a product increases in personal relevance, it becomes more important and adaptive to forming a reasoned veridical opinion. Therefore, people who have high involvement, rather than low, are more motivated to devote cognitive effort required to evaluate true merits of an issue or a product. One of the important outcomes of this research, on which I will reflect later, is that the nature of the product endorser had a significant impact on product attitudes only under low involvement, but not under high involvement.
According to Häfner and Trampe (2009) in impulsive systems, where people lack in motivation and/or capacity to devote their attention to the ad, behavior and evaluations are assumed to directly follow from perceptual input and subsequent spreading activation of the perceptual input. On the contrary, the reflective system is conceptualized as more indirect pathway since it is based on reasoning and correction. Specifically, explicit product evaluations (following reflective route) are more positive when round, next-door models rather than ultra-thin models are used to advertise the product. Here, people are willing to devote their cognitive effort to analyzing the ad and product evaluations are based on the liking of the campaign. Conversely, implicit evaluations are affected in the opposite direction: thinner, more beautiful models yield more positive product evaluations than round, less beautiful models. What is more, Häfner and Trampe (2009) confirm that equal attractiveness is necessary for equal product evaluations.
In this study, the relevance of the advertised product to participants is measured by the last store they have visited. Therefore, if women have visited some type of beauty store the concept of beauty and standard of thinness are made salient. In this way it can be argued that women will follow the implicit route and evaluate the ad in the peripheral way. They will be motivated to devote less attention to the ad and presumably positive product evaluations will result from the exposure to the ad with thin model. Conversely, exposing women to the ad with round, next-door model in the same situation of product relevance, can result in negative product evaluations.
On the other hand, if women have not visited any beauty store, and/or have visited store such as food and groceries, electronics and similar, beauty and thin-body standard is not made salient. If so, women will be under lower pressure of the ideal body. In this way women will assess the product without tendency to compare one’s self to the target model in the ad and accordingly product evaluation will be the result of general liking of the ad. Therefore, it can be expected that these women will be more positively influenced by the round model engaged in the ad, than by the thin model, resulting in higher product evaluations.
In summary the following hypotheses can be formulated:
H3a: When the product is considered to be relevant for women, they will reflect increased
product evaluation when exposed to the thin model, than when exposed to the round model.
H3b: When the product is considered to be relevant for women, they will reflect decreased
product evaluation when exposed to the round model, compared to when exposed to the thin model.
H3c: When the product is considered to be non relevant for women, they will reflect
2.7. Willingness to pay
Even though past research had its focus on product evaluations, little previous research has focused on the behavioral aspects of product evaluations. It is a common wisdom that people’s actions can subsequently differ from their words. People might express their attitudes and believes in one way, but their resulting behavior can follow a distinct route. If so, there is a great possibility that even though participants express higher product evaluations, their willingness to pay might be lower, and vice-versa.
Given this controversial effect, willingness to pay, as additional variable, was included in the present research. As behavioral measure willingness to pay is defined as the maximum amount of money that may be contributed by an individual to equalize a utility change. The main principle is that the maximum amount of money an individual is willing to pay for a commodity represents an indicator of the value to him/her of that commodity. Willingness to pay is positively associated with education, income and beliefs (www.pitt.edu).
In this study, willingness to pay is assumed to be positively correlated to the product evaluations. When people reflect positive product evaluations, it can be suggested that they will be willing to pay more for the product. Vice versa, women who have judged the product negatively or to some extent less positively are expected to have lower willingness to pay. According to these expectations, the following hypotheses can be suggested:
H4a: When the product is considered to be relevant for women, they will reflect increased
willingness to pay when exposed to the thin model, than when exposed to the round model.
H4b: When the product is considered to be relevant for women, they will reflect decreased
willingness to pay when exposed to the round model, than when exposed to the thin model.
H4c: When the product is considered to be non relevant for women, they will reflect
2.8. Food preferences
Due to the advertising body-thinness pressure, most cultures currently place a high value on thinness, and those who are overweight are often victims of stereotyping or stigmatization (Shapiro et al., 2007). Obesity is something that most people tend to avoid and the research (Crandall, 1994) has shown that we avoid behaviors associated with undesirable groups. The majority of the research has suggested that being fat is associated with a wide variety of negative characteristics. For instance, according to Crandall (1994) fat people are seen as aesthetically displeasing, alienated from their sexuality, and unattractive.
Consumer research has begun to show that the effects of social “others” are moderated by whether the person is a member of an aspirational or dissociative group. According to Hoyer and MacInnis (2007) aspirational reference groups are groups we admire and wish to be like, but we are currently not a member of. Dissociative reference groups are groups whose attitudes, values and behaviors we disapprove and do not wish to emulate.
Given the fact that people eat many times in social settings, it is surprising that little research has examined how our food preferences are shaped by world around us. Many of past research have their main focus solely on the effect that social and interpersonal influences exert on consumption choices (e.g. Polivy et al., 1979; Herman et al., 2003). One recent research that has extended the social influence on consumption choices is conducted by McFerran et al. (2010). They propose that food choice is strongly subject to interpersonal influences, with people choosing larger or smaller portions after seeing another consumer doing likewise, but that the size of the portions is moderated by the body type of the confederate. Their results point out that when the reference person (confederate) is observed taking a large quantity of food, participants choose and ate less when that confederate was heavy than when she was thin.
While the research outlined above was conducted in a social setting, and had its focus on consumers’ reaction to how much others eat or how the body type of the reference person influences quantity of food intake, little work has examined how people, specially women who are known to be over-concerned with their bodies, shape their food preferences after being exposed to advertising.
All of the previously mentioned suggests that we do not perceive round people in the same way as we do thin, underweight individuals, and thus we will not reflect the same food preferences after being exposed to each body type. Given that dissatisfaction with one’s appearance increases proneness to social comparison, I hypothesize that women who score low on self-evaluations will adjust their food preferences according to the body type of the model in the ad. However, among those women confident in their appearance, less social comparison should occur, resulting in less schematic choice of food.
Since the round model represents a dissociative reference group and research has showen that we avoid belonging to these types of groups, it can be expected that body dissatisfied women will chose a snack which is perceived as a healthier, in this case a fruit biscuit. What is more, as the thin model is perceived as assimilation reference group, and past results point out that we aspire to belong to these types of groups, it can be expected that body dissatisfied women will also have higher preference for the healthier snack, in this case biscuit.
On the other hand, it is to expect that body satisfied women will not perceive the model as belonging to either aspirational, or dissociative reference group, as these women do not aspire to have different bodies. In this instance it can be suggested that the body satisfied women won’t shape their preference according to the model that they have been exposed to. Rather, their choice will act as a random selection.
Consequently, it can be postulated that:
H5a: Women who exert lower self-evaluation after being exposed to either round or thin
model will form preferences towards healthier food, in this research presented as a fruit biscuit.
H5b: Women who exert higher self-evaluation after being exposed to either round or thin
model, will not form any preferences towards food, but their snack choice will be the result of random selection.
3. Methodology
3.1. Procedure and sample
To test the hypotheses an experiment was conducted. The questionnaire was used as the main instrument for collecting data. The research was executed in the main shopping street in Leiden, the Netherlands. Participants belonged to one of the six conditions of a 3 (mirror inside vs. mirror outside vs. no mirror) by 2 (thin vs. round model in the ad) design, from which the body size of a model was a manipulated, and position of the mirror was measured factor.
The field study included a sample of 125 women. 3 respondents’ answers were excluded from further analysis due to their incompleteness. All participants were Dutch, and they have all filled in a Dutch version of the questionnaire (please see Appendix 1 for the questionnaire). The underlying reason is that when people are asked questions in their native language they are able to reflect their thoughts immediately and more clearly. Therefore, their answers will not be intimidated with the language barriers.
Given that females are more sensitive to social comparison regarding body type (Trampe et al., 2007), the inquiry of this research was restricted to women only. What is more, as women get older they tend to show small decline in thin-ideal internalization, and also engage to some extent less in social comparison (Ditmar, 2004). For this reason women aged from 16 until 26 have been selected for this experiment, with exception for women aged from 27 to 29 who make 1.6 % of the whole sample. The great majority of participants were in the age group 16 to 18 (35.2%), followed by the group of participants aged between 22 and 24 (28.7%). The analysis of the education variable reviled that the majority of the participants were students (85.2%), while the rest (14.8%) was working. Consistent with the biggest age group of the sample (from 16 to 18 years), the majority of students were attending High School and MBO-level education, while the second largest group were Master students.
snacks. All questionnaires were completed in less than 5 minutes, and participants were successfully debriefed at the end.
Manipulations
Following the research of Häfner and Trampe (2009) deodorant was chosen as the target product, and the same ad was used in both thin and round conditions. To manipulate the body type of the model, he computer program Photoshop was used and the body of the model was noticeably enlarged, making her appearance “fat”. The ad was featuring a nonexistent product named “AVAN”. Purpose of using a nonexistent product is to prevent participants’ answers from being manipulated by preexisting brand attitudes. Both advertisements had the same simple green background and depicted the deodorant featuring the logo in the left part of the ad. The model (either thin or round) wearing lingerie was placed on the right side of the ad. What’s more the slogan “a source of freshness (een bron van frisheid)” was positioned in the left upper part of the ad. No further product information was featured in the ad. In addition, the model was the same ethnicity as most of the participants (please see Appendix 2 for the ads used in the research).
3.2. Questionnaire design
The women were asked to complete two set of questions containing the instruments relevant to the present study.
The first set of questions was designed to investigate whether the last store that the women had visited was a clothing store, and if they tried on any clothes. The underlying purpose of these questions was the need to separate those women who had visited a clothing store, but did not try on anything, from those who had visited a shop less related to clothes and appearance, such as home decoration store or an electronic store. The fundamental idea in examining these data is that the “beauty concept” is made salient when visiting stores such as clothing, cosmetics, jewelry etc., which influences self-awareness, as well as the way women evaluate product from the ad. This set of questions is designed to examine the validity of hypotheses H3 to H4.
In addition, a background set of questions was aimed at examining the position of the mirror in the dressing room. In order to establish the potential relationship between the presence and/or the position of the mirror, and dependent variables in this study, participants were asked to indicate whether they saw them selves in the mirror, and if so, what the exact position of the mirror was. The use of this background information is of immense importance and is spread through-out the whole research.
Second set of questions was designed to examine four measures including a) evaluations, b) product evaluations, c) willingness to pay, and d) public and private self-awareness. For each measure, except for the willingness to pay which is considered as the behavioral measure, a 7-point Likert scale was applied with scores ranging from 1 = “not at all” to 7 = “extremely”
3.2.1. Self-evaluation items
The first step in the development of a self-evaluation measure was to obtain suitable questions which will reflect participant’s personal feelings with regards to their appearance. Analyzing appropriate questions which will effectively indicate self-evaluations relied heavily on the study of “Development and Validation of a Scale for Measuring Self-Esteem” (Heatherton and Polivy, 1991).
Heatherton and Polivy (1991) have mainly settled their State Self-Esteem Scale (SSES) development on the scale of Janis and Field (1959), further abbreviated as JFS. Wide spectrum of past research has proven JFS to be one of the best multidimensional scales of self-esteem. Originally, scale is a 23-item test developed for use in attitude change research, and contains items about self-regards, academic abilities and social confidence (Heatherton and Polivy, 1991). The items of JFS have been modified many times by changing the scoring format (e.g., 5- or 7- point Likert scale), or adding items with regards to the other dimensions of self-evaluation, such as performance, appearance and general self-esteem in SSES (Heatherton and Polivy, 1991). SSES was applied in three studies, and provided substantial evidence that the scale is psychometrically sound and has a high degree of construct validity, which was the main reason for choosing items from this scale.
For the present study, items reflecting appearance dimensions from SSES were of vast interest. Moreover, Heatherton and Polivy (1991) showed that the appearance factor of the SSES would probably be more sensitive to manipulations that make physical appearance salient. This finding stands for the additional support for applying their items, since the present research relies heavily on the extent of salience of the physical appearance made by the presence (absence) and position of the mirrors.
Two appearance items from SSES were included in the questionnaire and one item was added. The SSES items were modified in the following way:
Item from SSES New modified item
I feel satisfied with the way my body looks right now.
appropriate to women. Such features can include gentleness, patience, and kindness. As such, this trait is considered to be one of the main predictors of self-evaluation. Femininity should not be confused with feminism, which is the belief that women deserve political and economic rights equal to men.
All the items were answered on 7-point Likert scale (1 = not at all, and 7 = extremely). The validity of each question as indicator of self-evaluation is confirmed by the reliability test.
3.2.2. Product evaluation items
In order to examine product evaluations, participants were asked to answer the three following questions: “Do you evaluate AVAN positively?”, “Do you think AVAN is a product of a good quality?”, and “Does AVAN make you feel good?”. These items were all adopted from the same research as the advertisement (Häfner and Trampe, 2009). The three items were used as the basis of forming an overall index of participants’ product evaluations. The idea of using these three items is examining under which conditions (thin vs. round model, and relevant vs. non-relevant store) women will reflect increased, or decreased product evaluations. Questions were all assessed on Likert scale ranging from 1 (not at all) to 7 (extremely).
3.2.3. Public and private self awareness items
As mentioned previously in this study, self-awareness has two major components: private and public. In order to examine both concepts of self awareness, two relevant items were adopted from Fenigstein, Scheier, and Buss’ Self-Consciousness Scale (1975).
The private self-awareness item was concerned with one’s inner thoughts and feelings, and was assessed with the question “To what extent are you aware of how you view yourself?” The original item from Self-Consciousness Scale was formulized as “Generally, I’m not very aware of myself”, and was modified due to its negative formulation which can possibly cause misinterpretation of the item.
3.2.4. Willingness to pay
Since people’s words and actions can subsequently differ, indicated product evaluations do not necessary reflect the same behavior of people. Due to this fact, one question was designed as a confirmation of increased or decreased product evaluations. The question was stated as a pure behavioral measure (“If AVAN is available in the store, how much would you be willing to pay for it?”), and was expected to indicate the amount of money participants would pay to attain AVAN, consistent with previously reflected product evaluations.
In identifying a price range for the advertised deodorant, the website www.drogisterij.net
was used. Since AVAN is non-existing product, participants will reflect low brand awareness. Therefore, in defining the standard price range for AVAN, prices of non-famous deodorant brands “Fenjal” and “Amando” have been assessed as reference products. Prices for these brands are found to vary from €3.99 to €4.13, while the bottle size varied from 150ml to 200ml. Since many well-known brands, such as “Odorex” are priced lower than “Fenjal” and “Amando” (€2.59 for a 150ml bottle), it can be expected that participants can also indicate prices lower than €3.99 and consider it as a standard price for AVAN. As a result, the standard price range for AVAN has been identified as starting from €2.70 and reaching up to €4.50. Prices indicated lower than the bottom price (€2.70) can be considered as reflecting lower willingness to pay, while prices which exceeded the upper boundary (€4.50) can indicate higher willingness to pay.
3.2.5. Food preferences
4. Results
4.1. Introduction
This chapter is designed to analyze data collected during the phase of empirical research. Firstly, the description of the data is presented, and the reliability of the applied scales is tested. Subsequently, the hypotheses suggested in the second chapter are examined, and this section ends up with concluding remarks with regards to testing of hypotheses.
4.2. Descriptive analysis
Mean scores, as well as standard deviations for “self-evaluation”, “product evaluation”, “public and private self-awareness”, “willingness to pay”, and “food preferences” are shown in Table 1. According to these results, the average responses on public self-awareness and private self-awareness were relatively high, with a mean above “5.00”, while self-evaluation and product evaluations scored something lower (respectively M = 4.89, and M = 4.25). In line with relatively lower product evaluation, the average amount of money that participants were willing to pay for the deodorant was M = 4.64. As for the food preferences, results show that relatively more women chose the chocolate bar over the fruit biscuit (M = 1.24). All items were assessed on a 7 point-Likert scale, except for the willingness to pay which is expressed in the money value.
Table 1. Measures of central tendency for dependent variables
Variables Mean Std. Deviation
4.2.1. Self-evaluation scale
Self-evaluation scale is comprised of three items for which means and standard deviations are shown in the Table 2.
Table 2. Measures of central tendency for Self-evaluation scale items
Items Mean Std. Deviation
How attractive do you feel at the moment? How satisfied are you with your body? How feminine do you feel?
The overall scale
4.57 4.84 5.26 4.89 1.220 1.255 1.335 1.049
With respect to the means as the main measures of central tendency, no big variances are found between these three items, which suggest that all items are strongly correlated to appearance. Women have scored the highest on the feminine feel measure (M = 5.26, SD = 1.335), which is expected since the feminine feel refers to the behavior generally associated with women. The overall mean of the scale is found to be M = 4.89, which proves that women have indicated more positive self-evaluations than negative.
With respect to the percentage of women who have scored lower than 4 (neutral) on individual items, indicating lower self-evaluations, the results are as follows:
a) 18.03% of women have scored lower than 4 (neutral) on the attractiveness item b) 13.11% of women have scored lower than 4 (neutral) on the body satisfaction item c) 12.30% of women have scored lower than 4 (neutral) on the feminine feel item The three items were computed into the new variable named “self evaluation” which served as a springboard for further analysis. In order to asses the reliability of the self-evaluation scale, the average correlation coefficient was calculated. Characteristics of the scale were examined with respect to the individual items, the overall scale and the relationship between the individual items and the overall scale. The reliability of the self-evaluation scale was found to be appropriate (alpha = 0.767). An examination of the item-total statistics indicated that no removal of items would enhance this reliability measure, proving the scale to be properly set up.
The distribution of the scores of the scale was assessed with the frequency histogram and is found to be normal. No missing values were reported.
Table 3. Means and Standard Deviations for Product evaluation scale items
Items Mean Std. Deviation
Do you evaluate AVAN positively?
Do you think AVAN is a product of good quality? Does AVAN make you feel good?
The overall scale
4.34 4.21 4.21 4.25 1.352 1.344 1.344 1.213
In order to assess the normality and reliability of the scale, a new variable has been computed and named “product evaluation”. The reliability test showed immense validity of the product evaluation scale, reporting a Cronbach’s alpha of 0.884.
Applying the graphical analysis, the distribution of the results is found to be normal.
4.2.3. Public and private self-awareness scale
Due to the time limitation of conducting the questionnaire in the shopping street, public and private self-awareness were assessed with one item-scale. With regards to the public-self awareness item, “To what extent are you aware of how other people view you?”, central tendency and variable measures were reported to be M = 5.03, and SD = 1.323 respectively, reflecting a slight salience of public self-awareness among women.
Mean and standard deviations for the private self-awareness item, “To what extent are you aware of how you view yourself?”, were found to be slightly higher than for its counterpart (M = 5.57, and SD = 1.192), indicating moderately increased private self-awareness.
Since both scales were designed as one-item scales, there was no need for reliability check.
4.2.4. Other
Position of the mirror
Table 4. Percentage distribution of participants per mirror condition
Condition (%)
Have not tried on clothes
50.8
Have tried on clothes
Body length mirror inside the fitting room
Face height mirror inside the fitting room and stayed in
Face height mirror inside and went out to look in a full body length mirror No mirror inside the fitting room and stayed in
No mirror inside and went out to out to look in a full body length mirror
33.6 8.0 4.9 0.0 9.8 Willingness to pay
Assessing scores on the willingness to pay item revealed a big deviation in amount of money that participants were ready to allocate on the advertised product. The minimum amount of money is found to be €0.00, while the maximum amount reached €20.00. Mean appointed that the average price respondents were willing to pay was M = 4.64, followed by SD = 3.54.
Out of the 122 respondents, 34.43% have indicated that they would pay the price lower than the defined standard price range (from €2.70 to €4.50), while 39.34% of respondents designated a higher price. The rest of respondents indicated willingness to pay within the standard price range.
The distribution of these scores was assessed through graphical analysis, and was found to be screwed to the left. In order to acquire a normal distribution of the data, the square root was computed for this variable values, which resulted in relatively normal distribution implying relevance of the data set. Please see Appendix 3 for comparison between histograms for original and computed variable “willingness to pay”.
Food preferences
4.3. Test of hypotheses
4.3.1. Hypotheses H1a, H1b and H1c
Hypothesis H1a states that seeing one self in the mirror inside the dressing room increases private self-awareness. Hypothesis H1b suggests that seeing one self in the mirror outside of the dressing room increases public self-awareness, while hypothesis H1c states that if one does not see her self in the mirror, self awareness (both public and private) will be lower comparing to the condition with the mirror.
In order to establish the potential link between the position of the mirror and self-awareness, new category variable was computed and named a “categorymirror”. Values for this variable were taken over from the item which indicated position of the mirror in the store. Three new values were recorded and indicated “1” as “no mirror” condition, “2” as “mirror inside the dressing room” condition, and value “3” as “mirror outside of the dressing room” condition. With the intention of testing hypothesis H1, a one-way analysis of variance (ANOVA) test was run. In testing the relationship between the position of the mirror and private self-awareness, variable “private self-awareness” was selected as the independent variable and the new category variable “category mirror” was selected as independent (factor) variable. The results are presented in Table 5. Given that F(2,119) = F(.606), p>0.05, no main effects were found of the position of the mirror on private self awareness. Therefore, H1a is not supported.
Table 5. ANOVA test for dependent variable private self-awareness
Sum of Squares df Mean Square F Sig
Between Groups 1.733 2 0.867 0.606 0.547
Within Groups 170.103 119 1.429
*Significant at p<0.05
In testing the effect that the position of the mirror has on public self-awareness, same procedure was followed, with the exception of selecting “public self-awareness” as dependent variable. Results are presented in the Table 6. A significant F-ratio F(2,119) = F(3.552) p<.05 was revealed, which indicates that alternative hypothesis H1b can be accepted.
Table 6. ANOVA test for dependent variable public self-awareness
Sum of Squares df Mean Square F Sig
Between Groups 11.934 2 5.967 3.552 0.032
Within Groups 199.934 119 1.680
Having obtained a significant result, further post-hoc analysis was done in order to determine where the significance lies; that which positions of the mirror significantly affects the public self-awareness. The results of the multiple comparisons between conditions are presented in Figure 1. Given the mean differences significance (p = 0.021) for the relationship between the 2 conditions (“no mirror”, and “mirror outside of the dressing room”) and public self-awareness, it can be concluded that:
- when participants saw themselves in the mirror outside of the dressing room, their public self-awareness was higher compare to the condition when they did not see themselves in the mirror, confirming hypothesis H1b
- when participants did not see themselves in the mirror, their public self-awareness was lower compare to when they saw themselves in the mirror outside of the dressing room, confirming hypothesis H1c.
Figure 1: Means and p-values between conditions (no mirror, mirror inside and mirror outside)
4.74 5.24 5.56 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 Means
No mirror Mirror isnside Mirror outside
Categorymirror
Mean
*p = 0.021
p = 0.058
4.3.2. Hypotheses H2a and H2b
With finding correlation between the position of the mirror and increased public self-awareness, the relationship between the ad type and self-evaluations, moderated by position of the mirror, needed to be examined. Namely, hypotheses H2a and H2b predicted respectively that seeing one self in the mirror and being exposed to the thin model causes decreased self-evaluations, compared to the state when the mirror is not present, while seeing one self in the mirror and being exposed to the round model causes increased self evaluations, compared to the state when mirror is not present.
In order to asses the main effect and interaction effect of the independent variables (mirror and ad type) on self-evaluations, a two-way ANOVA test was applied. The output presented in Table 8 illustrates that the main effects of the condition mirror/no mirror and type of the ad (thin vs. round model) are not significant (p>0.05). Therefore, neither the position of the mirror nor the body size of the model significantly influences women’s self-evaluation. Because neither main effect was significant, post-hoc analysis were not required.
The output also shows that there was no significant interaction effect (p>0.05) for the position of the mirror*add type. That is, the influence of the model’s body size on self-evaluations does not depend on whether the mirror is placed inside or outside of the dressing room.
However, with the intention of gaining further insight into the relation of variables, additional measures were included, such as Body Mass Index (BMI). The test was run to examine if there are any significant difference effects among women with lower than normal BMI (BMI under 19), as well as among women with higher than normal BMI (BMI over 25). No significant differences were found.
Consequently, hypothesis H2a and H2b were not supported.
Table 8. Test of between-subjects effects for dependent variable self-evaluations Source
Type III Sum
of Squares df Mean Square F Sig. mirror AdType mirror * AdType 0.991 0.105 0.023 1 1 1 0.991 0.105 0.023 0.886 0.094 0.021 0.349 0.760 0.886
4.3.3. Hypotheses H3a, H3b and H3c
Similar to the previous test of hypotheses, in order to asses whether the type of the ad, moderated by the effect of relevance of the product, has an impact on product evaluations, a two-way ANOVA analysis was used. In order to present the relevance of the product, new variable named “categorybeautyconcept” was created and its values “0” and “1” represented whether the last store participants have visited was beauty-related (“1”) or not (“0”). The distribution of the data for the newly computed variable was proven to be normal.
The univariate analysis of variance again did not show any significant results (p>.05) (see Table 9). That is the ad type and product relevance did not have any effect on participants’ product evaluations. Moreover, the interaction effect of the two independent variables was also not significant, in this way rejecting hypotheses H3a, H3b, and H3c.
Table 9. Test of between-subjects effects for dependent variable product evaluations Source
Type III Sum
of Squares df Mean Square F Sig. 0.066 1 0.066 0.045 0.832 1.036 1 1.036 0.708 0.402 AdType categorybeautyconcept AdType * categorybeautyconcept 3.474 1 3.474 2.374 0.126
*R Squared = 0.030 (Adjusted R Squared = 0.005)
4.3.4. Hypotheses H4a, H4b and H4c
To test whether there are significant differences in participants expressed product evaluations and their willingness to pay for AVAN, test of between-subjects effects was conducted. Root-squared willingness to pay was used as the dependent variable, while independent variables were kept the same as in testing product evaluations.
Table 10. Test of between-subjects effects for dependent variable willingness to pay Source
Type III Sum
of Squares df Mean Square F Sig. categorybeautyconcept 3.417 1 3.417 6.731 0.011 AdType 0.754 1 0.754 1.484 0.226 categorybeautyconcept * AdType 1.707 1 1.707 3.363 0.069
*R Squared = .067 (Adjusted R Squared = .043)
Figure 2 illustrates that for the exposure to the thin model, product relevance has no influence on willingness to pay, rejecting hypothesis H1a. However, with the exposure to the round model, product relevance has considerable impact: participants who have visited the last store which is related to the beauty concept exert the highest willingness to pay. Figure 2: Comparison of willingness to pay with respect to the interaction effect between the body
size of the model and the salience of beauty concept
categorybeautyconcept Yes No M e a n S Q W il li n g n e s s T o P a y 2.20 2.00 1.80 1.60 1.40 Round Thin
Thin or Round model
Interpretation based on a visual inspection of the plots was checked with inferential statistics. From the cell means presented in Table 11 the conclusion can be drawn that participants who had their last visit to one of the stores indicating salience of a beauty concept, and were exposed to the round model were ready to pay the highest amount of money in order to attain AVAN (M = 4.64), while participants who have not visited any beauty related store expressed less willingness to pay for AVAN (M = 2.01).
Table 11: One-way ANOVA for dependent the newly variable willingness to pay
95% Confidence Interval for Mean Conditions Mean Std. Deviation Lower Bound Upper Bound
No beauty store Beauty store 1.4169 2.1550 0.55135 0.78866 0.9931 1.9180 1.8470 2.3919
*Values are square-rooted since variable used in the analysis is square-rooted.
4.3.5. Hypotheses H5a and H5b
To test hypotheses H5a and H5b, indicating a relationship between food preferences and self-evaluations and type of the ad, Binary Logistic Regression test was employed.
Binary logistic regression assumes that the dependent variable is dichotomous (or so called Dummy variable), and that the outcomes are independent and mutually exclusive (Leech, Barrett, Morgan, 2005). In this research a binomial response is modeled that gives only zero’s (0) and one’s (1). The model accounts for the fact that only those two outcomes are possible, meaning that the participant can take healthy (“value 1”), or not take healthy (“value 0”) snack. Checking for multicolinearity was not necessary since only two covariates were used as predictors.
Summary of the second part of the analysis, called Beginning Block, is given in the Tables 11 and 12. From the Classification table (Table 12) it can be drawn that 79 participants did not take healthy snack, while 36 participants did. Thus, if only simply guessed that no participants took healthy snack, one would classify 68.7% participants by chance. Table 13 indicates how significantly each independent variable is related to the dependent variable. Thus, it can be concluded that neither self-evaluation (p=0.424), nor type of the ad (p=0.101) is significantly related to the choice of snack.
Table 12. Classification
Predicted
Healthyornot % Correct Observed Not taken Taken
Not taken 79 0 100.0
healthyornot
Taken 36 0 0.0
Step 0
Overal % 68.7
When both predictor variables were considered together, the Model was not significant (χ2 = 3.316, df = 2, p = 0.191), therefore not supporting hypotheses H5a and H5b. Table 14 represents a summary overview of interaction between variables in predicting who will choose a healthy snack. Odds ratio show that odds estimating correctly who chose healthy, and who chose unhealthy snacks were not improved if one participant indicated higher or lower self-evaluation, and if the participant was exposed to the thin or to the round model. Findings reported in these tests are not significant enough to support suggested hypotheses H5a and H5b.
Table 14. Logistic Regression predicting food preferences
Variable β S.E. Odds Ratio p
Self-evaluation -0.142 0.189 0.87 0.454
AdType -0.687 0.427 0.5 0.503
