Perceived Impact Of Thin Female Models In Advertising: A Cross-Cultural Examination Of Third Person Perception And Its Impact On Behaviors

Fang Wan Vanderbilt University Ronald J. Faber University of Minnesota Anthony Fung Chinese University of Hong Kong Introduction Contemporary western culture extols the virtues of slenderness while promoting a fear of fat (Dionne, Davis, Fox and Gurevich, 1995). In a 1997 Psychology Today Body Image Survey, eighty-nine percent of the women, and fifty-four percent of the men said they want to loose weight, and more than sixty percent said they are dissatisfied with their overall appearance-especially their body weight, hips and abdomen (Garner, 1997). Such dissatisfaction may cause individuals to resort to risky means to control their body weight and appearance. For example, a national study of women found that a 1 out of 100 females is anorexic and 3 out of 100 are bulimic (American Psychiatric Association, 1994). Other attempts to control body image include smoking, bingeing and purging, excessive exercise, dieting and surgical procedures. These practices suggest a worrying picture of young women who are obsessed with their appearance. Western society has been criticized for its emphasis on a slim physique and negative stereotyping of obese figures. The cultural ideals of thinness has been proposed to cause mass dissatisfaction with body shape and weight concerns among the female population, which ultimately lead to negative attitudes toward eating, a preoccupation with weight and dieting and a high prevalence of eating disorders (Tiggeman & Rothblum, 1988; Powell & Kahn, 1995). However, it is unclear if this phenomenon is unique to a Western cultural context and to Western women only. Do young women in Non-Western culture manifest similar patterns of weight concerns and weight-controlling behaviors? What role do idealized images in advertising play in the perception of body image and image-enhancement behaviors? This paper attempts to investigate the role of thin female models in advertising in the body image perception and image-enhancement behaviors in a cross-cultural context. It begins with a literature review of cross-cultural studies of body image perceptions and behaviors. It then employs a relatively new theoretical perspective to attempt to account for a less direct impact of the mass media on behavior. This is referred to as the third person effect . this model will be explained and discussed in both a North American and East Asian context. Lastly, it investigates the relationship between the third person perception of the impact of thin female models and image-enhancement behaviors in a cross-cultural context.

Cross cultural research on body image perceptions and eating attitudes: Cross-cultural research on eating disorder and body image perception has been limited, the majority of this literature has dealt with whether Western norms and values of female thinness influenced non-Western womens, (specifically Asian womens) eating patterns and perceptions of their body image (e.g., Dolan, 1991; Ahmad, Waller & Verduyn, 1994; McCourt & Waller, 1995; Akan & Grilo, 1995; Rosenthal & Feldman, 1990). Research in this area has focused on either women immigrated into a Western culture (e.g., Furham & Alibhai, 1983; Lake, Staiger & Glowinski, 2000), or women exposed to Western values in their home country (Mumford, Whitehouse & Choudry, 1992). Studies have found an increase in the incidence of eating disorders among non-Western women entering Western society (Dolan, 1991). Two opposing explanations have been offered for these findings. The first attributes an increase in maladaptive eating patterns to a culture clash experienced by individuals who migrate to a new country. This leads to pressure to adapt to a new culture which then causes problematic behavior. Research evidence supporting this view has shown that more traditional females (i.e., those having a stronger identity with their country of birth) are at a higher risk of developing eating disorders due to the difficulties experienced in growing up with conflicting cultural values (Ahmad, Waller & Verduyn, 1994; McCourt & Waller, 1995). A recent study by Lake and colleagues (Lake, Staiger & Glowinski, 2000) compared Australian-born and Hong Kong-born University students residing in Australia They found that Hong Kong-born subjects who demonstrated a higher level of Chinese identity (more traditional) were more affected by Western attitudes toward eating and body image than those who were influenced by Western values. Cultural assimilation provides an alternative view for explaining the development of eating disorders among non-Western women. This theory suggests that non-Western women demonstrate an increasing pattern of eating disorders when they move to a Western culture because they need to assimilate the host society norms and values. This includes assimilating to a new image of the ideal female body shape. Some literature has indicated non-Western women who have a higher level of assimilation of Western norms and values are at a higher risk of developing eating disorders (Akan & Grilo, 1995). Body image among Asian women Much of the existing literature on the body image of Asian respondents has examined people in Japan (Endo, 1992; Kowner & Ogawa, 1993; Shibata & Nobechi, 1991), or Asians living in Europe or North American (e.g., Arkoff & Weaver, 1966; Furnham & Baguma, 1994; Hill & Bhati, 1995; Marsella, Shizuru, Brennan, & Kameoka, 1981). More recently, there is a growing number of studies examining Chinese men and women in Hong Kong (Lee et al, 1997; Davis & Katzman, 1997; 1998; 1999). Davis & Katzman (1997) have argued that data on women from modernizing cultures in Asia such as Hong Kong is important for cross-cultural comparisons since these cultures are experiencing rapid social change in the face of an influx of Western values and ideals. When compared to European American male and female students, Chinese-born undergraduates in Hong Kong demonstrate a similiar pattern of dissatisfaction with various body

parts (Lee et al., 1993). Lee et al (1993) argued that young Chinese women share a similar ideal of slimness as their Western counterparts, although dieting and eating disorders remain rare. In a series of studies, Davis & Katzman found that Chinese women in HK held more negative views of themselves, their bodies and their mood when compared to Chinese women in the United States (Davis & Katzman, 1997; 1998). They also found Chinese subjects in Hong Kong reported significantly more body and weight dissatisfaction, lower self-esteem, higher depression, more dieting and less exercising as compared to their counterparts in the United States (Davis & Katzman, 1998). Davis & Katzman (1998) concluded that Asian students mirrored gender patterns previously reported in Caucasian samples with the respect to the relation of body image, self-esteem and mood. Interestingly, researchers (Lee et al, 1996; Katzman, 1996) found that even though the current weight of Chinese women is below the ideals of European and American women, they were still under pressure to diet. Again, the prevalence of body image and weight concerns in East Asia was ascribed to the increasing modernization and exposure to European and American ideals which leads to the devaluation of the oncesought-after fat body in traditional Asian culture (Buhrich, 1981). The limited research evidence has shown that Chinese women in HK demonstrate a similar pattern of body image worries and concerns compared to their counterparts in US (e.g., Lee et al, 1993). This study serves to further explore this issue. Therefore, the first goal of this study is to determine if there are differences in body image perception and image-related behaviors such as dieting, exercising and the likelihood of plastic surgery between young women in the U.S. and Hong Kong. RQ1: Do young women in Hong Kong and United States demonstrate the same level of body image concerns and disturbance and the same intensity of engaging in image-related behaviors? Third Person Effect: A Theoretical Framework Researchers have claimed that the ubiquitous ultra-thin ideal female images presented as the norm in the media may be responsible for perpetuating highly unrealistic ideals in women (e.g., Fallon, 1990; Harrison, 2000). These authors assume that this leads to the development of unhealthy behaviors. In contrast, the third-person effect provides a very different explanation of how media images may influence such actions. Postulated by Davison in 1983, the third-person effect is the perception that persuasive communication will exert a stronger impact on others than on the self. As Davison (1983, p.3) noted, individuals exposed to a mediated message typically believe that the message will not have its greatest effect on me (the first person), or you (the second person), but on them third persons. The robustness of the third person effect has been demonstrated across a broad range of media using survey and experimental methods (e.g., Cohen et al., 1988; Gunther, 1991, 1992, 1995; McLeod et al., 1997; Perloff, 1989; Salwen & Driscoll, 1997). The third-person effect framework postulates an intermediate process of media impact. Instead of assuming a direct link between media exposure and attitudes/behaviors, third-person effect lays out a two-step persuasion process. According to this framework, media exerts influence not through ones own interaction with media message, but through the projection of how others may respond to those messages. To some extent, the estimation and projection of others reactions to media messages is a constituent part of perceived social norm of media

persuasiveness (Wan, 2001). Thus people consistently demonstrate a perceptual gap between media susceptibility of themselves and others. It is the perceived belief that the media exert an influence on others that drives individuals to take some action (Wan, 2001). Research on the third-person effect has found that it only exists in certain situations. When a message is perceived negatively, or when being persuaded by it would be generally regarded as unintelligent, people estimate more that others will be more influenced than they will. However, when a message is considered positive, people attribute themselves as equally or more effected since they are smart enough to recognize its value (Cohen and Davis 1991; Gunther and Thorson 1992). Other perceptions or attitudes regarding the message also influence the size and direction of the perceptual gap. Some studies have found that there is a greater disparity between perceived effects on the self and others when the source of the message is judged to be negatively biased (Cohen et al. 1988; Gunther 1991) or when the audience attributes persuasive intent to the communicator (Gunther and Mundy 1993). Other research shows that those who consider an issue important (Matera and Salwen 1996), perceive themselves as experts (Lasorsa 1989), or are highly ego-involved in the message (Perloff 1989) tend to perceive that others will be more affected by message content. Embedded in the third person effect framework are additional intricacies in the relationships between self and others which can influence the perceptual gap. Research findings have shown that different conceptualizations and definitions of others influence the direction and magnitude of the third person effect. When others are defined in terms of geographical closeness to the first person, the discrepancy between perceived media effects on others and the self are not as large as when others are more distant (Cohen et al., 1988; Cohen and Davis, 1991; Gunther, 1991; Duck & Mulin, 1995). For example, Cohen et al. (1988) found that while subjects perceived that news stories would exert a greater impact on other Stanford students than on themselves, this difference was even greater when the self was compared with other Californians and still greater when compared to the public at large. When others are defined in terms of perceived similarity or social distance (e.g., Brosius & Engel, 1996; Eveland et al., 1999), individuals again tend to see less difference between themselves and similar groups (closer others) than with dissimilar others (more distant groups) (Eveland et al., 1999). For example, people tended to infer greater negative media effects on groups less similar to themselves (average others) than with more similar groups such as friends (Brosius and Engel, 1996). Differences in third-person effects can also be defined in terms of gender (Duck, 1997; Wan & Wells, 2001). In a study of perceived effect of commercial advertising, Duck (1997) found that both male and female respondents perceive themselves as less influenced by commercial advertising than othersmen in general and women in general. However, both male and female respondents perceived womenthe subordinate group in social stereotypeas more vulnerable than men were to the effects of commercial advertising (Duck, 1997). Wan & Wells (2001) also defined others in terms of gender identity. Consistent with Ducks finding, Wan & Wells (2001) found that both female and male respondents perceived ads with sexual female images to have less impact on themselves than others. However, they found an

interesting effect of similarity of the other. A gender similar group (for male respondents, men in general; for female respondents, women in general) was perceived to be less influenced by sexual image ads than gender outgroup (for male respondents, women in general; for female respondents, men in general). Part of the goals of this study were to examine how people would perceive the impact of thin female models on different others. Both gender identity and social distance were used to define others. Specifically, five different others were examined: a) female friends; b) male friends; c) most women; d) most men; and e) my current/potential romantic partner. These sets of others are assumed to be important agents in a persons social interactions. They range from being close (romantic partner; female friends; male friends) to distant (most men; most women) and from being same gender in-group members (female friends; most women) to outgroup members (male friends, most men). To generate specific hypotheses regarding third person effects, we need to consider the social desirability of the medias influence. When the media content is seen as having socially undesirable outcomes others are perceived as more vulnerable to this effect than ones-self (third person effect). However, when the media content is judged to be positive, either no difference, or a belief that the impact will be greater on ones-self than on others (the first person effect) is likely. Social critics, media educators, and others have vehemently criticized media and advertising for their unrealistic portrayal of female images and the negative impact those idealized images have on young people (e.g., lowered self-esteem; eating disorder likelihood; excessive exercising). Therefore, it is reasonable to assume that susceptibility to idealized images in advertising will be viewed as having a negative connation and be seen as socially undesirable. Hence, people should deny that they themselves are susceptible while being more likely to assume that thin female models portrayed in advertising influence others. We would then expect a classic third person effect: H1a: Thin female models in advertising are perceived to have less impact on self than on others (third person effect). However, as noted earlier, the magnitude of the third person effect is moderated by social distance. If others are more distant, the perceived impact of thin female models on them should increase. Since, people have limited information about distant others, they adhere more to social stereotypes to estimate the impact. In this situation, social stereotypes seem to be that thin female models in advertising set the standards of how women should look. So, when estimating the impact of the idealized images on these distant others, people would retrieve those social stereotypes and infer that others are more vulnerable to the idealized images than themselves. However, when others are relatively closer to one self, for example, friends or romantic partner, people would attribute more similarity in abilities and thoughts to them. Therefore, we would expect that: H1b: The gap between perceived impact of thin female models in advertising on self and others is smaller when others are close to self (i.e., female friends; male

friends; romantic partner) than when others are more distant to self (i.e., most men; most women). In addition, as discussed earlier, gender identity can also influences the third person perception. Female respondents should think that men are more influenced by thin female models than they themselves or other women are (Wan & Wells, 2001). This is because ingroup members are perceived to be more similar to others, and therefore would be similarly influenced by media messages (Duck, Hogg & Terry, 2000). Therefore, we would expect that: H1c: The gap between the perceived impact of thin female models in advertising on self and others is smaller when others belong to ones gender ingroup (i.e., female friends, most women) than when others belong to gender outgroup (romantic partner, male friends, most men). A cross-cultural inquiry into the third person effect In a comprehensive review of third person effect research, Perloff (1999) summarized various factors influencing the third person effect. Among these factors are message desirability, social distance of others, media use orientations, knowledge, issue importance, self-esteem and ego-involvement. However, he pointed out that there is a need to further examine possible contextual variations in the third person effect. One of the contextual factors he mentioned is cultural context. To date, there are no published study examining a cross-cultural difference in the third person effect. Cross-cultural research on this phenomenon is important to determine its universality. Perloff (1999) speculated that third-person effects may be less pronounced in Asian cultures. He based this belief on the fact that Eastern cultures emphasize the interrelatedness of people with others in their social environment, while in Western culture, the focus is more on the uniqueness and autonomy of individuals. Literature in cultural psychology, especially in the area of different self-construal in Eastern versus Western culture, support Perloff s proposition. In general, the key differences between Asian and Western culture is the definition of selfhood or self construal (Markus& Kitayama, 1991; Kitayama, Markus and Kurokawa, 1994; Kim & Markus, 1999). Markus and Kitayama (1991) defined the independent construal of self as characterized by a bounded and autonomous sense of self that is relatively distinct from others and the environment. This view is best exemplified by North American and Western European cultures. In these cultures, individuals strive to assert their individuality and uniqueness and stress their separateness from the social world. In contrast, the interdependent construal of self emphasizes the interrelatedness of the individual to others and to the environment (Markus and Kitayama, 1991). It is only within the contextual fabric of social relationships, roles, and duties that the self has meaning. This construal of self is most representative of Asian cultures. Miller (1993) employed another concept co-subjectivityto describe the interdependent self-construal in Asian culture.

According to Miller (1993), co-subjectivity in Japan is the formation of subjects through identification with a group or community (p.482). In a culture characteristic of interdependent self-construal, attending to others, fitting in, and harmonious relations are valued and socially rewarded (Markus and Kitayama, 1991). Therefore, we would expect that people from such a culture would tend to focus on the similarity between self and others. In contrast, in a culture characteristic of independent construal of self, inner attributes of a person such as self-sufficiency, autonomy and actualizations of ones goals, desires and needs are valued. Therefore, when engaging in a social comparison process, we would expect that people from this type of culture will emphasize the dissimilarity between self and others. The above proposition gains some support from a cross-cultural study on unrealistic optimism (Heine & Lehman, 1995). They found that people from cultures representative of an interdependent construal of self do not engage in self-enhancement to the same extent as people from cultures characteristic of an independent self. They believe that self-enhancing biases serve to buttress ones self-assessments to help achieve a norm valued by an independent culture. However, in an interdependent culture, people realize their cultural ideals by self-effacement. This minimizes their distinguishing, and potentially alienating, features and allows them to maximize their sense of belongingness (Heine & Lehman, 1995). Additional research findings with American subjects suggest that the self is judged to be more dissimilar to other than other are to the self (Kitayama, Markus, Tummala, Kurokawa, & Kato, 1990; Holyoak & Gordon, 1983; Srull & Gaelick, 1983). This pattern, however, is reversed for Eastern subjects, who judged the self somewhat more similar to others than the other is to the self. It appears that, for more interdependent subjects, knowledge about others is relatively more elaborated and distinctive than knowledge about the self. Applying this to third person effect leads to an expectation that there will be cultural differences. In Asian cultures, it is assumed people will focus on their belongingness to and interrelatedness with others, and will therefore yield similar estimation of media vulnerability for themselves and others. However, for Westerners, differences between the self and others will be intensified to enhance ones superiority over others. This will lead to a larger difference in perception of the medias influence on self versus others. Thus, we would expect that: H2: The gap between perceived impact of thin female models in advertising on self and others is larger in an independent culture (e.g., United States) than in an interdependent culture (e.g., Hong Kong) Behavioral consequences of third Person Effect: A considerable amount of research on the third person effect has demonstrated that perceptual gaps exist for a wide range of messages and social contexts. However, explorations examining behavioral components of the third person effect have been meet with far more limited success. By far, the most commonly documented behavioral impact of a gap in thirdperson perceptions is a greater willingness to support censorship of content that is deemed undesirable (Perloff, 1999).

Researchers have found that the difference between the perceived impact on others and on ones-self is related to peoples willingness to advocate censoring media materials such as pornography (Gunther, 1995; Rojas, Shah, & Faber, 1996), violence on TV (Gunther & Hwa, 1996), advertising for controversial products (Shah, Faber, & Young, 1999; Youn, Faber, Shah, 2000) and violent and misogynic rap lyrics (McLeod et al., 1997). However, in some cases regarding news and political communications, a behavioral effect has not been found. For example, third person perceptions did not predict willingness to limit press coverage of the O.J.Simpson trial (Salwen & Driscoll, 1997). Similarly, this gap failed to significantly predict support for an independent commission to regulate political communications (Rucinski & Salmon, 1990), or opposition to printing a Holocaust-denial advertisement (Price, Hung & Tewksbury, 1997). Testing whether third person perceptions predict willingness to censor media content is at best only a small subset of the possible behavioral components of the third person effect. Yet little evidence for other aspects of this hypothesis have previously been found. Recently, Wan (2001) proposed that the belief that the perceived impact of idealized images in advertising on others can influence an individuals behaviors. These behaviors may include dieting, exercising and the likelihood of plastic surgery. Integrating social influence theory with the third person effect, she suggested that the third person perception be conceptualized as the perceived social norm of media persuasiveness (Wan, 2001). That is, if media influence is desirable for a social group, and an individual identifies with that group, he/she would comply to the group norms by acting in accordance with the media messages. An individuals conformity to perceived norm of media influence results either from their internalization of group norms and values, or from a simple self-presentation motive-- to project an appropriate image in order to fit in. It should be noted that a different psychological mechanism is predicted here than is at work in censorship effects. In the case of the third person perception and willingness to censor, a paternalistic motivation is perceived to be the driving force. People believe others will be harmed by media messages and need to be protected from the effect of those messages. However, here it is believed that the third person perception and behavior linkage is driven by a desire to conform to group norms that are thought to be shaped by media presentations. This study attempts to empirically test this component of the third person model. Here, if ones peers (male and female friends), romantic partners and larger social group (most men or most women) are perceived to be influenced by thin female models in advertising, then acceptance by these groups will depend on adhering to these norms. Thus, even if advertising images do not directly influence ones own perceptions, they can still create a need for behavioral change. Hence, it is predicted that: H3: The gap between perceived impact of thin female models in advertising on self and others will predict image-related behaviors. The final goal of this study is to test the third person model of behavioral impact in a cross-cultural context. Since the role of others is defined differently for interdependent and independent construal of self, there could be a cultural difference in terms of the level of social

conformity. According to Markus & Kitayama (1999), others in independent cultures serve primarily as standards of reflected appraisal, or as sources that can verify and affirm the inner core of the self. However, others in an interdependent culture are assigned more importance, carry more width, and are relatively more focal in directing ones behavior. To maintain a connection to others, interdependent people are constantly aware of others needs, desires and goals. Additionally, they are more likely to internalize the goals of others and make them their own. Furthermore, in an interdependent culture, following norms is a core cultural goal that fosters group harmony (Hsu, 1948). As Markus & Kitayama (1991) pointed out, in an interdependent culture, there is a need as well as a strong normative demand to know and understand others thinking, feeling and action in the context of ones relationships to them. Compared to people in an independent culture, the most important element of selfconstrual in interdependent societies is the self-in-relation-to-others element. This could intensify the pressure to conform to group norms and strengthen the relationship between perceived social norm and ones own behavior. Therefore, compared to people in an independent culture, people in interdependent cultures tend to demonstrate a higher level of social conformity (Kim & Markus, 1999). It is, therefore, theoretically reasonable to expect that the relationship between third person perceptions and image enhancement behaviors will be stronger in an interdependent culture than in an independent culture. However, the value of prior research in this area is limited by the fact it has only examined public behavior. Most cross-cultural research on social conformity has addressed ones expressed opinion in front of a group after the group opinion has been presented (e.g., Huang & Harris, 1973; Meade & Barnard, 1973; Bond & Smith, 1996). This is a very public indication of behavior. However, the behaviors examined in this study are private in nature and lack an overt confrontation with group norms. Thus this hypothesis must be considered very tentative and is, instead, presented in the form of a research question. RQ2: Is the relationship between the third person perceptions of thin female models and image-related behaviors stronger in an interdependent culture than in an independent culture? Methodology: Sample Altogether, 505 female respondents participated in the study. Chinese respondents came from an introductory communications class at a major university in Hong Kong. There were a total of 122 Chinese women who participated. The questionnaires distributed in Hong Kong were written in English, given the high English proficiency of Hong Kong students. In fact, at many universities in Hong Kong, English is the required language for instruction. Respondents from Hong Kong sample were all between the ages of 19 to 28. The U.S. data were collected by the students enrolled in an advertising course at a Midwest university as partial fulfillment of a course project. Members of the class who participated in this project each interviewed 10 women between the ages of 18 to 30. A total of 383 U.S. women were interviewed.

Sample Comparison: The mean age of the U.S. respondents was 21.6 , while the average age of the Hong Kong respondents was 20.6. The average height of the Hong Kong women was 159.8m or 5 feet 3 inches. Their average weight was 49.1 mg or 108 lbs. For the U.S. respondents, the mean height was 167.8m or 5 feet 6 inches, and their average weight was 62.3 mg or 137.2 lbs. Thus, the U.S. women were both heavier (F(1,483)=183.9, p<.001) and taller (F(1,483)=116.8, p<.001) than the Chinese women. This difference is consistent with findings of other researchers (Lee et al., 1993). A relatively high percentage of both samples (80.9% of the U.S. women and 72.6% of the H.K. women indicated that they wanted to lose weight. Measures: Perceived first and third person effects. Most third-person studies have relied on single items to assess perceived impact of the media on self and others. To try to improve on this approach, the current study used multi-item scales to assess the perceived impact of thin models in advertising. For each person assessed, the perceived impact of advertising models was measured with four separate questions. These were: a) Thin female models in advertising have a powerful impact on _______; b) Seeing thin female models in advertising makes _______ feel that women should be thinner; c) Thin female models in advertising influence ________ perception of how women should look; and d) Seeing thin female models in advertising makes _______ feel less satisfied with how I (or they or their girlfriends) look. The respondents were asked to estimate the impact of thin female models (measured by the four questions) for themselves and 5 different others. These five others were: their female friends, their male friends, most women, most men and their current/potential romantic partner. These four questions by six possible people lead to a total of 24 items to assess the third person questions. Each item was phrased as a statement and respondents were asked to indicate their level of agreement using a five point scale ranging from 1 strongly disagree to 5 strongly agree. To create a scale of perceived impact of thin female models on self and various others, the four items measuring perceived impact on each person was subjected to an exploratory factor analysis. Principal component analysis with Varimax extraction method was adopted. Reliability tests were also conducted to examine internal consistency of each of the scales. When the four items measuring perceived impact of thin female models on me was submitted to a factor analysis, a single factor solution was extracted, explaining 73.5% of total variance. A reliability tests on these four items yielded an acceptable level of reliability (=.88). Therefore, averaging the four items created an index of perceived impact of thin female models on me. Averaging allowed the index that retained the range of 1 to 5. Similarly, factor analyses were conducted on the four items measuring the perceived impact of advertising models on each of the various others examined in this study. In all cases, this produced single factor solutions. These separate analyses explained between 57.5% and 72.4% of the total variance for each of the others. Reliability scores were then computed for each set of items. In all cases, these demonstrated acceptable levels of reliability ranging from =.75 to .90. As a result, each of the four item sets were averaged to produce an indicator of advertisings perceived impact on male friends, female friends, most men, most women and romantic partner.

10

Outcome measures. Three kinds of potential outcome measures were assessed in the study. Two of these were behavioral measures (dieting and exercising) and one was a behavioral intent measure (likelihood of having plastic surgery). Dieting was measured by a single item: How frequently have you been on a diet? The respondents were asked to choose from the response: never, rarely, sometimes, often, and always. Exercising was measured by two items. The respondents were asked to indicate how frequently they engage in exercising. This question also used the response categories of never, rarely, sometimes, often and always. The second exercise question asked respondent on average, how many hours per week do you spend exercising? They were given the following eight response categories to choose from: less than 1 hour per week, at least 1 hour, but less than 2 hours, at least 2 hours, but less than 3 hours, at least 3 hours, but less than 4 hours, at least 4 hours, but less than 5 hours or more per week. We assigned the values ranging from 1 to 8 to the response categories in this measure. A reliability test yielded an alpha of .79 for the two exercise items. Based on this, we standardized each of these scores and summed the standardized scores. Thus scores could range from 0 to 2. Actual scores on this exercise scale ranged from .13 to 1.80 (standard deviation=.44). The third behavior variable was the likelihood of performing plastic surgery. A behavioral intention variable was used here since actual prior behavior may have been limited by financial or other considerations. This outcome variable was assessed with two items: If I did not like the way some feature of my face looked, I would consider having plastic surgery to change it, and if I did not like the way some feature of my body looked, I would consider having plastic surgery to change it. The respondents were asked to use a five-point scale with 1 indicating strongly disagree and 5 strongly agree. These two items were then averaged to create a final index of willingness to have plastic surgery. Body Image Disturbance. Body image disturbance was measured by five items selected from Cooper, Taylor, & Fairburn (1987) s seventeen-item index of Body Image Disturbance. The five chosen items were: Have you been afraid that you might become fat or fatter? Have you limited what you eat to stay thin? Have you felt depressed about your shape? Have you avoided wearing clothes that make you particularly aware of the shape of your body? Have you been particularly self-conscious about your shape when in the company of other people? We chose the five items because they formed the subscale of negative body image. The respondents were asked to respond to these questions by choosing one of the response categories: never, rarely, sometimes, often and always. Exploratory factor analysis of the five items with Principal Component Analysis and Varimax Extraction method yielded a single factor solution, explaining 62.7% of the total variance of the underlying construct. The five items had reasonable reliability (alpha=.85) and were therefore averaged to create a body image disturbance index. Results To answer the first research question, we conducted a series of ANOVA analyses with culture as a fixed between-subject factor. The results are shown in Table 1.

11

(Insert Table 1. about here) Cross-cultural differences in five dependent variables were examined. These were body image disturbance, dieting, exercising every day, exercising hours per week and likelihood of having plastic surgery. The results in Table 1 demonstrate significant main effects of culture for three of the five variables examined. Specifically, compared to HK respondents, the U.S. respondents reported a higher frequency of exercising everyday than HK sample (F(1, 504)=26.97, eta-squared=.05, p<.01), and exercised for a greater number of hours per week (F(1, 500)=63.79, eta-squared=.11, p<.01). Additionally, the U.S. respondents indicated a higher likelihood of having plastic surgery if they are not satisfied with some features of their body and face (F(1, 502)=23.66, eta-squared=.05, p<.01). However, the HK and U.S. samples did not differ in terms of body image disturbance (F(1, 504)=.59, n.s.). Additionally, the groups did not differ on the frequency of dieting (F(1, 505)=.83, n.s.) with both the HK and U.S. respondents reporting that they occasionally diet (the cell means fell between rarely and sometimes). Hypothesis 1a H1a predicts a classic third person effect (perceived impact of thin female models on others is larger than on self) regardless of cultural context. To test the hypothesis, we performed a series of paired t-tests to compare perceived impact of thin female models in advertising on self and on different others respectively within each culture. The results are shown in Table 2. (Insert Table 2 here) In order to demonstrate the direction and magnitude of the third person effect, we calculated mean difference between perceived impact on me minus the mean impact on others. The results in Table 2 show the classic third person effect exists when others are female friends, most women and most men. This is true for respondents in both cultures. Thus, both the HK and U.S. respondents perceived that thin female models in advertising exert more influence on their female friends, most women and most men than on themselves. However, the respondents of both culture also demonstrated a surprising first person effect (greater impact on self than other) when the others were their male friends and their romantic partner. Hypothesis 1b and Hypothesis 2 To examine H1b, We first combined three variables (perceived impact on male friends, female fiends and romantic partner) by taking the average of them and created a new variable perceived impact of thin female models on close others. Similarly, we collapsed two variables (perceived impact of thin female models on most men and most women) by taking the average of the two and created another variableperceived impact of thin female models on close others. To examine the impact of social distance of others and culture on the third person gap, we ran a repeated measures MANOVAs with one within-subject factor (social distance of others; 2 levels: close versus distant others) and one between-subject factor (HK versus US). The results are shown in Table 3. (Insert Table 3 here) Hypothesis 1b stated that when others are socially closer to respondents (e.g., friends and romantic partners), the respondents would think that thin female models in advertising will affect

12

both themselves and others similarly. However, when others are phrased in a more socially distant way (such as most men or women), respondents will perceive that thin female models in advertising will influence others to a greater extent than themselves. Hypothesis 1c Hypothesis 1c stated that there would be a significantly smaller gap when the other was ones gender in-group than when the other was composed of gender out-group members. To examine this hypothesis, two variables were created. Perceived impact of thin female models on gender out-group (male others) was created by averaging three variables--perceived impact on male friends, most men and romantic partner). Taking the average of perceived impact on female friends and most women created the impact of advertising models on gender in-group. Again, a repeated measure MANOVA was computed with one within-subject factor (social distance of others; 2 levels: close versus distant others) and one between-subject factor (HK versus US) to examine the impact of gender identity of others and culture on the third person gap. The results are shown in Table 4. (Insert Table 4 here) First, within each culture, class third person effect exists when others are women. The negative and significant gaps between perceived impact on self and on female others suggested that the respondents from both HK and US agreed that the thin female models in advertising would affect other women to a greater extent than themselves. However, when others are men, first person effect emerges for US women and HK women demonstrated no difference between perceived impact on me and male others. Furthermore, MANOVA analysis yielded a significant main effect of gender identity on the third person gap (F(1,489)=417.29, Eta-squared=.46, p<.01), demonstrating that when others belong to gender ingroup (i.e., most women and female friends), the third person effect emerges. However, when others are members of gender outgroup (i.e., most men, male friends and romantic partner), first person effect emerges. This is opposite to what H1c predicts. The findings reveal that respondents in both culture tend to think that women are more influenced by the thin female models while men are less likely to adopt the standards of thin female models in advertising to judge women around them in real life than respondents themselves and women. Interestingly, gender identity moderated the direction of the third person effect. If others are gender outgroup members, first person effect emerges; if others belong to gender ingroup, third person effect surfaces. Hypothesis 2 Hypothesis 2, predicting a cultural difference in the third person gap, was only partly supported. No significant differences between the U.S. and HK women were found for the impact of advertising models on themselves. However, cultural differences were found in the perceptions of the impact of advertising models on female friends and most women. In both cases, the U.S. women perceived a greater impact on other females than did the Chinese women. No significant difference by culture was found for the perceived impact on males, with the exception of romantic partner. Here, HK women perceived a greater effect on their romantic partner than did the American women.

13

A contrast between distant and close others in terms of a third person gap is more prominent in the U.S. sample than among HK respondents. This is indicated by the significant interaction effect of culture and social distance of others on third person gap (F(1,489)=6.34, Etasquared=.013, p<.05). This difference appears to be primarily due to the fact that American women saw the media as having slightly less impact on close others than themselves, while the HK respondents perceived close others as being slightly more affected than oneself. The gap for distant others in both cultures was very similar. Moreover, a contrast between female and male others in terms of a third person gap is more prominent in the U.S. sample than among HK respondents. The difference is evidenced by a significant interaction effect of culture and gender identity on the third person effect (F(1,489)=19.42, Eta-squared=.04, p<.01). Since US and HK sample do not differ in terms of the third person gap between self and female others, the interaction effect is largely due to the fact that HK women perceived that male others were as equally affected by thin female models in advertising as themselves while US women saw media as having less impact on male others than themselves. Hypothesis 3 To examine whether the third person gap predicts behaviors and whether the prediction varies according to cultural context, we conducted three path models for HK and US sample respectively. Each model examined a separate dependent variable (dieting, exercise and likelihood of getting plastic surgery). In the path models, third person perceptions were treated as exogenous variables directly predicting both body image disturbance and the specific behaviors mentioned above. Body image disturbance also mediates the relationship between third person perception and behaviors creating an indirect path. We examined alternative path models in which body image disturbance was another exogenous variable along with the third person perceptions rather than an endogenous variable mediating the relationship between the third person perceptions and behaviors. However, the chi-squares of the alternative models are significantly larger than the mediation models (average chi-square difference=40.70, df=6, p<.01), suggesting a poorer model fit. Therefore, just the mediation model is presented here. The results for path analysis predicting dieting, exercising and likelihood of plastic surgery are presented in Tables 5, 6 & 7. The path models are also diagramed in Figures presented in an appendix. Unstandardized path coefficients were reported in all figures. (Insert Table 5 here). In general, the third person perceptions did not significantly predict body image disturbance, nor dieting behaviors for HK sample. For these respondents, the only significant path was for the perceived impact of advertising models on self with body image disturbance. Not surprisingly, this was also a significant path for the U.S. respondents. However, for the U.S. sample a few significant paths were also found between some of the gap measures and dieting or body image disturbance. The gap in perceived advertising effect between self and male friends significantly predicts body image disturbance (B=.16, p<.05). In other words, contrary to expectations, if the respondents perceive that thin female models influence themselves more than their male friends, they would demonstrate a higher level of body disturbance. More importantly, the perceived gap between self and most women, and between self and romantic

14

partner significantly predict dieting behavior for US sample. Surprisingly, the direction of the gaps for these others was different. The more thin female models were perceived to influence ones-self as compared to most women, the more the respondents would diet (unstandardized coefficient=.21, p<.05). On the other hand, when ones romantic partner is perceived to be more influenced by thin female models than ones-self, the respondents would diet. (Insert Table 6 here). The results for exercising behavior demonstrated a different pattern (see Table 6). For this variable, the third person perceptions predict behavior significantly for HK sample, but not for US sample. Among the HK respondents, the frequency and intensity of doing exercise are partly driven by the concerns of how thin models in advertising influence others and whether others would use models in advertising to judge themselves or other women. Specifically, in HK sample, if the respondents perceive that thin female models influence others, including their female and male friends, most women, and their romantic partner, to a larger extent than themselves, they are more likely to exercise. For these respondents, their exercising behavior is partly driven by their perception of media influence on their peers (female friends, male friends), more distant others (most men) and closer others (romantic partner). As we discussed earlier, for an individual from East Asian culture, in which interdependent self concept is dominant, concerns about and attention to others feelings, thinking, and experience tend to direct ones own behavior. However, US respondents did not demonstrate a linkage between the third person perception and exercising behaviors. (Insert Table 7 here). Finally, the third person perceptions seem to predict the likelihood of having plastic surgery in both countries when others are most men and romantic partner. Specifically, if respondents perceive that they are more influenced by thin female models than most men, they would consider having plastic surgery to change their looks and body shape. The likelihood of having plastic surgery can also be triggered by the perception that ones romantic partner is more affected by thin models than ones-self. In this situation, having plastic surgery to improve ones look and body shape appears to stem from a desire to conform to the perceived expectations of ones romantic partner. Conclusions and discussions: The goals of this study were to: a) examine the cross-cultural difference in body image perception and behaviors; b) examine cultural variations of the third person effect and c) test the linkage between the third person effect and behaviors in a cross-cultural context. The results indicated that HK and US respondents were equally bothered by their body images and demonstrated similar levels of body image disturbance. Additionally, dieting was equally common among women in both cultures. However, the U.S. women exercised more, and were more likely to have considered having plastic surgery than were the Chinese women. The findings seem to suggest that U.S. women are somewhat more active in their efforts to improve their body shape and are more willing to consider more extreme measures to achieve a more idealized look. The finding here are consistent with literature that finds that body image concerns, weight concerns and eating disorders are more prevalent among Western women than among non-Western women.

15

However, it is important to note that the results here represent just a single point in time. They cannot determine the degree to which these cultures may be stable or changing. The findings show that approximately three-fourths of the women in both cultures desire to lose weight. Thus, it may be that in a few more years, similar percentages from both countries will endorse more extreme methods to achieve idealized looks. This study also attempted to examine notions regarding the third person effect in the context of perceived impact of thin female models in ads. U.S. women perceived both female friends and most women to be influenced by media images to a greater extent than did Chinese women. However, the perceived impact of media images on self did not vary across these two cultural groups. Thus, the difference in the two cultures is manifested only in expectations of the effect on other women. This may be due to difference in interpersonal communication or in the cultural norms and beliefs operating in each society. In both countries, a third person effect gap existed when comparisons were made between self and female friends, most women and most men. Surprisingly, a reverse gap appeared for both groups when comparisons were made with romantic partner. A reverse gap was also found for the U.S. sample when comparisons were made with the impact on male friends. This reverse gap for romantic partner and male friends was surprising, especially given that it occurred in both cultures. While the reason for it is unclear, a couple of possibilities exist. It is possible that the males these women know dont appear to pay attention to female models in ads. They dont read fashion magazines or perhaps simple choose not to make comments about the appearance of these models. Alternatively, this may represent a self-protection belief system among women in both cultures. If they believe that the males they actually know (and care about) arent being influenced by these images, then they wont be as worried about being judged harshly by these friends. Future research exploring what leads to these beliefs about the impact of media images on male friends and romantic partners would be valuable. The path analysis examining the linkage between the third person perception and image related behaviors yielded interesting findings. For both the HK and U.S. women, behaviors related to physical appearance were partly triggered by the perceptions that their romantic partners are more susceptible to the influence of thin female models in ads than the respondent herself. This difference significantly predicted dieting among the U.S. respondents and approached significance in predicting exercising among the HK women. For both cultural groups, the gap in perceived influence between self and romantic partner significantly predicted willingness to consider plastic surgery. These findings suggest that the third-person effect can lead to behavioral outcomes beyond just censoring media messages. This provides important additional support for Davisons original claims regarding this theory. However, the results here suggest that behavioral effects are highly dependent on who the other is. Interestingly, in both cultures, romantic partner was seen as the other least influenced by advertising images. The perceived impact of seeing thin models was considered to be less for ones romantic partner than for the respondents themselves. However, the more ones romantic partner was seen as being influenced relative to ones-self, the more women engaged in behaviors to try to achieve an idealized look. This suggests that behavior is at least partially affected by an indirect influence

16

mediated by the belief that advertising affects ones romantic partner. Thus, the perceptions of media influence on different others are not equally important in influencing ones own behavior. Instead, perceived beliefs about others only exert influence when one highly identifies with or cares about the opinions of certain others (as in this context with romantic partners). While some differences existed, the findings here were remarkably consistent across cultural groups. Given previous research, this may be somewhat surprising. The lack of cultural difference could be due to the fact that we are examining private behaviors. In contrast to public behavior (such as express ones opinion publicly), private behaviors are less subject to normative pressure, which may be true regardless of cultural contexts. An alternative explanation could be a ceiling effect. The cultural ideal of female thinness and physical attractiveness have long been valued in Western society and internalized by Western women. When it comes to behaviors such as dieting, exercising and performing plastic surgery, women in Western culture have already been socialized to accept a willingness to engage in these behaviors in order to look good. The findings here may be due to the fact that Hong Kong women are catching up on these attitudes. The findings suggest that the use of the third-person effect theory to explain reactions to the media and their impact on behavior is applicable across cultures, at least in this context. Future studies are needed to see if it also holds up across other forms of consumptions behaviors and in a greater variety of cultural settings. Nonetheless, as media becomes more universal, it does suggest that beliefs and perceptions regarding how the media affects others may be an important determinant in how people will act. The findings here indicate that perceived effects on some types of others matter more to respondents. Future cross-cultural applications of the third person effect should attempt to determine if there are any cultural differences on who the important others are in various contexts.

17

Reference: Ahmad, S., Waller, G., and Verduyn, C. (1994). Eating attitudes among Asian schoolgirls: The role of perceived parental control. International Journal of Eating Disorders, 15, 91-97. Akan, G.E.& Grilo, C.M. (1995). Sociocultural influences on eating attitudes and behaviors, body image and psychological functioning: A comparison of African-American, Asian-American and Caucasian college women. International Journal of Eating Disorders, 18, 181-187. Arkoff, A., & Weaver, H. (1996). Body image and body dissatisfaction in Japanese-Americans. Journal of Social Psychology, 68, 323-330. Bond,R., & Smith, P.B.(1996). Culture and conformity: A meta-analysis of studies using Aschs (1952b, 1956) line judgment task. Psychological Bulletin, 119, 111-137. Brosius, H. B. & Engel. E (1996). The causes of third-person effects: unrealistic optimism, impersonal impact, or generalized negative attitudes towards media influence? International Journal of Public Opinion Research, 8, 142-161. Buhrich, N. (1981). Frequency and presentation of anorexia in Malaysia, Australian and New Zealand, Journal of Psychiatry, 115, 153-155. Cohen, J., Mutz, D., Price, V., & Gunther, A. (1988). Perceived impact of defamation: An experiment on third-person effects. Public Opinion Quarterly, 52, 161-173. Cohen, J & Davis, R.G. (1991). Third-person effects and the differential impact in negative political advertising. Journalism Quarterly, 68, 680-688. Cooper, P., Taylor, M., Cooper, Z., & Fairburn, C. (1987). The development and validation of the body shape questionnaire. International Journal of Eating Disorders, 6, 485-494. Davis, Cindy & Katzman, Melaine (1999). Perfection as acculturation: Psychological correlates of eating problems in Chinese male and female students living in the United States. International Journal of Eating Disorder, 25, 65-70. Davis, Cindy & Katzman, Melaine (1997). Charting new territory: Body esteem, weight satisfaction, depression, and self-esteem among Chinese males and females in Hong Kong. Sex Roles: A Journal of Research, 36, 449-459. Davis, Cindy & Katzman, Melaine (1996). Chinese men and women in the United States and Hong Kong: Body and self-esteem rating as a prelude to dieting and exercise. International Journal of Eating Disorder, 23, 99-102. Davis, C.& Katzman, M.(1998). Chinese men and women in the USA and Hong Kong: Body and self-esteem ratings as a prelude to dieting and exercise. International Journal of Eating Disorders,25, 99-102.

18

Davison, W.P.(1983). The third-person effect in communication. Public Opinion Quarterly, 47, 1-15. Dionne, M., Davis, C., Fox, J., & Gurevich, M. (1995). Feminist ideology as a predictor of body dissatisfaction in women. Sex Roles, 33:277-287. Dolan, B. (1991). Cross-cultural aspects of anorexia nervosa and bulimia: A review. International Journal of Eating Disorders, 10, 67-78. Duck, J.M. (1997). Gender identity and the perceived impact of commercial advertising. Working paper. Duck, J. M., Hogg, M.A. and Terry, D.J. (2000). The perceived impact of persuasive messages on us and them. In Terry, D.J. & Hogg, M.A (Eds.) Attitudes, Behaviors and Social Context: The Role of Norms and Group Membership. New Jersey: Lawrence Erlbaum AssociatesEndo, Y. (1992). Negative ideal self as a standard of self-esteem. Japanese Journal of Psychology, 63, 214-217. Eveland Jr., Amy I. Nathanson, Benjamin H. Detenber, Douglas M. McLeod.(1999). Rethinking the social distance corollary: perceived likelihood of exposure and the third-person perception, Communication Research, 26, 275-279. Fallon, A. (1990). Culture in the mirror: Sociocultural determinants of body image. In T.F.Cash & T. Pruzinsky (Eds.), Body Image, Development and Change (pp.80-109). New York: Guilford Press. Furnham, A., and Alibhai, N. (1983). Cross-cultural differences in the perception of female body shape. Psychological Medicine, 13, 829-837. Garner, D. M (1997). The 1997 Body Image Survey Results. Psychological Today, 30(1),3384. Gunther, A.G. (1995). Overrating the X-rating: The third-person perception and support for censorship of ponography. Journal of Communication, 45, 27-37. Gunther, A.G. (1991). What we think others think: courses and consequences in the third-person effect. Communication Research, 18, 355-372. Gunther, A. G. & Thorson, E. (1992). Perceived persuasive effects of product commercials and public service announcements: Third-person effects in new domains. Communication Rsearch,19, 574-596. Gunther, A.G. & Mundy, P. (1993) Biased optimism and the third-person effect. Journalism Quarterly, 70, 58-67. Gunther, A.G. & Hwa, A.P. (1996). Public perceptions of television influence and opinions

19

about censorship in Singapore. International Journal of Public Opinion Research, 8, 248-265. Harrison, Kristen (2000). The body electric: Thin-ideal media and eating disorders in adolescents. Journal of Communication, 51 (2), 119-143. Heine, Steven J. & Lehman, Darrin R. (1995). Cultural variation in unrealistic optimism: Does the West feel more invulnerable than the East? Journal of Personality and Social Psychology, 68, 595-607. Hill, A.J. and Bhati, K. (1995). Body shape perception and dieting in preadolescent British Asian girls: Links with eating disorders. International Journal of Eating Disorders, 17, 175-183. Hsu, F.L.K. (1948). Under the Ancestors Shadow: Chinese Culture and Personality. New York: Columbia University. Huang, L.C., & Harris, M.B.(1973). Conformity in Chinese and Americans: A field experiment. Journal of Cross-Cultural Psychology, 4, 427-434. Katzman, M. (1996). Asia on my mind: Are eating disorders a problem in Hong Kong? Eating Disorders: Journal of Treatment and Prevention, 3, 379-380. Kim, Heejung & Markus, Hazel R. (1999). Deviance or uniqueness, harmony or conformity: A cultural analysis. Journal of Personality and Social Psychology, 77, 785-800. Kim, Heejung & Markus, Hazel R. (1999). Deviance or uniqueness, harmony or conformity: A cultural analysis. Journal of Personality and Social Psychology, 77, 785-800. Kitayama, S., Markus, H.R., & Kurokawa, M. (1994). Cultural views of self and emotional experience: Does the nature of good feelings dependent on culture? Unpublished Manuscript, Kyoto University, Kyoto, Japan. Kowner, R., and Ogawa, T. (1993). The contrast effect of physical attractiveness in Japan, Journal of Psychology, 127, 51-64. Lake, Amelia J. , Staiger, Petra K. & Glowinski, Huguette (2000). Effect of western culture on womens attitudes to eating and perceptions of body shape. International Journal of Eating Disorder, 27, 83-89. Lasorsa, D. L. (1992). Policymakers and the Third-Person Effect, In J. D. Kennamer (Ed.), Public Opinion, The Press, and the Public Policy, New York, Praeger, 163-175. Lee, S. (1996). Reconsidering the status of anorexia nervosa as a western cultural bond syndrome. Social Science Medicine, 42, 21-34.

20

Lee, S., Leung, T/, Lee, A.M., Yu, H., and Leung, C.M. (1996). Body dissatisfaction among Chinese undergraduates and its implications for eating disorders in Hong Kong. International Journal of Eating Disorders, 20, 77-84. Lee, S., Ho, P., & Hsu, L.K.G.(1993). Fat phobic and non-fat phobic anorexia nervosa: A comparative study of 70 Chinese patients in Hong Kong. Psychological Medicine, 23, 999-1017. Markus, H. R. & Kitayama, S.(1991). Culture and the self: Implications of cognition, emotion and motivation. Psychological Review, 98, 224-253. Markus, Hazel R. & Kitayama, Shinobu (1991). Culture and the self: Implications of cognition, emotion and motivation. Psychological Review, 98, 224-253. Maynard, Michael L. & Taylor, Charles R. (1999). Girlish images across cultures: Japanese versus U.S. Seventeen magazine ads. Journal of Advertising, 28, 39-50. Matera, F. R. and Salwen, M. B. (1996). Unwieldy questions? Circuitous answers? Journalists as panelists in presidential election debates. Journal of Broadcasting & Electronic Media, 40, 309-317. McLeod, D. M., Eveland, W. P. Jr. & Nathanson, A. L. (1997). Support for censorship of violent and misogynic rap lyrics: An analysis of the third-person effect. Communication Research, 24, 153-174. McCout, J.M. & Waller, G. (1996). The influence of sociocultural factors on the eating psychology of Asian women in British society. European Eating Disorders Review, 4, 73-83. Meade, R.D. & Barnard, W.(1973). Conformity and anti-conformity among Americans and Chinese. Journal of Social Psychology, 89, 15-25. Miller, Mara (1993). Canons and the challenge of gender. The Monist, 76 (4), 477-493. Mukai, Takayo, Kambara, Akiko & Sasaki, Yuji (1998). Body dissatisfaction, need for social approval, and eating disturbance among Japanese and American College Women. Sex Roles: A Journal of Research, 39, pp.751-758. Perloff , R.M.(1989). Ego-involvement and the third person effect of television news coverage. Communication Research, 16, 236-262. Perloff, Richard M. (1999). The third-person effect: A critical review and synthesis. Media Psychology, 1, 353-378. Powell, A.D., and Kahn, A.S. (1995). Racial differences in womens desire to be thin. International Journal of Eating Disorders, 17, 191-195.

21

Price, V., Huang, L and Tewksbury, D. (1997). Third-person effects if news coverage: Orientations toward media. Journalism and Mass Communication Quarterly, 74, 525540. Reiger, Elizabeth, Touyz, Stephen W., Swain, Tony & Beumont, Peter J.V. (2001). Crosscultural research on anorexia nervosa: Assumptions regarding the role of body weight. International Journal of Eating Disorder, 29, 205-215. Rojas, H., Shah, D. V., & Faber, R. J. (1996). For the good of others: Censorship and the thirdperson effect. International Journal of Public Opinion Research, 8(2), 163-185. Rosenthal, D.A., & Feldman, S.S. (1990). The acculturation of Chinese immigrants: Perceived effects on family functioning of length of residence in two cultural contexts. Journal of Genetic psychology, 154, 495-514. Rucinski, D., & Salmon, C.T. (1990). The other as the vulnerable voter: A study of the thirdperson effect in the 1988 U.S. presidential campaign. International Journal of Public Opinion Research, 2, 345-368. Salwen, M.B. & Driscol, P.D. (1997). Consequences of third-person perception in support of press restrictions in the O.J. Simpson trial. Journal of Communication, 47, 60-78. Shah, D. V., Faber, R. J., & Youn, S. (1999). Susceptibility and severity: Perceptual Dimensions Underlying the Third-Person Effect, Communication Research, 26 (2), 240-267. Shibata, T., & Nobechi, M. (1991). The cognition of masculinity-femininity of ones own body in adolescence, Japanese Journal of Educational Psychology, 39, 40-46. Tiggeman, M. & Rothblum,E.D.(1988). Gender differences in social consequences of perceived overweight in the United States and Australia. Sex Roles, 18, 75-86. Wan, F. (2001). The Impact of Idealized Images in Advertising. Unpublished Dissertation, University of Minnesota. Wan, F. and William D. Wells (2001). How women and men perceive the thin female models in media differently: Another exploration of the third-person effect. Paper presented to Mass Communication Division at the 51st Annual Conference of International Communication Association in Washington, D.C., May 24-28, 2001. Youn, S., Faber, R. J., & Shah, D. V. (2000). Restricting Gambling Advertising and the ThirdPerson Effect. Psychology and Marketing.

22

Table 1 Culture difference in terms of image-related behaviors

US Body image disturbance Dieting Frequency of exercising Hours spent exercising Plastic surgery 2.98 (.88) 2.44 (1.23) 3.02 a (1.11) 3.64 a (2.23) 2.16 a (1.01) HK 3.05 (.74) 2.32 (1.14) 2.45 b (.84) 1.95 b (1.26) 1.66 b (.85) Between Culture Analysis F(1, 504)=.59, n.s. F(1, 505)=.83, n.s. F(1, 504)=26.97, p<.01 F(1, 500)=63.79, p<.01 F(1, 502)=23.66, p<.01

Notes: 1. 2. Numbers in parentheses are standard deviations. Mean scores with different superscript letters indicate a significant difference (p<.01) between US and HK

23

Table 2 Perceived impact of thin female models in advertising on self and others
US Perceived impact of thin female models on Self Female friends Self vs. Female friends Most women Self vs. Most women Male friends Self vs. Male friends Most Men Self vs. Most men Romantic Partner Self vs. Romantic partner
Notes: 1. Cells with italic numbers refer to mean difference between perceived impact on self and others within each culture. ** Within-culture comparison significant at p<.01. * Within-culture comparison significant at p<.05 For cross-cultural comparison, cells with different superscript letters indicate that the mean difference between the US and HK sample is statistically significant at p<.05.

HK

Between Culture Analysis F(1, 499)=.24, n.s. F(1, 498)=4.32, p<.05 F(1, 498)=1.20, n.s. F(1, 499)=6.18, p<.05 F(1, 499)=1.31, n.s. F(1, 499)=.10, n.s. F(1, 499)=.76, n.s. F(1, 499)=1.25, n.s. F(1, 499)=2.34, p<.10 F(1, 497)=13.33, p<.01 F(1, 497)=16.37, p<.01

3.34 3.83a -.50** 4.03a -.69** 3.19 .14** 3.43 -.09* 2.86a .47**a

3.28 3.69b -.41** 3.88b -.59** 3.21 .06 3.50 -.22** 3.15b .14*b

2.

24

Table 3 The third person gap by social distance

US Perceived impact of thin female models on Self Close others Self vs. close others Distant others Self vs. distant others
Notes: 1. Cells with italic numbers refer to mean difference between perceived impact on self and others within each culture. ** Within-culture comparison significant at p<.01. * Within-culture comparison significant at p<.05. For cross-cultural comparisons, cells with different superscript letters indicate that the mean difference between the US and HK samples is statistically significant at p<.05.

HK

Between Culture Analysis F(1, 499)=.24, n.s. F(1, 489)=.90, n.s. F(1, 489)=2.56, n.s. F(1, 489)=.50, n.s. F(1,489)=.10, n.s.

3.34 3.30 .04 3.73 -.40**

3.28 3.35 -.07 3.69 -.41**

2.

25

Table 4 The third person gap by gender identity

US Perceived impact of thin female models on Self Gender ingroup (female others) Self vs. Female others Gender outgroup (male others) Self vs. Male others HK Between Culture Analysis F(1, 499)=.24, n.s. F(1, 489)=5.90, p<.05 F(1, 489)=1.40, n.s. F(1, 489)=3.95, p<.05 F(1, 489)=5.35, p<.05

3.33 3.93a -.59** 3.16a .18* a

3.28 3.79b -.50** 3.29b .01 b

Notes:

.Cells with italic numbers refer to mean difference between perceived impact on self and others within each culture. ** Within-culture comparison significant at p<.01. * Within-culture comparison significant at p<.05. For cross-cultural comparisons, cells with different superscript letters indicate that the mean difference between the US and HK samples is statistically significant at p<.05.

26

Table 5 Coefficients of path analysis predicting dieting behavior

Body image disturbance (Direct effect) HK US Perceived impact of thin female models on Self Self vs. Female friends Self vs. Most women Self vs. Male friends Self vs. Most men Self vs. Romantic partner .45** .13 .01 -.05 -.04 .10 / 18.7 .61** .04 -.01 .16* -.14 -.06 / 36.2 Dieting behavior (Direct effect) HK -.04 .01 .09 -.07 .07 -.04 1.09** 48.4 US -.01 -.06 .21* .05 -.06 -.16* .93** 46.6 Dieting behavior (Total effect) HK .45 .16 .10 -.12 .02 .07 1.09 / US .56 -.03 .20 .20 -.19 -.22 .93 /

Body Disturbance Total R2 (%)

Notes:

1. The path coefficients reported in the cells are unstandardized estimates. . **p<.01. * p<.05.
2. 3. Model of fit measures for HK sample: CMIN=193.5 DF=7 (p<.01) NFI=.89 CFI=.89 RMSEA=.47 Model of fit measures for US sample: CMIN=591.8 DF=7 (p<.01) NFI=.89 CFI=.89 RMSEA=.47

27

Table 6 Coefficients of path analysis predicting exercising behavior

Body image disturbance (Direct effect) HK US Perceived impact of thin female models on Self Self vs. Female friends Self vs. Most women Self vs. Male friends Self vs. Most men Self vs. Romantic partner .45** .13 .004 -.05 -.04 .10 / 18.7 .61** .04 -.01 .16* -.14# -.07 / 36.2 Exercising behavior (Direct effect) HK US .53* -.23# -.17# -.24* .03 -.21# .09 7.6 .003 -.01 -.004 .001 .02 -.01 .02 .03 Exercising behavior (Total effect) HK US .57 -.22 -.16 -.25 .03 -.20 .09 / .02 -.01 .001 -.001 .01 -.01 .02 /

Body Disturbance Total R2 (%)

Note: 1. 2.

The path coefficients reported in the cells are unstandardized estimates. . **p<.01. * p<.05. Model of fit measures for HK sample: CMIN=198.1 DF=13 (p<.01) NFI=.90 CFI=.90 RMSEA=.30 Model of fit measures for US sample: CMIN=595.00 DF=13 (p<.01) NFI=.91 CFI=.91 RMSEA=.34

28

Table 7 Coefficients of path analysis predicting likelihood of plastic surgery

Body image disturbance (Direct effect) HK US Perceived impact of thin female models on Self Self vs. Female friends Self vs. Most women Self vs. Male friends Self vs. Most men Self vs. Romantic partner .45** .13 .01 -.05 -.04 .10 / 18.7 .61** .04 -.01 .16* -.14# -.07 / 36.3 Plastic surgery (Direct effect) HK -.03 -.09 .08 -.06 .28* -.39** .31** 14.4 US .15 .05 .09 -.10 .28* -.36** .19* 18.2 Plastic surgery (Total effect) HK .11 -.05 -.08 -.07 .27 -.36 .31 / US .27 .06 .09 -.07 .26 -.37 .19 /

Body Disturbance Total R2 (%)

Note: 1. 2.

The path coefficients reported in the cells are unstandardized estimates. . **p<.01. * p<.05. Model of fit measures for HK sample: CMIN=193.12 DF=7 (p<.01) NFI=.89 CFI=.89 RMSEA=.41 Model of fit measures for US sample: CMIN=590.84 DF=7 (p<.01) NFI=.89 CFI=.89 RMSEA=.44

29