|The Cultural Construction of Self-Enhancement:
An Examination of Group-Serving Biases
Steven J. Heine and Darrin R. Lehman
University of British Columbia
Please address correspondence to:
Steven J. Heine or Darrin R. Lehman
Department of Psychology
2136 West Mall
University of British Columbia
Vancouver, B.C. Canada V6T 1Z4
E-mail: firstname.lastname@example.org or email@example.com
In press: Journal of Personality and Social Psychology: ASC
Running Head: CULTURE AND GROUP-SERVING BIASES
Self-serving biases, found routinely with Western samples, have not been observed with Asian samples. Yet given the orientation toward individualism and collectivism in these 2 cultures, respectively, it is imperative to examine whether or not parallel differences emerge when the target of evaluation is the group. It may be that Asians show a group-serving bias parallel to the Western self-serving bias. In two studies, group-serving biases were compared across European-Canadians, Asian-Canadians, and Japanese. Study 1 revealed that Japanese evaluated a family member less positively than did both groups of Canadians. Study 2 replicated this pattern with students’ evaluations of their universities. The data suggest that cultural differences in enhancement biases are robust, generalizing to individuals’ evaluations of their groups.
The Cultural Construction of Self-Enhancement:
An Examination of Group-Serving Biases
Research in cultural psychology has underscored the notion that many psychological processes are not universal, that is, that culture plays an important role in influencing the ways that people think, feel, and view themselves. However, interpreting cultural differences obtained in psychological experiments is a nontrivial task. The favored interpretation of cultural psychologists is that the psychological process under study is shaped by the culture and hence varies across cultures (e.g., Markus & Kitayama, 1991b; Shweder & Bourne, 1984). Competing with this cultural view is a more prosaic interpretation: That is, the methodology of the experiment was more meaningful and relevant to one of the cultures, thereby indicating superficial and misleading cultural differences (Berry, 1969; Hui & Triandis, 1985; Triandis, 1978). Given that many social psychological paradigms have emerged within Western, “individualistic” cultures, we may indeed have stacked the deck against finding comparable results when studying individual-based psychological processes in Eastern, “collectivistic” cultures. Perhaps some failures to replicate common Western findings are attributable to the “imposed etic” of our individually-based methodologies (Berry, 1969).
One way of reducing this cultural bias in our Western methodologies is to explore targets of evaluation that are more meaningful within Eastern cultures. In contrast to the individualistic view of self common in Western cultures, Eastern cultures are characterized by a view of self that encompasses the important groups to which people belong (Markus & Kitayama, 1991b; Triandis, 1989). Accordingly, employing individuals’ groups as targets is a significant step in moving beyond the limitations inherent in individual-centered methodologies.
Cultural differences in self-enhancement
One domain of cross-cultural research that may be particularly susceptible to the potential imposed-etic effects of an individual-based methodology is self-enhancement. Self-enhancement research typically examines how individuals view themselves as individuals in unrealistically positive terms. The results of such research have revealed the various ways in which North Americans distort their views of themselves such that they appear, for example, overly competent and optimistic, and more in control (for reviews, see Blaine & Crocker, 1993; Greenwald, 1980; Taylor & Brown, 1988). As is the case in the bulk of research on the self, self-enhancement research has focused almost exclusively on Westerners.
Research in cross-cultural psychology and cultural psychology, however, has demonstrated that such unrealistically positive views of the self are not common in Eastern cultures, particularly in Japan. For example, Markus and Kitayama (1991a) found that Japanese do not demonstrate a false-uniqueness bias (i.e., the tendency to see oneself as uniquely talented or better than most people on a given dimension), although this effect has consistently been observed among North Americans (e.g., Campbell, 1986; Marks, 1984). Moreover, Japanese do not show this effect even when evaluating themselves on attributes which they view as most important to succeeding in their culture (Heine & Lehman, 1996c). A number of studies have demonstrated that Japanese fail to show self-serving attributional biases (for a review, see Kitayama, Takagi, & Matsumoto, 1995). This is despite the marked tendencies of North Americans to attribute success to internal factors and failure to external factors (for a review, see Zuckerman, 1979). In contrast to the robust unrealistic optimism effects found with North Americans (e.g., Perloff & Fetzer, 1986; Weinstein, 1980), Japanese do not show much evidence for optimism biases, and in some situations are actually unrealistically pessimistic (Heine & Lehman, 1995a). Furthermore, Kitayama, Markus, Matsumoto, and Norasakkunkit (in press) demonstrated that whereas Americans view everyday life in terms of opportunities for self-enhancement, Japanese view it in terms of opportunities for self-criticism. We know of no studies that have demonstrated consistent self-serving biases with Japanese, or, for that matter, with people from any Eastern culture.
One interpretation of these cultural differences in self-serving biases is that they reflect differences in motivations to view oneself as especially positive or competent (Heine & Lehman, 1995a, 1996a; Heine, Lehman, Kitayama, & Markus, 1996; Kitayama, Markus, & Lieberman, 1995; Markus, Kitayama, & Heiman, in press). Heine et al. (1996) argue that these cultural differences in self-enhancement are observed because Western (particularly North American) culture encourages people to think positively about themselves as a means to approach the culturally-defined ideals of independence and autonomy. They maintain that the construction of the typical North American’s identity as a meaningful cultural entity hinges on the identification and confirmation of positive internal attributes of the self. The elaboration of positive self-attributes and the denial or neglect of negative attributes are thus rewarded in North America by a cultural validation of the individual. Hence, enhancing one’s self-assessments enables North Americans to maintain the subjective feeling of being authenticated by their culture (cf., Greenwald, 1980; Taylor & Brown, 1988).
In contrast, Eastern (particularly, Japanese) culture encourages people to strive to fit in with their groups. The Japanese self is defined as a relational entity that is made meaningful in reference to the pertinent social relationships to which the self is part (Hamaguchi, 1985; Markus & Kitayama, 1991b; Nakamura, 1964). Therefore the construction of the typical Japanese individual’s identity as a meaningful cultural entity involves the validation of the individual’s social relationships by constantly seeking to identify and confirm shared expectations and norms. The identification of positive internal attributes of the self does not aid individuals in gaining a sense of belongingness. Possessing, let alone enhancing, a positive evaluation of the individual self ought not be a primary concern for Japanese. Hence, this cultural explanation argues that the very process of self-enhancement is not as meaningful to Japanese as North Americans.
A very different interpretation of the cultural differences in self-enhancement is that they result from a methodological artifact. Given that cross-cultural studies of self-enhancement typically employ a target of evaluation that is theoretically more meaningful to those from Western cultures (i.e., the individual self), it is reasonable to be concerned that studies demonstrating an absence of self-enhancing tendencies among Japanese are due to their evaluation of a target that is inconsequential to them. Perhaps people from both Western and Eastern cultures have similar tendencies to enhance themselves, yet they enhance the view of self most meaningful to them. That is, individualistic North Americans may be motivated to enhance their individual selves, whereas Japanese may be motivated to enhance their collective selves. This view suggests that past cultural differences in self-enhancement might reflect differences in content (the target of the evaluation) rather than differences in process (the motivation to see the self, or one’s group, in a positive light; Greenfield, in press; Fiske, Kitayama, Markus, & Nisbett, 1996). If this is the case, Japanese should exhibit group-serving biases at a comparable level to North Americans’ self-serving biases.
Perspectives reflecting the “differences in content” position have emerged in the psychoanalytic literature of the Japanese. For example, Roland (1988) maintains that there is a parallel between Americans being concerned with their “I-self regard” and Japanese being concerned with their “we-self regard.” Similarly, Johnson (1993) argues that Japanese are socialized to transform their feelings of personal narcissism and vanity into a sense of group-pride and collective narcissism. And De Vos (1973) contends that Japanese aspire towards succeeding as a group in contrast to Americans’ aspirations to succeed as individuals. These perspectives suggest fundamental similarities in underlying psychological processes between Japanese and North Americans, yet differences in the nature of the self that these processes sustain.
Culture and Group-Serving Biases
These two opposing views of the nature of previously detected cultural differences in self-serving biases make opposite predictions regarding cultural differences in group-serving biases. The “differences in process” view assumes that North American culture places greater importance on viewing oneself positively than does Japanese culture (Heine et al., 1996; Kitayama et al., in press). To the extent that one’s groups reflect upon the individual, this view predicts that North Americans would exhibit group-serving biases to a greater extent than Japanese. That considerable research conducted in the West shows that feelings of self-worth are promoted by positive evaluations of individuals’ groups (e.g., Brown, Collins, & Schmidt, 1988; Lemyre & Smith, 1985; Tajfel & Turner, 1986) suggests that group-serving biases may indeed serve to enhance the individual self as well.
In contrast, the “differences in content” view assumes the existence of similar motivations for self-enhancement across cultures. An implication of this view is that self-serving biases should be most prominent when individuals evaluate their most meaningful view of self. Given the collectivist, group-oriented nature of the Japanese (Hamaguchi, 1985; Kondo, 1987; Markus & Kitayama, 1991b), switching the target from the individual to the individual’s group should yield a view of self more meaningful to Japanese than to North Americans. To the extent that motivations to enhance the self are similar across cultures, Japanese should show more pronounced group-serving biases than North Americans.
Past research on group-serving biases
A review of studies of group-serving biases with Asian (albeit primarily non-Japanese) and North American participants is more in line with the “differences in process” view. Group-serving bias studies with Asian participants have yielded inconsistent results (Fletcher & Ward, 1988). We located three studies that found evidence for group-serving biases in people from Asian cultures. Taylor and Jaggi (1974) found that Hindus made internal attributions for other Hindus who performed socially desirable acts and external attributions for those who acted in socially undesirable ways. In contrast, when Hindus made attributions for the behavior of Muslims the reverse pattern was found. This study is problematic, however, because Muslims are a minority group of lower status than Hindus, and the reciprocal attributions from this group had not been solicited (Hewstone & Ward, 1985).
Hewstone, Bond, and Wan (1983) reported that Chinese students from two universities in Hong Kong made group-serving attributions favoring their respective universities. In a later study, somewhat limited evidence of group-serving biases in terms of sex-typed behaviors was found for Hong Kong Chinese students (Bond, Hewstone, Wan, & Chiu, 1985). However, the robustness of group-serving effects with Chinese participants is challenged. Bond et al. (1985) discovered that American students displayed a more pronounced group-serving bias for sex-typed behaviors than did Chinese, and a study of Chinese in Singapore by Hewstone and Ward (1985) found no evidence for group-serving biases at all. Hence, studies of group-serving biases with Chinese have not presented a clear picture.
Two recent studies with Japanese participants have failed to demonstrate group-serving biases. Kitayama, Palm, Masuda, Karasawa, and Carroll (1996) measured perceptions of vulnerability to earthquakes in two high-risk cities both in Japan and in the United States. Results showed that Americans from both cities demonstrated group-serving tendencies by stating that the other city was slightly less prepared for earthquakes than their own. In contrast, Japanese participants from both cities reported that their own city was significantly less prepared than the other. That is, Japanese showed unrealistic pessimism towards their own city.
Heine and Lehman (1995a) presented Japanese and Canadians with a list of possible negative future life events that specifically threatened the individual’s interpersonal network. They found that Japanese demonstrated even less unrealistic optimism (or more unrealistic pessimism) for these events than they did for events that simply threatened the individual. This pattern was not obtained for the Canadian participants. At least for studies exploring perceived risk, then, Japanese have not been found to exhibit group-serving biases. Japanese feel that their groups are at least as threatened, if not more so, than other people’s groups. Taken together, the past evidence for group-enhancing biases among Asians is not compelling.
In contrast, there is considerable evidence that North Americans do exhibit group-serving biases. North American research has shown that attributional biases, whereby individuals take credit for successes and explain away failures (see Zuckerman, 1979), generalize to the group level: For example, players and coaches of baseball and football teams make more attributional biases regarding their wins and losses than do sportswriters (Lau & Russell, 1980); individuals are at least as, if not more, self-serving when they interpret their spouses’ outcomes as when they interpret their own outcomes (Fincham, Beach, & Baucom, 1987); and people make group-serving attributions when their group succeeds and jointly absolve each other of responsibility when their group fails (Forsyth & Schlenker, 1977). Other self-serving biases have been shown to generalize from oneself to one’s friends. For example, individuals view positive personality traits to be more characteristic of their close friends than they are for others (Brown, 1986), and people believe that future negative life events are more likely to happen to the average other than to their close friends (Perloff & Fetzer, 1986).
A study by Crocker, Thompson, McGraw, and Ingerman (1987, Study 2) demonstrated group-serving biases among sorority members. They found that members of sororities evaluated rival sororities more negatively than their own, particularly when the participants were members of low-status sororities themselves. Brown et al. (1988) investigated group-serving biases in a minimal-groups paradigm. They demonstrated that individuals tended to evaluate products made by their in-groups more positively than those made by out-groups and that this effect was more pronounced after participants received negative feedback. Cialdini and Richardson (1980) found that students viewed their university more positively than a rival university, particularly following failure feedback. Cialdini et al. (1976) demonstrated that individuals affiliate themselves more with successful than unsuccessful groups. They argued that individuals feel good about themselves when they are associated with positively-viewed groups because they are able to bask in the reflected glory of the group’s success. Conversely, when their groups are viewed negatively, individuals are motivated to distance themselves from this reflected failure (e.g., Cialdini et al., 1976; Snyder, Lassegard, & Ford, 1986; Taylor & Mettee, 1971).
For North Americans, affiliating themselves with positively-viewed groups and holding unrealistically positive views of their groups appear to enhance their self-evaluations (Tajfel & Turner, 1986). Group memberships, even for “independently-oriented” North Americans, form an important part of their individual self-concepts (e.g., James, 1950/1890), thereby suggesting that individual self-evaluations are served by enhancing one’s group-evaluations. Research (Luhtanen & Crocker, 1992) showing that collective self-esteem (i.e., the extent to which one views one’s social groups positively) is correlated positively with global self-esteem corroborates this relation. Moreover, that some of the aforementioned group-enhancement studies (e.g., Brown et al., 1988; Cialdini et al., 1976; Cialdini & Richardson, 1980; Crocker et al., 1987) demonstrated increased group-serving tendencies when participants were confronted with threats to the self underscores the self-enhancing role that they play.
We sought to pit the “differences in process” view and the “differences in content” view against each other by investigating how individuals evaluate the people to whom they are connected and the groups to which they belong. In the first study, we measured how Canadians and Japanese evaluated a close family member relative to others and how they evaluated themselves relative to others. In the second study, we compared how Canadian and Japanese students evaluated their own and a rival university. In both studies we assessed collective self-esteem as another means to investigate how positively people evaluate their groups.
Because the Canadian data were collected in Vancouver, a city with a large Asian community, we partitioned the Canadian data for both studies into those of European and Asian ancestry. This third cultural group, “Asian-Canadians,” although heterogeneous in terms of country of origin and length of time/number of generations in Canada, falls between the groups of European-Canadians and Japanese in terms of exposure to Western cultural values (Heine & Lehman, 1996b). To the extent that culture mediates evaluations of one’s groups, we anticipated that Asian-Canadians would exhibit group-serving biases intermediate to those of European-Canadians and Japanese.
The Japanese sample consisted of students from Ritsumeikan University in Kyoto who completed the questionnaire packet 1 month before leaving Japan for a 7-month study-abroad program in Canada. Of the 93 students in the program, 82 (55 females and 27 males) agreed to participate in the study. It deserves mention that these students, in choosing to live abroad for an academic year, may be more “Western-oriented” than the average Japanese. If anything, however, this should reduce the likelihood of observing cross-cultural differences.
The Canadian sample consisted of 151 students enrolled in introductory psychology classes at the University of British Columbia (UBC). This sample was separated by ethnic background to further examine cultural differences. Forty-four (28 females and 16 males) declared themselves to be of Asian heritage and formed what we term the Asian-Canadian sample. Seventy-five (57 females and 18 males) declared themselves to be of European heritage and formed what we term the European-Canadian sample. The remaining 32 students were of varied ethnic backgrounds (e.g., mixed-ethnicities, Latin-American descent, African descent, etc.) and were not included in the analyses.
All data were collected by questionnaire. Following some demographic questions, participants were asked to write down the name of, and their relation to, the member of their family to whom they felt closest. They were then asked to indicate how close they were to this family member on a scale from 1 (not at all close) to 10 (extremely close). Next, they were asked to complete a section of the questionnaire to assess their degree of family-member serving biases: They were asked to estimate the percentage of the population of the same age and sex as their chosen family member that was better than this family member with respect to 10 traits. Five of these traits were chosen to be particularly meaningful to the independent view of self (attractive, interesting, independent, confident, and intelligent) and five traits were chosen to be particularly meaningful to the interdependent view of self (cooperative, loyal, considerate, hard-working, and dependable; Markus & Kitayama, 1991b).
Subsequently, participants were asked to complete a section of the questionnaire that assessed their degree of self-serving biases: They were asked to estimate the percentage of the population of the same age and sex as themselves that was better than them with respect to the same 10 traits. Last, participants completed Rosenberg’s (1965) 10-item Global Self-Esteem Scale and Luhtanen and Crocker’s (1992) Collective Self-Esteem (CSE) Scale. The CSE Scale is composed of four 4-item subscales: membership, private, public, and identity. Membership CSE refers to the extent to which individuals feel that they are worthy members of their social groups. Private CSE indicates how satisfied one is about being a member of her or his social group. Public CSE assesses how individuals feel that others view their social groups. Finally, Identity CSE measures the importance of an individual’s social group to her or his self-concept. Both Rosenberg’s and Luhtanen and Crocker’s self-esteem scales were completed using 5-point Likert scales from 1 (Strongly Disagree) to 5 (Strongly Agree).
The materials were originally produced in English and then translated into Japanese. Then, after an independent translator back-translated the Japanese version into English, three translators discussed and resolved any inconsistencies between the versions.
Results and Discussion
Comparability of the Samples
There was a significant difference in the average ages of the three samples, F(2, 198) = 5.44, p < .01, with post-hoc comparisons (Tukey’s HSD for unequal ns for all post-hoc comparisons in Studies 1) revealing that the European-Canadian sample (M = 21.2 years) was significantly older than either the Asian-Canadian sample (M = 19.9 years) or the Japanese sample (M = 20.0 years). However, given that correlational analyses revealed that age did not significantly relate to any of the dependent variables (all rs between -.10 and .03, ns), it is unlikely that this age difference confounded the cultural comparisons. Each of the samples was predominantly female; 67% of Japanese, 64% of Asian-Canadians, and 76% of European-Canadians were female. Sex was included as a factor in all analyses; however, for the sake of brevity, main effects for sex, and sex culture interactions, are reported only when they reached conventional levels of significance.
Self-Serving and Family-Serving Biases
Self-serving biases (SSBs) and family-serving biases (FSBs) were operationalized as the discrepancy between participants’ overall estimates and what would be expected if participants answered accurately. It is difficult, if not impossible, to ascertain precisely what entails “accurate” responding by individuals. If we can assume a normal distribution of the population, and that our samples are not substantially different from the general population at large with respect to the traits, then, on average, there should be approximately 50% of the population better than the research participants and their family members. We realize that these assumptions are debatable. Samples consisting of university students may reasonably be argued to be “better than average” on certain traits. However, lacking more objective criteria, we adopted the 50% benchmark as the touchstone from which to describe the magnitude of enhancement biases. This operationalization of “bias” exists simply for illustrative purposes. Our primary concern is with the cultural differences in the magnitude of self-enhancing biases, not the absolute magnitude of the biases per se. The lack of precision of the comparative benchmark does not compromise the cultural comparisons. Employing this 50% benchmark, SSBs and FSBs were calculated by subtracting each of the participant’s population estimates for the 10 traits from 50%. Any estimate that was significantly less than 50% across the entire sample was termed an enhancement bias.
The 5 interdependent and the 5 independent traits were analyzed separately to assess whether participants responded to them differently. Reliability tests for each set of 5 traits were conducted for both measures of SSBs and FSBs. Cronbach’s alphas were .81 and .79 for interdependent and independent FSB traits, and .84 and .86 for interdependent and independent SSB traits, respectively. This indicates that, within each set of traits, participants tended to estimate that roughly the same percentage of people were better than either them or their family member regardless of the traits under consideration. Therefore, the 5 traits were averaged for both sets of independent and interdependent traits for both measures of SSBs and FSBs to give an overall estimate of the degree of bias.
Self-Serving Biases. First, examining the independent traits for the SSB measure, a significant Sex Culture interaction emerged, F(2, 193) = 3.29, p < .04. Simple-effects analyses revealed that Japanese women estimated that significantly more people were better than them (52.5%) than did Japanese men (42.1%; F[1, 79] = 8.09, p < .01). Japanese women did not show evidence of SSBs for the independent traits as their estimates were not significantly different from 50%, t < 1. In contrast, Japanese men showed a significant overall SSB, t(25) = 2.33, p < .05. No sex differences emerged for either Asian-Canadians, F < 1, or European-Canadians, F(1, 73) = 2.97, p < .09, and thus males and females were examined together (see Table 1). Both Asian- and European-Canadians’ self-estimates were significantly less than 50%, t(42) = 5.80 and t(74) = 13.12, respectively, both ps < .001, across the 5 traits, thereby exhibiting a pronounced overall SSB for the independent traits.
A highly significant main effect for culture emerged, F(2, 193) = 27.65, p < .001, which post-hoc comparisons revealed was the result of both Asian- and European-Canadians demonstrating a more pronounced self-serving bias than Japanese, and Asian-Canadians falling squarely between. Hence, a clear relation between SSBs and culture was observed for the independent traits.