Cross-observer agreement and self-concept consistency across cultures: Integrating trait and cultural psychology perspectives

Journal of Research in Personality 47 (2013) 78–89

Contents lists available at SciVerse ScienceDirect

Journal of Research in Personality journal homepage: www.elsevier.com/locate/jrp

Cross-observer agreement and self-concept consistency across cultures: Integrating trait and cultural psychology perspectives Marcia S. Katigbak a, A. Timothy Church a,⇑, Juan M. Alvarez a, Congcong Wang b, José de Jesús Vargas-Flores c, Joselina Ibáñez-Reyes c, Rina Mazuera Arias d, Brigida Carolina Rincon d, Lei Wang e, Fernando A. Ortiz f a

Department of Educational Leadership and Counseling Psychology, Washington State University, United States Department of Teaching and Learning, Washington State University, United States Iztacala National School of Professional Studies, National Autonomous University of Mexico, Mexico d Catholic University of Táchira, Venezuela e Department of Psychology, Peking University, People’s Republic of China f Counseling Center, Gonzaga University, United States b c

a r t i c l e

i n f o

Article history: Available online 29 September 2012 Keywords: Culture Cross-observer agreement Social Relations Model Self-construals Dialecticism Tightness-looseness

a b s t r a c t The Social Relations Model was used to compare cross-observer agreement and self-concept consistency in trait judgments in the United States, Mexico, Venezuela, and China. Target participants recruited friends and family members who comprised separate friend and family round-robin groups. Consistent with trait perspectives, in all cultures, (a) consensus and self-other agreement in trait judgments was found for most traits within the friend and family contexts, (b) across-context consensus was observed for at least some traits, and (c) self-concept consistency across contexts was substantial. Consistent with cultural psychology perspectives, consensus was generally greatest in the United States, intermediate in Mexico and Venezuela, and lowest in China. However, measures of dialecticism, self-construals, and cultural tightness failed to account for the cultural differences. Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction In a series of studies we have sought to integrate trait and cultural psychology perspectives, two dominant approaches in the study of culture and personality (Church, 2000, 2009). While some theorists have questioned their compatibility (Markus & Kitayama, 1998; Shweder, 1991), there is support for both perspectives (Church, 2000; Heine, 2012; McCrae, 2000; Spencer-Rodgers, Williams, & Peng, 2010b). This suggests that an integrated approach will lead to a more comprehensive understanding of the relationship between culture and personality. In the present study, we applied this integrated approach in the study of crossobserver agreement and self-concept consistency in personality judgments, a topic of considerable importance for trait psychology and interpersonal perception generally (Kenny, 1994; Kenny & West, 2010; McCrae et al., 2004). Researchers have differentiated two types of cross-observer agreement: consensus and self-other agreement (Kenny & West, 2010). Consensus refers to agreement between two or more raters ⇑ Corresponding author. Address: Department of Educational Leadership and Counseling Psychology, Cleveland Hall, Washington State University, Pullman, WA 99164-2136, United States. Fax: +1 509 335 6961. E-mail address: [email protected] (A. Timothy Church). 0092-6566/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jrp.2012.09.003

of a third person target, whereas self-other agreement refers to agreement between the target and other raters. In the present study, self-concept consistency refers to consistency in self-ratings across contexts (Boucher, 2011; Church et al., in press). Consensus, self-other agreement, and self-concept consistency all imply a degree of consistency in trait-relevant behavior and thus the existence of stable traits. At the same time, cultural psychology theory suggests that cross-observer agreement and self-concept consistency will be lower in less Western or individualistic cultures, calling into question the relative importance of traits in these cultures. In the present study, we compared cross-observer agreement and self-concept consistency in four cultures while testing hypotheses based on an integration of trait and cultural psychology.

1.1. Trait and cultural psychology perspectives on agreement and consistency From the perspective of trait psychology, we expect a degree of agreement in personality trait judgments between observers in all cultures. The existence of heritable and stable traits (Jang, McCrae, Angleitner, Riemann, & Livesley, 1998; Terracciano, Costa, & McCrae, 2006), combined with an ecological-realist perspective on person perception (e.g., Baron & Misovich, 1993; Funder,

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89

1995), leads to the prediction that traits will be perceived with some accuracy by self and others in all cultures, leading to agreement in trait judgments. In contrast, cultural psychologists—who emphasize the ‘‘mutual constitution’’ of culture and personality—have proposed various cultural dimensions that might underlie cultural variation in cross-observer agreement and self-concept consistency. One theoretical perspective distinguishes independent versus interdependent self-construals, which are thought to be more prevalent in individualistic and collectivistic cultures, respectively (Markus & Kitayama, 1991; Suh, 2002). People with independent selfconstruals are believed to have a greater need to express their traits and act in accordance with personal preferences and goals. This should lead to greater behavioral consistency and hence greater cross-observer agreement and self-concept consistency. In contrast, for people with interdependent self-construals, behaviors are more responsive to social obligations and situational contexts, reducing behavioral consistency and hence crossobserver agreement and self-concept consistency (Heine, 2001; Markus & Kitayama, 1998). A second cultural psychology perspective attributes lower behavioral and self-concept consistency, and hence lower crossobserver agreement, to East Asian dialecticism (Boucher, 2011; Spencer-Rodgers et al., 2010b). Dialecticism is a system of thought rooted in Eastern philosophical traditions and characterized by acceptance of contradiction, expectations of complexity and change, and holistic thinking (Peng & Nisbett, 1999). One aspect of dialecticism is a greater tolerance and expectation of cognitive and behavioral change, which could lead to reduced cross-observer agreement and self-concept consistency in trait judgments, at least across different situational contexts. A third theoretical framework addresses the cultural dimension of tightness versus looseness. As defined by Gelfand, Nishii, and Raver (2006), cultural tightness refers to ‘‘the strength of social norms and the degree of sanctioning within societies’’ (p. 1226). Implicit in this framework is the expectation of reduced behavioral consistency in tight cultures where situational constraints on behavior are greater (Gelfand et al., 2011). Again, the predicted result would be reduced cross-observer agreement and self-concept consistency in tight cultures, as compared to loose cultures. It is possible that cross-observer agreement and self-concept consistency will be lower in collectivistic, dialectical, or tight cultures even within particular contexts (e.g., with different family members) to the extent that one interacts differently with different individuals within this context. In any case, if cultural psychology theory is correct, we should expect lower cross-observer agreement and self-concept consistency in trait ratings across interpersonal contexts in these cultures (e.g., with friends versus family members).

1.2. Available cross-cultural evidence 1.2.1. Cross-observer agreement Studies in the United States have generally found moderate cross-observer agreement when judges are well-acquainted (Biesanz, West, & Millevoi, 2007; Connelly & Ones, 2010; Funder & Colvin, 1988; Funder, Kolar, & Blackman, 1995; Kenny, 1994; Kenny & West, 2010; Malloy & Albright, 1990; Watson, Hubbard, & Wiese, 2000). Indeed, some agreement is found even when judges have had little if any prior interaction with each other, particularly for the trait of extraversion (Albright, Kenny, & Malloy, 1988; Ambady & Rosenthal, 1992; Borkenau & Liebler, 1992; Kenny, 1994). Agreement across contexts has rarely been investigated, but Malloy, Albright, Kenny, Agatstein, and Winquist (1997) found weaker agreement across friend, family, and cowor-

79

ker contexts than within contexts, indicating that trait perceptions are somewhat context dependent even in Western cultures. Cross-cultural studies of cross-observer agreement are fairly rare, and we could identify only one cross-cultural study that examined agreement across interpersonal contexts. Malloy, Albright, Díaz-Loving, Dong, and Lee (2004) investigated wellacquainted college students and found that agreement in trait ratings, both within and across social contexts (i.e., friends vs. family), was higher in Mexican students (and similar to American levels) than in Chinese students. Malloy et al. concluded that Mexicans, because of their cultural norm of simpatía, have general prescriptions for social behavior that lead to trait consistency across friend and family contexts. Both Mexicans and Americans, they argue, are socialized to evaluate others in terms of traits. In contrast, Chinese have dyadic prescriptions for social behavior that derive from Confucian traditions of filial piety, leading to behavior that is dyad specific. Malloy et al.’s results would support cultural psychology predictions regarding individualism-collectivism only if Mexican culture is relatively individualistic. Accordingly, Malloy et al. argued that one cannot predict the extent of cross-observer agreement from the individualism-collectivism dimension. McCrae et al. (2004) also questioned whether there are cultural differences in agreement between individualistic and collectivistic cultures. They summarized existing cross-observer agreement data with the NEO Personality Inventory and other Big Five measures and concluded that agreement is as high in less individualistic cultures (e.g., China, Russia, Korean students in the United States) as in North America. Limitations of these data included the use of single rather than multiple raters and the fact that the Asian findings involved self/spouse agreement, which tends to be higher than agreement between other judges (Connelly & Ones, 2010). More consistent with cultural psychology perspectives were studies by Heine and Renshaw (2002), who found higher self-other agreement among well-acquainted students in the United States than in Japan, and Church et al. (2006b), who found somewhat lower consensus and self-other agreement in Filipino students than in European American, Asian American, and Mexican students. In a zero-acquaintance study, Albright et al. (1997) found significant consensus among Chinese students, but noted that the level of consensus was lower than typically found in the United States. In contrast, Yik, Bond, and Paulhus (1998) found high levels of consensus in a Hong Kong sample, perhaps because of greater Western influences. 1.2.2. Self-concept consistency A few studies have investigated cultural differences in selfconcept consistency, generally by correlating participants’ trait self-ratings in different roles or relationships. Consistent with trait perspectives, these studies have reported substantial selfconcept consistency across cultures, as well as cultural differences that support cultural psychology perspectives. Suh (2002) attributed the reduced self-concept consistency of Koreans, as compared to Americans, to differences in self-construals. English and Chen (2007, 2011), who found lower self-concept consistency in Asian Americans than in European Americans, and Boucher (2011), who found lower self-concept consistency in Chinese than Americans, attributed the cultural differences to dialecticism. Finally, Church and colleagues (Church et al., 2008, in press) found similar levels of self-concept consistency in Americans, Australians, Mexicans, Venezuelans, Filipinos, Malaysians, and Chinese, whereas Japanese exhibited significantly lower consistency than the cultural groups. In summary, there is some evidence of both cross-observer agreement and self-concept consistency across cultures, which is consistent with trait perspectives. Evidence of cultural differences is more mixed and has most often involved comparisons with se-

80

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89

lected Asian cultures. None of the cross-observer agreement studies directly measured the cultural dimensions hypothesized to account for the cultural differences, such as individualism-collectivism (or associated self-construals), dialecticism, or tightness. In addition, only Malloy et al. (2004) investigated cross-observer agreement both within and across interpersonal contexts and the study included only two cultures. Finally, only two cross-cultural studies applied the Social Relations Model (SRM; Kenny, 1994) to quantify cross-observer agreement (Albright et al., 1997; Malloy et al., 2004). Thus, there is a strong need for additional studies of cross-observer agreement and self-concept consistency in a diversity of cultures applying the SRM. 1.3. Social relations model In most studies, consensus has been quantified by correlating the judgments of two or more perceivers of the same target individual. To quantify self-other agreement, self-ratings are typically correlated with the average of two or more other raters of the target. However, these trait judgments contain multiple components, which can confound interpretation of the agreement indices. For example, trait ratings can reflect individual differences between the rated targets (i.e., target effects), but also individual differences in how particular raters judge all individuals (i.e., perceiver effects). To partition the various components that can impact trait judgments, Kenny and colleagues developed the Social Relations Model (SRM), which estimates these components in research designs that involve reciprocal trait judgments (Kenny, 1994; Kenny & La Voie, 1984). Such judgments can be obtained, for example, using round-robin designs, in which each member of a group rates each other member of the group. In the SRM, perceiver i’s rating of target j on a particular trait can be decomposed as follows:

X ij ¼ l þ ai þ bj þ cij þ eij

ð1Þ

where l is the overall mean level of ratings on the trait in a roundrobin group; ai is the mean rating that person i gives to members of the round-robin group (i.e., the perceiver effect); bj is the mean rating that person j receives (i.e., the target effect); cij is the rating deviation due to person i’s unique perception of person j (i.e., the relationship effect); and eij represents error of measurement (Kenny, 1994; Paulhus & Reynolds, 1995). To the extent that particular targets are rated similarly by different perceivers (and discriminated from each other), the variance of the target effects (r2b ) will be large. Thus, in the SRM, the target variance provides an estimate of consensus within a group that is not confounded by the other components of trait judgments (Kenny, Albright, Malloy, & Kashy, 1994). Similarly, self-other agreement can be quantified as the correlation between the target effects for each individual and their self-ratings. When particular targets are rated in two or more interpersonal contexts, agreement in trait judgments across these contexts can be quantified by correlating the within-group target effects (bs) obtained for individuals in each context (Malloy et al., 1997). Finally, if self-ratings are provided by round-robin participants in two or more contexts, the self-ratings in different contexts can be correlated to obtain an index of self-concept consistency. 1.4. Overview of the present study The purpose of this study was to compare cross-observer agreement and self-concept consistency across cultures while testing hypotheses based on an integration of trait and cultural psychology. We applied the SRM and a round-robin design to estimate cross-observer agreement both within and across friend and family groups in four cultures, the United States, Mexico, Venezuela, and China. We used the SOREMO program (Kenny, 2007) to estimate

the SRM target effects and variances used to quantify consensus and self-other agreement. Self-ratings provided in the friend and family contexts were correlated to quantify self-concept consistency. Finally, we examined the ability of individualism-collectivism, dialecticism, and cultural tightness to account for cultural differences in levels of agreement. Although it would be difficult, if not impossible, to sample cultures that represent all combinations of these three cultural dimensions, our results in a previous study showed that these four cultures differed along these three dimensions (Church et al., in press). For both dialecticism and tightness, China averaged higher than the United States and Mexico, who, in turn, averaged higher than Venezuela. China and Venezuela averaged higher than the United States and Mexico in collective self-construal. China averaged lowest in independent self-construal, as expected, but both Mexico and Venezuela averaged higher in independent self-construal than the United States. The results for independent self-construal were inconsistent with the traditional view of Mexico and Venezuela as relatively collectivistic, but the Mexican results are consistent with some previous studies, which suggest that Mexican culture may be relatively individualistic (Church et al., 2006b; Morling & Lamoreaux, 2008). Drawing on trait psychology perspectives, we hypothesized that in all four cultures cross-observer agreement would be observed both within and across friend and family contexts and that substantial self-concept consistency across contexts would be observed. Drawing on cultural psychology perspectives and some previous findings (Albright et al., 1997; Boucher, 2011; Heine & Renshaw, 2002), we predicted lower consensus and self-other agreement in China than in the United States, particularly across friend and family contexts. Given Malloy et al.’s (2004) finding of greater agreement in Mexico than China, and the uncertain status of Mexico and Venezuela along the individualism-collectivism dimension, we did not make specific predictions of cultural differences involving Mexico and Venezuela. However, by including Mexican and Chinese samples, we sought to replicate Malloy et al.’s findings and extend them to additional cultures. Although cultural psychology theory predicts cultural differences in selfconcept consistency, the few available studies suggest that reduced self-concept consistency (relative to American samples) may be limited to selected Asian cultures such as Japan (Church et al., 2008) and Korea (Suh, 2002), while the results for China are mixed (Boucher, 2011; Church et al., in press). Thus, we did not propose an a priori hypotheses regarding cultural differences in selfconcept consistency. Finally, consistent with cultural psychology perspectives, we hypothesized that some of the cultural differences in cross-observer agreement would be mediated by differences in individualism-collectivism, dialecticism, or cultural tightness. 2. Method 2.1. Participants Target participants were college students who were asked to recruit three friends who knew the target and the other friends well, plus three family members who knew the target and the other family members well. Complete friend and family round-robin groups were not obtained for all target participants due to the demanding nature of round-robin designs. Usable round-robin data was obtained from 66% to 71% of the targets in the United States, Venezuela, and China, and from 83% of the targets in Mexico. Reasons for nonusable data included nonreturn of surveys by one or more friends or family members despite reminders; recruited friends reporting that they did not know each other well; and overlap between different friend groups. In the United States and Mexico, all round-robin groups were comprised of four members. For a few target participants in Venezuela and China,

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89

complete round-robin groups were obtained for friend or family groups but not both. These were retained to maximize sample sizes for the within-context analyses. For Venezuela and China, we also retained some round-robin groups with three rather than four members, which is acceptable because SOREMO analyses revealed no dyadic reciprocity effects in the ratings (Kenny, 1994, p. 34).1 In Venezuela, 8 of 45 friend groups and 7 of 41 family groups contained three members. In China, 9 of 33 friend groups and 9 of 35 family groups contained three members. Follow-up SOREMO analyses revealed only trivial changes in the target variance estimates when the three-person round-robin groups were excluded. In the United States, respondents were 57 target participants (22 male, 35 female; M age = 20.53, SD = 1.34) at Washington State University, 171 of their friends (72 male, 99 female; M age = 20.91, SD = 4.51), and 171 of their family members (72 male, 99 female; M age = 41.98, SD = 16. 07), resulting in 399 total respondents. These respondents constituted 57 friend and 57 family groups. In Mexico, the respondents were 59 target participants (26 male, 33 female; M age = 20.07, SD = 2.12) at the National Autonomous University of Mexico at Iztacala, 177 of their friends (73 male, 104 female; M age = 20.72, SD = 3.40), and 177 of their family members (72 male, 105 female; M age = 39.06, SD = 13. 30), resulting in 413 total respondents. These respondents constituted 59 friend and 59 family groups. In Venezuela, respondents were 47 target participants (17 male, 30 female; M age = 20.60, SD = 3.62) at the Catholic University of Táchira, 127 of their friends (43 male, 84 female; M age = 22.27, SD = 5.79), and 116 of their family members (42 male, 74 female; M age = 33.62, SD = 11.79), resulting in 290 total respondents. These respondents constituted 45 friend groups and 41 family groups. In China, participants were 36 target participants (12 male, 24 female; M age = 21.61, SD = 1.66) at Peking University in Beijing, 90 of their friends (42 male, 48 female; M age = 21.90, SD = 2.20), and 96 of their family members (44 male, 52 female; M age = 36.97, SD = 12.20), resulting in 222 total respondents. These respondents constituted 33 friend groups and 35 family groups. The number of round-robin groups and total sample sizes in each culture compare very favorably to the numbers in previous studies (Kenny, 1994, Appendix A; Kenny & West, 2010). Friends and family members rated how well they knew each other on a 5-point scale (not at all, slightly, fairly well, very well, extremely well). In an ANOVA, we found cultural mean differences in acquaintance ratings for friends (F[3, 2211] = 26.16, p < .01). Follow-up Tukey tests indicated that American (M = 4.01) and Venezuelan (M = 3.86) friends reported knowing each other better than did the Mexican (M = 3.68) and Chinese (M = 3.53) friends. On average, however, friends in all four cultures indicated that they knew each other fairly well to very well. There were also cultural differences in acquaintance ratings for family members (F[3, 2194] = 78.66, p < .01). Follow-up Tukey tests indicated that American family members (M = 4.57) reported knowing each other better than Venezuelan family members (M = 4.33), who, in turn, knew each other better than Chinese (M = 3.98) and Mexican (M = 3.95) family members. Again, however, family members in all four cultures indicated that they knew each other very well to extremely well, on average. 2.2. Instruments 2.2.1. Translation and measurement equivalence The backtranslation method was used to translate all instruments from English into Spanish and Chinese. Minor modifications 1 Dyadic reciprocity is present when person i’s unique judgments of person j are correlated with person j’s unique judgments of person i. The absence of dyadic reciprocity in the present study is consistent with Kenny’s (1994) conclusion that ‘‘there is little or no evidence of reciprocity of perceptions in trait judgments’’ (p. 110).

81

to the initial translations were made as necessary. Church et al. (in press) conducted multigroup confirmatory factor analysis to test the cross-cultural measurement equivalence of the cultural measures and a longer version of the trait-rating form used in the present study. Acceptable metric (loading) invariance and partial scalar (intercept) invariance were demonstrated, indicating that the instruments can be applied in the four cultures, but that mean comparisons with the cultural dimensions should be interpreted with some caution (see Church et al., in press for details). 2.2.2. Trait rating form Participants rated themselves and others on 15 trait adjectives. We selected three trait adjectives, including some reverse-keyed (r) traits, for each of the Big Five dimensions, as follows: for Extraversion, energetic, extraverted, and quiet (r); for Agreeableness, kind, helpful, and selfish (r); for Conscientiousness, disciplined, organized, and lazy (r); for Emotional Stability, relaxed, calm, and moody (r); for Openness to Experience, imaginative, intelligent, and shallow (r) (Goldberg, 1992; Saucier, 1994). Participants in the friend groups (which included the target participants) first rated how accurately each trait describes them when they are with their friends and then rated each of the other members of their friend group on the 15 traits, one friend at a time. Ratings were made on a five-point scale that ranged from 1 = Not at all accurate or descriptive to 5 = Extremely accurate or descriptive. The analogous procedure was used in the family groups. Only the target participants were members of both a friend and family group. They provided separate self-ratings describing their traits when interacting with their friends and when interacting with their family members. Thus, target participants provided a total of 120 trait ratings, whereas friends and family members who were not target participants provided a total of 60 trait ratings. 2.2.3. Measures of cultural dimensions 2.2.3.1. Individualism-Collectivism. Only the target participants filled out the cultural measures. To assess individualism-collectivism we administered 14 items from Singelis’ (1994) Independent Self-construal scale and 13 collectivism items from Kashima and Hardie’s (2000) RIC Self-aspects scale and Yamaguchi’s (1994) Collectivism scale. Participants indicated their level of agreement using a 6-point scale that ranged from 1 = strongly disagree to 6 = strongly agree. Across the four cultures, a reliabilities ranged from .69 to .81 for the Independent Self-construal scale and .61 to .90 for the Collectivism measure. 2.2.3.2. Dialecticism. We administered 20 items from the Dialectical Self Scale (DSS; Spencer-Rodgers et al., 2010a), which included 14 items from the Abbreviated DSS scale, plus 6 additional items from the original 32-item measure. Items assess acceptance of contradiction (e.g., believing that opposing sides of an argument can both be correct), tolerance of cognitive change (e.g., being willing to change one’s beliefs), and willingness to adapt one’s behavior to fit circumstances. Participants rated their level of agreement on a 7-point scale that ranged from 1 = strongly disagree to 7 = strongly agree. Alpha reliabilities ranged from .68 to .76 across the four cultures. 2.2.3.3. Cultural Tightness-Looseness. Gelfand et al. (2011) constructed a 6-item measure to assess participants’ perceptions of the strength of social norms and the degree of sanctioning of behavior within their country. We wrote nine new items to improve reliability and the balance of positive- and reverse-keyed items. We deleted one item with low item-total correlations, resulting in a reliabilities that ranged from .63 to .78 across the four cultures.

82

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89

2.3. Procedure Target participants were recruited in classes or research participant pools. Target participants provided the research team in each country with the first name and e-mail address(es) of three friends and three family members. In the United States, Venezuela, and China, a member of the U.S.-based research team with fluency in the relevant language then e-mailed an invitation to participants and a link to the relevant on-line survey. Each participant received a tailored survey containing the first names of the individuals they were expected to rate. Completion of the on-line surveys was monitored daily and participants were sometimes contacted by e-mail, for example, to finish an incomplete survey. Periodic reminders were sent via e-mail, as necessary. Because of more limited internet access in Mexico, particularly for family members, paper surveys were used. Each target participant was given a packet containing the tailored surveys for themselves and for their friends and family members. Target participants were given the responsibility to distribute the paper surveys and return envelopes to each friend and family member. To ensure confidentiality, friends and family members were asked to return the forms directly to the Mexican research team or, if returned via the target participant, to place their survey in a sealed envelope with their initials across the seal. In all four countries, local members of the research team also followed up with target participants, as necessary, in order to obtain as many complete round-robin data sets as possible. Payments were made to target participants from whom complete round-robin data was obtained. Payments were determined in consultation with country collaborators and reflected the varying costs of living in the respective countries: $50 in the United States, 525 pesos in Mexico, 175 bolivares in Venezuela, and 200 yuan in China.

3. Results 3.1. Within-context agreement 3.1.1. Consensus Drawing on trait perspectives, we hypothesized that cross-observer agreement would be observed within the friend and family contexts in all cultures. Drawing on cultural psychology perspectives, we predicted lower agreement in China than in the United States. For each of the 15 specific traits, Table 1 shows the proportion of target variance—the SRM index of consensus—within the friend and family contexts. Statistically significant consensus (p < .05) was found for all of the specific traits in the U.S. and Mexico and for the majority of traits in Venezuela and China.2 These relative target variances can be interpreted as the predicted correlations between pairs of different raters judging the same targets (Kenny, 1994, p. 31). The mean values in the United States and Mexico exceeded the 15% value described by Kenny (1994, p. 84) as typical in SRM studies and the mean values in Venezuela and China are close to this typical value. To formally compare the cultures, we conducted MANOVAs with the absolute target variances for the 15 traits as dependent variables. The main effect for culture was statistically significant in the family context (Wilks’ Lambda = .57, F[45, 518] = 2.38, 2 In SOREMO, target variance is calculated by estimating the absolute target variance within each round-robin group and pooling across all groups. The target variances are theoretical estimates of what the variance would be if there were an infinite number of raters (Kenny, 1994, p. 31). Significance tests are one-tailed tests comparing the absolute variances across groups to zero with group as the unit of analysis. SOREMO then divides the pooled absolute variances by all other sources of variance in the model (i.e., perceiver, relationship, and error variance) to obtain the relative variances shown in Table 1.

p < .001) but not in the friend context (Wilks’ Lambda = .75, F[45, 524] = 1.18, p = .21). Follow-up ANOVAs in the family context revealed significant cultural differences of modest to moderate size for 12 of the 15 traits (g2 range = .04 to .12). In Table 1, mean target variances that exhibited cultural differences in Tukey tests are labeled with subscripts. Target variances that share a subscript were not significantly different (p > .01). Consistent with our hypothesis, the relative target variances were greater in the US than in China for 14 of the 15 specific traits; the differences were statistically significant for six of the traits. The target variances in the Mexican sample, which did not significantly differ from those in the US sample, were also greater than the target variances in China for most traits, and significantly so for eight traits. The target variances in Venezuela were frequently lower than those in the US and Mexico and higher than those in China, although many of the differences were not statistically significant. Although the target variance differences in the friend context were not large enough to produce a statistically significant cultural effect in the overall MANOVA, most of the target variances in the US and Mexico were larger than the target variances in China in the friend context as well. The bottom of Table 1 shows comparable results for the Big Five composites, each of which was measured by three of the specific traits. SOREMO estimates both stable and unstable components of target variance for each construct, which are separated by slashes (/) in Table 1. Stable target variance, which is shown to the left of the slashes, is variance that replicates across the three specific traits and will be larger when the traits are highly correlated. If the three traits are not highly correlated, target variance will be largely unstable (i.e., unique to the individual traits). In Table 1, stable target variance that was statistically significant (p < .05) is indicated by asterisks.3 Reliable stable variance was generally found for the Big Five constructs, with the primary exception of Openness to Experience. Inspection of the SOREMO correlations relating the target variances for the three Openness to Experience traits (imaginativeness, intelligent, and shallow) revealed lower correlations (e.g., less than .25 in some cases) than for the traits measuring the other Big Five constructs, which generally correlated .60 or higher. Other researchers have also observed that Openness to Experience traits are less tightly related than traits that define the other Big Five domains (Trapnell, 1994).4 Given the imperfect replication of target variances across indicators of the Big Five traits (i.e., the presence of unstable target variance), the relative target variances for the individual traits in the top part of Table 1 give a better indication of the degree of consensus that can be obtained in these cultures with the traits and rating methods used in the present study. For this reason, we emphasize the specific traits rather than the Big Five composites in our crosscultural comparisons of cross-observer agreement. Nonetheless, the stable target variances for four of the Big Five constructs, which were usually statistically significant, again indicate that observers can agree to some extent in their personality judgments of friends and family members. MANOVAs comparing these variances across cultures revealed significant cultural effects in both the friend (Wilks’ Lambda = .87, F[15, 514] = 1.72, p < .05) and family

3 SOREMO does not compute significance tests for the Big Five constructs, so the absolute target variances were output from the program and tested for statistical significance using SPSS. One-tailed t-tests determine whether the mean target variances are significantly different from 0 with group as the unit of analysis and the target variances weighted by group size minus 1 (Kenny, 2007, p. 12). 4 The few other cases of low stable target variance for the Big Five constructs can also be explained by low correlations between the relevant trait indicators. This includes the Emotional Stability traits in China and the Extraversion traits in the family context in Mexico and Venezuela. By comparison, the specific traits measuring the Agreeableness and Conscientiousness constructs were almost always highly correlated in each culture.

83

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89 Table 1 Consensus in trait judgments within social contexts in four cultures. Traits

Friends

Family

US

Mexico

Venezuela

China

US

Mexico

Venezuela

China

Energetic Extroverted Quiet Kind Helpful Selfish Disciplined Organized Lazy Relaxed Calm Moody Imaginative Intelligent Shallow Mean

.27* .18* .31* .15* .20* .17* .30* .25* .24* .21* .12* .32* .22* .18* .12* .22

.14* .26* .26* .16* .18* .27* .32* .28* .22* .10* .22* .19* .26* .16* .23* .22

.16* .26* .21* .08 .10* .06 .12 .18* .09 .20* .14* .20* .17* .14* .17* .15

.12 .32* .33* .02 .04 .04 .24* .19* .14* .10 .20* .06 .20* .15* .03 .15

.41* .32* .38*a .20*a,b .29* .26*a .37*a,b .46*a .29*a,b .24*a,b .32*a .33*a .33*a .22*a,b .18*a.b .31

.24* .18* .21*a,b .25*a .18* .21*a,b .35*a .34*a .35*a .28*a .24*a,b .32*a .33*a .25*a,b .26*a .27

.17* .18* .20*a,b .13*a,b .19 .10a,b .26*b .22*b .24*a,b .14*b .10*a,b .36*a .24*a,b .20*b .13a,b .19

.26* .22* .17*b .06*b .11* .04b .20*b .10b .16*b .15*b .08b .08b .14*b .32*a .05b .14

Big Five compositesa Extraversion Agreeableness Conscientiousness Emotional Stability Openness Mean

.19*/.05 .14*a,b/.03 .19*a,b/.07 .13*a/.09 .03/.15 .14/.08

.14*/.08 .17*a/.04 .22*a/.05 .09*a,b/.09 .04/.18 .13/.09

.13*/.08 .05b/.03 .13*b/.01 .15*a/.03 .06/.11 .10/.05

.25*/.02 .05b/.00 .15*a,b/.03 .01b/.11 .04/.09 .10/.05

.28*a/.09 .20*a/.05 .28*a,b/.10 .23*a/.07 .02/.24 .20/.11

.08*b/.13 .14*a,b/.07 .27*a/.08 .17*a/.12 .09*/.19 .15/.12

.06b/.13 .10*a,b/.05 .16*b,c/.08 .05*b/.16 .06/.12 .09/.11

.18*b/.04 .04*b/.03 .07c/.08 .00b/.14 .07*/.10 .07/.08

Note: Table entries are relative target variances estimated by the SOREMO program. Within the friend and family contexts, target variances that share a subscript are not significantly different (p > .05). In rows without subscripts, the cultures did not differ significantly (p > .05). a For the Big Five composites, entries to the left and right of the slash (/) are stable and unstable target variances, respectively. * p < .05.

(Wilks’ Lambda = .67, F[15, 508] = 5.35, p < .001) contexts. In follow-up ANOVAs, statistically significant (p < .05) culture effects were found for Agreeableness, Conscientiousness, and Emotional Stability in the friend context and for Extraversion, Agreeableness, Conscientiousness, and Emotional Stability in the family context, with effect sizes somewhat larger in the family context (g2 range = .06 to .15) than in the friend context (g2 range = .04 to .05). Subscripts in Table 1 again indicate the pattern of cultural differences in Tukey tests. Both the pattern of cultural differences and the mean sizes of the stable target variances in the last row of Table 1 again suggest that consensus is somewhat higher overall in the United States and Mexico than in Venezuela and China. Finally, some researchers have found that consensus is greater for Extraversion than for the other Big Five traits and have suggested this may be due to the greater visibility of extraverted behaviors (Connelly & Ones, 2010; Funder & Dobroth, 1987; Kenny & West, 2010; Park & Judd, 1989; Watson et al., 2000). However, inspection of the target variances in the present study revealed no consistent trend for consensus to be greater for extraverted traits, except perhaps in China. Thus, our within-context analyses support Kenny’s (1994) proposal that consensus may only be greater for Extraversion when observers are not wellacquainted. 3.1.2. Self-other agreement Cross-observer agreement can also be operationalized as selfother agreement. For each culture, Table 2 shows the disattenuated self-other correlations in the friend and family contexts estimated by SOREMO. Kenny (2007) has recommended interpretation of these correlations only if the target effects that are correlated with the self-ratings are themselves statistically significant. However, significance levels are a function of sample size, which varied across the four cultures, so we also set a fixed minimum amount of relative target variance—ten percent or more—for interpretation of these correlations in each culture (Kenny & West, 2010). In Table 2, excluded correlations based on this criterion are indicated by

dashes (—). Because non-zero self-other correlations require target variances that are significantly greater than zero, the excluded correlations were treated as zero in the computation of significant tests and the mean correlations across traits. In all four cultures, the self-other correlations were generally statistically significant and revealed moderate to high self-other agreement for most of the specific traits (SOREMO does not test the significance of the self-other correlations for the Big Five composites).5 The size of the self-other correlations is similar to those reported in previous studies (Connelly & Ones, 2010; de Vries, 2010; Kenny, 1994; Paulhus & Reynolds, 1995). Thus, if we treat observer ratings as a criterion for accuracy, target participants in all four cultures were generally able to judge their traits with some accuracy. In the friend context, the general pattern of cultural differences seen previously for consensus was again evident. Although there were exceptions for specific traits, self-other agreement was generally strongest for the Americans, followed by the Mexicans and Venezuelans, then the Chinese. However, self-other agreement with friends was relatively strong in the Chinese sample for extraverted traits, particularly energetic, extraverted, and the Big Five Extraversion composite. The correlation differences between cultures for the specific traits and Big Five composites were often not statistically significant because of the limited number of round-robin groups (i.e., the unit of analysis) in each culture (z-tests comparing independent correlations; Hays, 1973, p. 663). Different subscripts in Table 2 indicate any significant cultural differences within the friend and family contexts. However, excluding

5 We cannot draw conclusions about the relative degree of consensus versus selfother agreement in the present study by comparing the SOREMO estimates reported in Tables 1 and 2. The SOREMO target variance estimates reported in Table 1 can be interpreted as the correlation between two observers (i.e., one-with-one correlations; Kenny et al., 1994). In contrast, the self-other agreement correlations reported by SOREMO relate self-ratings with the ratings of multiple observers and are also disattenuated to take into account the reliability of the perceiver and target effects. Therefore, we focus on the pattern of cultural differences in consensus and self-other agreement rather than the relative degree of consensus versus self-other agreement.

84

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89

Table 2 Self-other agreement in trait judgments within social contexts in four cultures. Traits

Friends

Family

US

Mexico

Venezuela

China

US

Mexico

Venezuela

China

Energetic Extroverted Quiet Kind Helpful Selfish Disciplined Organized Lazy Relaxed Calm Moody Imaginative Intelligent Shallow Meana

.50*a .54*a,b .65* .32* .03 .29* .35* .61* .43*a .46* .45* .35*a,b .48* .47*a .28* .42

.43*a .37*b .56* .20 .13 .25* .53* .52* .37*a,b .58* .35* .35*a,b .42* .10b .31* .37

.32*a .48*b .54* — .09 — .33* .54* —c .33* .52* .43*a .46* .27*a,b .19 .31

.79*b .73*a .44* — — — .22 .48* .05b,c .40 .39* —b .52* .28a,b — .32

.58*a .52* .64*a .24* .23*b .29* .42* .52* .37* .44* .36*a .41*b .36* .36* -.12b .39

.32*b .50* .56*a,b,c .27* .41*a,b .41* .32* .51* .54* .49* .50*a .64*a .30* .41* .34*a .44

.49*a,b .55* .34*b,c .43* .64*a .20 .37* .56* .41* .67* .43*a .49*a,b .41* .35* .38*a .46

.29*a,b .51* .27c .29 .48*a,b — .53* .50* .49* .16 —b —c .45* .55* —b .32

Big Five compositesb Extraversion Agreeableness Conscientiousness Emotional stability Openness Meana

.80 .63a .75 .44b — .58

.73 .10b .63 .51a,b — .44

.67 —b .59 .68a — .43

.79 —b .59 —c — .34

.79a .29a,b .60a .53b — .49

.59b .48a .61a .84a .26 .59

—c .30a,b .74a .64b — .38

.49b .04b —b —c .20 .15

Note: Within the friend and family contexts, correlations that share a subscript are not significantly different (p > .05). In rows without subscripts, the cultures did not differ significantly (p > .05). Dash (—) entries are shown when the relative target variances that were correlated with self-ratings were not statistically significant or were lower than 10% (see text). The dash entries were treated as zero in the computation of means and in cultural comparisons. a Fisher’s r to Z transformations were used to compute the mean correlations. b SOREMO does not test the significance of the self-other correlations for the Big Five composites. * p < .05.

the three extraversion traits, for which the Chinese exhibited strong self-other agreement, we found marginally significant cultural differences in the average of the remaining 12 self-other correlations in and ANOVA (F[3, 44] = 2.29, p = .09, g2 = .14; Fisher’s r-to-z transformations were used). Follow-up t-tests revealed that the average of these self-other correlations was significantly lower in the Chinese sample than in the American sample (t[22] = 2.42, p < .05). With the small number of traits in this analysis, none of the other pair-wise mean comparisons were statistically significant. The usual pattern did not hold in the family context. For many of the specific traits, self-other agreement was as strong or stronger in Mexico and Venezuela as in the United States. In an ANOVA comparing the mean correlations for the 15 specific traits, we again found a marginally significant cultural effect (F[3, 56] = 2.31, p < .09, g2 = .11). Follow-up t-tests suggested that the average correlation across the 15 specific traits was significantly lower in the Chinese sample than in the Mexican (t[28] = 2.03, p < .06) and Venezuelan (t[28] = 2.22, p < .05) samples. To speculate, the higher self-other correlations in the family context in Mexico and Venezuela might reflect the greater influence of family members’ perceptions on targets’ self-perceptions in cultures characterized by close family ties (Diaz-Loving & Draguns, 1999).

3.2. Across-context consensus If traits show some consistency across contexts, then some consensus in trait judgments should be observed across the friend and family contexts. At the same time, drawing on cultural psychology perspectives, we expected even greater cultural differences in consensus across contexts, as compared to within contexts. Consensus across contexts was quantified by correlating the target effects estimated for the target participants in the friend and family contexts. The correlations are shown in Table 3. Dash (—) entries again

Table 3 Across-context correlations between target effects in four cultures. Traits

US

Mexico

Venezuela

China

Energetic Extroverted Quiet Kind Helpful Selfish Disciplined Organized Lazy Relaxed Calm Moody Imaginative Intelligent Shallow Meana

.52**a .53**a .41**a .27* .29*a,b .13 .49**a .42** .08 .31* .45**a .33*a,b .25 .42**a .13 .34

.27*a,b .16b .32*a,b .07 .04b .07 .36**a .46** .29* .20 .01b .43**a .02 .22a,b .09 .20

.18b .44**a,b .08b,c — .45**a — .02b .45** — .31 .16a,b .45**a .31 .07b .11 .21

.43*a,b .59**a .52**a — —b — -.07b .28 .02 -.01 —b —b .11 .19a,b — .15

Big Five composites Extraversion Agreeableness Conscientiousness Emotional Stability Openness Meana

.58**a .29* .44**a .40**a,b .32* .41

.29*b .17 .45**a .19b .09 .24

.29b .43** .20a,b .52**a .24 .34

.63**a .22 .09b .13b .09 .25

Note: Dash (—) entries are shown when the relative target variances that were correlated with self-ratings were not statistically significant or were lower than 10% (see text). The dash entries were treated as zero in the computation of means and in cultural comparisons. Within each row, correlations that share a subscript were not significantly different (p > .05). In rows without subscripts the cultures did not differ significantly (p > .05). a Fisher’s r to Z transformations were used to compute mean correlations. * p < .05. ** p < .01.

indicate traits for which target variance was not statistically significant in the relevant culture or failed to account for at least 10% of the total variance in the trait judgments.

85

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89

As evident in Table 3, there were substantial differences in the number of statistically significant across-context correlations found in the four cultures. Statistically significant across-context correlations were found for 11 of 15 specific traits and all of the Big Five composites in the United States, 6 of 15 specific traits and 2 of 5 Big Five composites in Mexico, 4 of 15 specific traits and 2 of 5 Big Five composites in Venezuela, and 3 of 15 specific traits and one Big Five composite in China. The proportion of specific traits with statistically significant correlations was significantly greater in the United States than in Venezuela (v2[1] = 6.53, p < .05) and China (v2[1] = 8.57, p < .01), as marginally greater than in Mexico (v2[1] = 3.39, p < .07). Direct pair-wise comparisons of the correlations in Table 3 for each specific trait and Big Five composite often revealed statistically significant cultural differences (different subscripts in a row indicate significant differences; z-tests comparing independent correlations; Hays, 1973, p. 663). Although the pattern of cultural differences varied somewhat across the various traits, the mean of the correlations for the 15 specific traits was largest in the American sample, intermediate in the Mexican and Venezuelan samples, and smallest in the Chinese sample. In an ANOVA the culture effect was statistically significant (F[3, 56] = 3.12, p < .05, g2 = .14). Follow-up t-tests indicated that the across-context correlations were significantly larger, on average, in the American sample than in the Mexican (t[28] = 2.57, p < .05), Venezuelan (t[28] = 2.19, p < .05), and Chinese samples (t[28] = 2.77, p < .05). When we excluded the three extraversion traits—the only traits for which the Chinese exhibited across-context consensus—the average of the remaining acrosscontext correlations were also significantly higher in the Mexican and Venezuelan samples than in the Chinese sample (t[22] range = 2.28 to 2.46, p < .05). Thus, the results supported our hypothesis that across-context consensus is stronger in the U.S. than in China, and the Mexican and Venezuelan samples can be viewed as intermediate in across-context consensus. Although the Chinese sample showed the least across-context consensus overall, the consistent across-context correlations in China for extraverted traits and the Big Five Extraversion composite indicates that across-context consensus is possible for some traits in that culture. Thus, the results support our hypothesis that across-context consensus in trait judgments can be observed in all cultures, at least for some traits. Finally, in the US, most of the specific traits for which within-context consensus was observed also showed significant across-context consensus. This was less the case in Mexico and Venezuela, and least so in China. Thus, as anticipated, cultural differences were greater for across-context consensus than for within-context consensus. 3.3. Self-concept consistency across contexts The across-context results revealed cultural differences in the extent to which friends and family members view the traits of the target participants differently. Given this finding, it is of interest to examine whether the target participants themselves view their traits differently across the two contexts. Table 4 shows the correlations between the target participants’ ratings of their own traits in the two contexts, which were computed for each trait across target participants. Most of the correlations were somewhat higher in the United States and Venezuela than in Mexico and China, but only a minority of the correlation differences were statistically significant (z-tests on differences between independent correlations; different subscripts indicate significant differences). In an ANOVA comparing the means of the correlations for the 15 specific traits, the culture effect was marginally significant (F[3, 56] = 2.27, p = .09, g2 = .11). Follow-up t-tests revealed that the correlations in the Mexican sample were significantly lower, on average, than the correlations in the Venezuelan (t[28] = 2.37,

Table 4 Across-context correlations for self-ratings in four cultures. Traits

US

Mexico **

**

Venezuela **

China

Energetic Extroverted Quiet Kind Helpful Selfish Disciplined Organized Lazy Relaxed Calm Moody Imaginative Intelligent Shallow Meana

.57 .70**a .61** .55**b .39**b .60** .51** .75** .71** .56**a,b .63** .62**b,c .80**a .50** .79** .63

.53 .41**b .61** .62**a,b .25b .53** .53** .69** .48** .29*b .57** .50**c .52**b .63** .74** .54

.60 .72**a .50** .48**b .74**a .62** .44** .74** .59** .65**a .51** .83**a .68**a,b .70** .68** .65

.38* .54**a,b .55** .80**a .35*b .53** .52** .62** .59** .37*a,b .54** .77**a,b .60**a,b .58** .70** .58

Big Five composites Extraversion Agreeableness Conscientiousness Emotional Stability Openness Meana

.71** .64** .80** .75**a,b .75** .73

.54** .49** .73** .52**c .62** .59

.54** .72** .77** .84**a .76** .74

.61** .54** .71** .60**b,c .75** .65

Note: Within each row, correlations that share a subscript were not significantly different (p > .05). In rows without subscripts the cultures did not differ significantly (p > .05). a Fisher’s r to Z transformations were used to compute mean correlations. * p < .05. ** p < .01.

p < .05) and American (t[28] = 2.06, p < .05) samples. Overall, however, fairly high correlations in all four cultures indicate that target participants in each culture rated their traits in a fairly consistent manner across the two contexts. Thus, despite the fact that friends and family members viewed the target participants differently on some traits, particularly in China, self-perceptions of trait consistency across contexts was substantial in all four cultures. 3.4. Cultural dimensions and mediation We hypothesized that some of the cultural differences in consensus would be mediated by differences in individualism-collectivism, dialecticism, or cultural tightness. In a MANOVA with these cultural measures as dependent variables, we found a significant culture effect (Wilks’ Lambda = .54, F[12, 506] = 9.97, p < .01). Follow-up ANOVAs revealed significant cultural differences for dialecticism (F[3, 194] = 9.85, p < .01, g2 = .25), tightness (F[3, 194] = 3.47, p < .01, g2 = .13), independent self-construal (F[3, 194] = 15.65, p < .01, g2 = .20), and collective self-construal (F[3, 194] = 5.63, p < .01, g2 = .08). Table 5 shows the results of Tukey tests on the cultural means. Means that share the same subscript were not significantly different. Because only the target participants completed the cultural measures, the sample sizes were relatively small for this analysis. Nonetheless, the pattern of cultural differences was very similar to that reported in our previous study with large samples (Church et al., in press). Chinese again averaged higher in dialecticism and cultural tightness than the other three cultural groups. The Chinese and Venezuelans again averaged highest in collective self-construal, although they were not significantly higher than the American sample. Mexicans and Venezuelans again averaged higher in independent self-construal than Americans and Chinese. Most importantly, inspection of the pattern of means in Table 5 for the cultural dimensions revealed that they did not exhibit the same rank-order as the cultural differences generally found in the within- and across-context agreement results, where agreement was generally greatest in the United

86

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89

Table 5 Comparison of cultural dimensions. Dimension

US

Mexico

Venezuela

China

g2

Dialecticism M SD

3.66b .65

3.48b,c .67

3.24c .77

4.40a .57

.25

Tightness M SD

3.68b .57

3.43b .64

3.62b .69

4.11a .50

.13

Independent Self-construal M 4.23b SD .61

4.67a .64

4.70a .66

3.93b .53

.20

Collective Self-construal M 4.35a,b SD .51

4.09b .73

4.53a .64

4.54a .69

.08

Note: Means in each row that share a subscript are not significantly different (p > .05) in Tukey tests. g2 = g2 (i.e., ANOVA effect size).

States, intermediate in Mexico and Venezuela, and lowest in China. Therefore, the cultural dimensions could not mediate the consensus differences between the four cultures.6 4. Discussion Both cross-observer agreement and self-concept consistency provide persuasive evidence for the existence of stable traits, particularly when demonstrated across contexts (Connelly & Ones, 2010; McCrae et al., 2004). At the same time, cultural differences in these variables would support cultural psychology perspectives by suggesting that trait-relevant consistency is reduced in some cultures, particularly if these differences can be explained in terms of the cultural dimensions proposed by cultural psychologists (Markus & Kitayama, 1991, 1998; Spencer-Rodgers et al., 2010b). We found considerable support for trait perspectives. In four diverse cultures, we found significant within-context consensus and self-other agreement for the majority of traits, across-context consensus for some traits, and substantial self-concept consistency. Cultural differences were also observed. With the exception of the self-other correlations in the family context, cross-observer agreement was generally strongest for the Americans, followed by the Mexicans, Venezuelans, and Chinese, in that order. However, measures of dialecticism, self-construals, and cultural tightness failed to mediate these cultural differences. 4.1. Comparisons with previous cross-cultural studies 4.1.1. Cross-observer agreement Of the few cross-cultural studies of agreement, the present investigation is most similar to the study by Malloy et al. (2004), who applied the SRM in samples of well-acquainted Mexicans and Chinese, two of the cultural groups in the present study. Several findings in the two studies were similar. Foremost was the finding of consensus within both friend and family contexts for most traits in both cultures, as well as the substantially greater within-context consensus in Mexico than in China. Also, in both studies, across-context consensus was substantially greater in Mexico than in China, and the largest across-context correlations in China involved extraversion traits. Our self-other agreement findings were also similar to those of Malloy et al. Both studies found significant self-other agreement for most Big Five traits in Mexico but non-significant or weak agreement for agreeableness 6 Using procedures described by Baron and Kenny (1986) we did test for mediation of cultural differences in consensus between the United States and China, because these two cultures differed in the same direction on cross-observer agreement and the cultural dimensions. However, none of the cultural dimensions mediated the consensus differences between these two cultures.

traits in the friend context. This suggests that Mexican college students’ self-perceptions of agreeableness traits conform less well to the consensus of their peers than traits associated with the other Big Five dimensions. Finally, both studies found that in China self-other agreement was highest for extraverted traits in the friend context. These similar results increase confidence in the replicability and meaningfulness of the findings, despite some differences in methods between the two studies. Specifically, in the present study we used on-line and paper surveys, whereas Malloy et al. employed phone interviews. Albright et al. (1997) also applied the SRM in a Chinese sample and obtained target variance estimates that ranged from .08 for Conscientiousness to .16 for Extraversion (M = .11). In our Chinese sample, we obtained higher target variance estimates for most traits, probably because our participants were well-acquainted, whereas Albright et al.’s participants made their ratings based solely on visual exposure to the other participants (i.e., a zeroacquaintance study). Finally, Malloy et al. (1997) used the SRM to examine across-context consensus, but only in a United States sample. The size of the across-context (friend vs. family) correlations in that study were quite similar to those observed in our United States sample. These comparisons with other SRM studies of consensus suggest that our results are not atypical. Recall that Kenny (1994) proposed a 15% ‘‘rule’’ for the anticipated amount of relative target variance found in consensus studies with the SRM. The target variance estimates in the present study exceeded this ‘‘rule,’’ on average, in the United States and Mexico and were about this size in Venezuela and China. It is more difficult to compare our results with the remaining cross-cultural studies of agreement because they did not apply the SRM model. In these studies target effects were potentially confounded with perceiver effects (Kenny, 1994). Nonetheless, our finding of greater agreement in the United States than in China resembles Heine and Renshaw’s (2002) finding of greater self-peer agreement in the United States than in Japan. Our results also resemble those of Church et al. (2006a), who reported higher consensus and self-other agreement in American and Mexican students than in Filipino students. In contrast, our results are less consistent with McCrae et al.’s (2004) review of agreement studies with the NEO-PI-R and other Big Five measures. They reported cross-observer correlations in Chinese adults (Yang et al., 1999) and Korean students in the United States (Spirrison & Choi, 1998) that were similar in size to those in North American samples. However, only self/spouse agreement was reported in these Asian samples, which may be more similar across cultures than consensus among friends and other family members (Connelly & Ones, 2010). In addition, the studies summarized by McCrae et al. did not apply the SRM and assessed self-other agreement with entire scales, rather than single trait adjectives. Overall, there is now sufficient evidence to conclude that crossobserver agreement—including both consensus and self-other agreement—may be lower in some Asian cultures than in the United States (Albright et al., 1997; Church et al., 2006a; Heine & Renshaw, 2002; Malloy et al., 2004). However, the basis for these differences in less definitive, as we discuss later. 4.1.2. Self-concept consistency Given the fairly consistent and substantial pattern of cultural differences in consensus, particularly across contexts, the finding of substantial self-concept consistency across contexts in all four cultures is noteworthy. In particular, although friends and family members judged each other with less consensus in China than in the other three cultures, the Chinese target participants nonetheless perceived their own traits as fairly consistent across the two contexts.

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89

Boucher (2011) found lower self-concept consistency in Chinese, as compared to Americans. However, our results are consistent with two previous studies in which Chinese showed comparable levels of self-concept consistency as Americans, Australians, Mexicans, Venezuelans, Filipinos, and Malaysians (Church et al., 2008, in press). In these studies only Japanese showed lower self-concept consistency than the other cultural groups. Thus, reduced self-concept consistency may only apply consistently to selected East Asian cultures including Japan (Church et al., 2008; Kanagawa, Cross, & Markus, 2001) and Korea (Suh, 2002). A likely explanation for substantial self-concept consistency— concomitant with more limited across-context consensus between friends and family members—is provided by role identity theory (Wood & Roberts, 2006). According to role identity theory, individuals formulate self-perceptions of their traits (i.e., general identities) by aggregating their identities or trait perceptions in different roles. In turn, the consistent self-ratings across friend and family contexts would reflect the impact of individuals’ general identities on their trait ratings across contexts. Self-concept consistency could also reflect a motive (Lecky, 1945) or bias (Leising, 2011; Sadler & Woody, 2003) for self-consistency, leading individuals in all cultures to perceive themselves in different contexts in a manner that reflects their general view of themselves. Even in non-Western cultures, consistent self-perceptions may provide a sense of coherence in identity that is important for psychological well-being (Church et al., 2008). Of course, friends and family members who observe target individuals in different contexts would be less able to formulate general trait perceptions of target individuals that generalize across contexts, leading to lower across-context consensus than self-concept consistency. 4.2. Interpretation of cultural differences In general, we found cross-observer agreement to be greatest in the US, followed by Mexico, Venezuela, and China, in that order. One plausible explanation for the observed cultural differences is consistent with our original hypothesis linking greater crossobserver agreement to the individualism-collectivism distinction. The finding that agreement was generally strongest in the United States is consistent with the cultural psychology hypothesis that behavior is more ‘‘traited’’ or consistent in Western or individualistic cultures (Church, 2000; Heine, 2001; Markus & Kitayama, 1998). Some recent evidence suggests that Mexican society, or at least more educated and urban Mexicans, may also be relatively individualistic. Internationally, Mexicans are above average in Intellectual Autonomy values (Schwartz, 2002) and are similar to the United States in the individualism of their cultural products (Morling & Lamoreaux, 2008). In previous studies, we have found that Mexican college students consistently average as high or higher than American students on measures of independent selfconstrual (Church et al., 2003, 2006a). The generally lower agreement in the Venezuelan and especially Chinese samples would be consistent with their higher collective self-construal scores in the present sample and in a previous study (Church et al., in press). In this explanation, the observed cultural differences in agreement reflect differences in individualism-collectivism, despite the failure of the self-construal measures to mediate these differences. A significant exception to the general pattern in our Chinese results involved extraversion traits, which were the only traits to show high across-context consensus in that culture. This might be due to the greater temperamental nature of these traits (Clark & Watson, 2008) or their greater visibility (Connelly & Ones, 2010; Funder & Dobroth, 1987; Kenny, 1994; Kenny & West, 2010). A second plausible explanation of the cultural differences is suggested by Malloy et al.’s (2004) explanation of their own findings in Mexico and China. Malloy et al. argued that prescriptions

87

for social behavior in Mexico are general rather than dyadic and that Mexicans, like Americans, are socialized to evaluate others in terms of traits. Similarly, Diaz-Loving and Draguns (1999) argued that Mexican ‘‘interaction styles brewed in the family environment spill over into the realms of friends and acquaintances’’ (p. 111). In contrast, Chinese have dyadic prescriptions for social behavior that derive from Confucian traditions of filial piety (Ho, 1996) and are more likely to rely on situational cues to evaluate behavior (Choi, Nisbett, & Norenzayan, 1999; Morris & Peng, 1994). This explanation would account for the generally higher consensus in the United States and Mexico, as compared to China. In this interpretation, Venezuela might be considered more similar to Mexico than China in their general versus dyadic prescriptions for social behavior, although we could not identify any studies that address this point. It is noteworthy that the two explanations we offer for the pattern of cultural differences observed in the study are compatible if our Mexican sample is, in fact, relatively individualistic, resulting in behavior that is more traited or consistent than in more collectivistic cultures such as Venezuela and China.

4.3. Failure of mediation tests A limitation of the present study was the failure of the cultural dimensions to mediate the cultural differences in cross-observer agreement. Cross-cultural researchers have emphasized the need to directly measure the cultural dimensions that may mediate cultural differences (van de Vijver & Leung, 1997). However, a number of cross-cultural measurement issues complicate the assessment of these dimensions in an equivalent or comparable manner across cultures, reducing their effectiveness as mediators (Church, 2010; van de Vijver & Leung, 1997). Although full metric (loading) and partial scalar (intercept) equivalence were demonstrated for these cultural measures in a previous study (Church et al., in press), such analyses do not eliminate the possibility that scores were impacted by cultural differences in response styles (e.g., acquiescence) or the reference groups used to make one’s ratings (Heine, Lehman, Peng, & Greenholtz, 2002). Such factors may have contributed to our finding that the four cultures were not ordered along the cultural dimensions in the same order as they generally differed in crossobserver agreement. The American and Chinese samples did show the expected mean differences on the cultural dimensions, but even their differences in cross-observer agreement were not mediated by the cultural dimensions. In previous studies we have been more successful using the same cultural measures. For example, Church et al. (submitted for publication) found that dialecticism and independent selfconstruals accounted for differences between Asian and non-Asian cultures in self-ratings of need satisfaction and well-being. Several other studies involving Asian cultures have reported successful mediation effects using the DSS scale (Boucher, 2011; SpencerRodgers et al., 2010b), although none of these studies investigated cross-observer agreement. A possible explanation for the success of the dialecticism and self-construal measures as mediators in our previous studies, but not in the present study (e.g., in comparisons of the Americans and Chinese), is that these measures assess self perceptions and beliefs. As such, they may be better able to mediate cultural differences in various self-perceptions (e.g., self-ratings of one’s need satisfaction and well-being) than cultural differences involving the perceptions of others (e.g., cross-observer agreement). Although the unsuccessful mediation tests led to the rejection of one of our hypotheses, it is nevertheless a potentially important finding because it may help to clarify the psychological phenomena that are (e.g., well-being) and are not (e.g., consensus) accounted for by these cultural dimensions.

88

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89

4.4. Strengths and limitations Compared to most studies of cross-observer agreement, strengths of the present study included application of the SRM; sampling of more cultures; inclusion of more specific traits, as well as Big Five composites; analyses of more round-robin groups; and direct measurement of the cultural dimensions hypothesized to account for cultural differences. There were also several limitations. First, complete round-robin data was obtained for only 66-83% of the target participants across the four cultures. It is possible that potential participants who did not respond were less conscientious. However, the focus of the study was on cross-observer agreement not the level of the trait ratings. Second, it is possible that respondents in the United States, Venezuela, and China had differential access to, or familiarity with, the internet. However, the similar response rates in these three cultures may alleviate this concern. In addition, our Mexican and Chinese results were similar to the findings of Malloy et al. (2004), who used phone surveys. Third, some of the specific trait indicators, particularly for Openness to Experience, did not cohere as well as anticipated with other indicators of the same Big Five domain. This reduced stable target variance for the Big Five composites and made cross-cultural comparisons at the level of these composites less useful than comparisons involving the specific traits. There are trade-offs in the selection of Big Five indicators. Researchers will find greater stable target variance if they select highly correlated indicators, but at the cost of representativeness and breadth of coverage. Fourth, the proportions of target variance in the study, although typical of most previous studies (Kenny, 1994), might have been increased by disallowing ties in the participants’ ratings (Paulhus & Reynolds, 1995) or by using personality inventories rather than trait adjectives to obtain personality judgments (de Vries, 2010). Both of these options, however, place increased demands on participants. Finally, there is ethnic and geographical diversity in each of the countries examined, perhaps especially in China, so our findings may not generalize to all ethnic or regional groups. 5. Conclusion We sought support for both trait and cultural psychology perspectives on cross-observer agreement using the Social Relations Model (SRM). In all four cultures, consensus and selfother agreement was observed within friend and family contexts, across-context consensus was demonstrated for some traits, and self-concept consistency across contexts was substantial. Each of these findings is consistent with the cultural universality of stable traits. Cultural differences in levels of cross-observer agreement were also observed and may reflect differences in individualismcollectivism or prescriptions for social behavior within and across contexts. Cross-cultural studies of cross-observer agreement and self-concept consistency are still rare, particularly using the SRM, and future studies should sample an even greater variety of cultures. Such research has important implications for trait theory and for interpersonal perception generally. Acknowledgment The research was supported by National Science Foundation Grant 0953940. References Albright, L., Kenny, D. A., & Malloy, T. E. (1988). Consensus in personality judgments at zero acquaintance. Journal of Personality and Social Psychology, 55, 387–395. Albright, L., Malloy, T. E., Dong, Q., Kenny, D. A., Fang, X., Winquist, L., et al. (1997). Cross-cultural consensus in personality judgments. Journal of Personality and Social Psychology, 72, 558–569.

Ambady, N., & Rosenthal, R. (1992). Thin slices of expressive behavior as predictors of interpersonal consequences: A meta-analysis. Psychological Bulletin, 111, 256–274. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Baron, R. M., & Misovich, S. J. (1993). Dispositional knowing from an ecological perspective. Personality and Social Psychology Bulletin, 19, 541–552. Biesanz, J. C., West, S. G., & Millevoi, A. (2007). What do you learn about someone over time? The relationship between length of acquaintnance and consensus and self-other agreement in judgments of personality. Journal of Personality and Social Psychology, 92, 119–135. Borkenau, P., & Liebler, A. (1992). Trait inferences: Sources of validity at zero acquaintance. Journal of Personality and Social Psychology, 62, 645–657. Boucher, H. C. (2011). The dialectical self-concept II: Cross-role and within-role consistency, well-being, self-certainty, and authenticity. Journal of CrossCultural Psychology, 42, 1251–1271. http://dx.doi.org/10.1177/00220221 10383316. Choi, I., Nisbett, R. E., & Norenzayan, A. (1999). Causal attribution across cultures: Variation and universality. Psychological Bulletin, 125, 47–63. Church, A. T. (2000). Culture and personality: Toward an integrated cultural trait psychology. Journal of Personality, 68, 651–703. Church, A. T. (2009). Prospects for an integrated trait and cultural psychology. European Journal of Personality, 23, 153–182. http://dx.doi.org/10.1002/per.700. Church, A. T., Anderson-Harumi, C. A., del Prado, A. M., Curtis, G. J., Tanaka-Matsumi, J., Valdez Medina, J. L., et al. (2008). Culture, cross-role consistency, and adjustment: Testing trait and cultural psychology perspectives. Journal of Personality and Social Psychology, 95, 739–755. Church, A. T., Alvarez, J. M., Katigbak, M. S., Mastor, K. A., Cabrera, H. F., TanakaMatsumi, J., et al. (2012). Self-concept consistency and stability in eight cultures. Journal of Research in Personality, 46, 556–570. http://dx.doi.org/ 10.1016/j.jrp.2012.06.003. Church, A. T., Katigbak, M. S., Locke, K. D., Zhang, H. -S., Shen, J., Vargas-Flores, J. d., & et al. (in press). Need satisfaction and well-being: Testing self-determination theory in eight cultures. Journal of Cross-Cultural Psychology. Church, A. T., Katigbak, M. S., del Prado, A. M., Ortiz, F. A., Mastor, K. A., Harumi, Y., et al. (2006a). Implicit theories and self-perceptions of traitedness across cultures: Toward integration of cultural and trait psychology perspectives. Journal of Cross-Cultural Psychology, 37, 694–716. http://dx.doi.org/10.1177/ 0022022106292078. Church, A. T., Katigbak, M. S., del Prado, A. M., Valdez-Medina, J. L., Miramontes, L. G., & Ortiz, F. A. (2006b). A cross-cultural study of trait self-enhancement, explanatory variables, and adjustment. Journal of Research in Personality, 40, 1169–1201. http://dx.doi.org/10.1016/j.jrp. 2006.01.004. Church, A. T., Ortiz, F. A., Katigbak, M. S., Avdeyeva, T., Emerson, A. M., Vargas, J. J. F., et al. (2003). Measuring individual and cultural differences in implicit trait theories. Journal of Personality and Social Psychology, 85, 332–347. http:// dx.doi.org/10.1037/0022-3514.85.2.332. Church, A. T. (2010). Measurement issues in cross-cultural research. In G. Walford, M. Viswanathan, & E. Tucker (Eds.), The Sage handbook of measurement (pp. 151–177). Thousand Oaks, CA: Sage Publications. Clark, L. A., & Watson, D. (2008). Temperament: An organizing paradigm for trait psychology. In O. P. John, R. W. Robins, & L. A. Pervin (Eds.), Handbook of personality: Theory and research (3rd ed., pp. 265–286). New York: Guilford. Connelly, B. S., & Ones, D. S. (2010). An other perspective on personality: Metaanalytic integration of observers’ accuracy and predictive validity. Psychological Bulletin, 136, 1092–1122. de Vries, R. E. (2010). Lots of target variance. An update of SRM using the HEXACO Personality Inventory. European Journal of Personality, 24, 169–188. Diaz-Loving, R., & Draguns, J. (1999). Culture, meaning and personality in Mexico and in the United States. In Y. T. Lee, C. McCauley, & J. Draguns (Eds.), Personality and person perception across cultures (pp. 103–126). Mahwah, NJ: Lawrence Erlbaum. English, T., & Chen, S. (2007). Culture and self-concept stability: Consistency across and within contexts among Asian Americans and European Americans. Journal of Personality and Social Psychology, 93, 478–490. English, T., & Chen, S. (2011). Self-concept consistency and culture: The differential impact of two forms of consistency. Personality and Social Psychology Bulletin, 37, 838–849. http://dx.doi.org/10.1177/0146167211400621. Funder, D. C. (1995). On the accuracy of personality judgment: A realistic approach. Psychological Review, 102, 652–670. Funder, D. C., & Colvin, C. R. (1988). Friends and strangers: Acquaintanceship, agreement, and the accuracy of personality judgment. Journal of Personality and Social Psychology, 55, 149–158. Funder, D. C., & Dobroth, K. M. (1987). Differences between traits: Properties associated with interjudge agreement. Journal of Personality and Social Psychology, 52, 409–418. Funder, D. C., Kolar, D. C., & Blackman, M. C. (1995). Agreement among judges of personality: Interpersonal relations, similarity, and acquaintanceship. Journal of Personality and Social Psychology, 69, 656–672. Gelfand, M. J., Nishii, L. H., & Raver, J. L. (2006). On the nature and importance of cultural tightness-looseness. Journal of Applied Psychology, 91, 1225–1244. Gelfand, M. J., Raver, J. L., Nishii, L., Leslie, L. M., Lun, J., Lim, B. C., et al. (2011). Differences between tight and loose cultures: A 33-nation study. Science, 332, 1100–1104. http://dx.doi.org/10.1126/science.1197754.

M.S. Katigbak et al. / Journal of Research in Personality 47 (2013) 78–89 Goldberg, L. R. (1992). The development of markers for the Big-Five factor structure. Psychological Assessment, 4, 26–42. Hays, W. L. (1973). Statistics for the social sciences (2nd ed.). New York: Holt, Rinehart and Winston. Heine, S. J. (2001). Self as a cultural product: An examination of East Asian and North American selves. Journal of Personality, 69, 880–906. Heine, S. J. (2012). Cultural psychology (2nd ed.). New York: W.W. Norton. Heine, S. J., Lehman, D. R., Peng, K., & Greenholtz, J. (2002). What’s wrong with crosscultural comparisons of subjective Likert scales? The reference-group effect. Journal of Personality and Social Psychology, 82, 903–918. Heine, S. J., & Renshaw, K. (2002). Interjudge agreement, self-enhancement, and liking: Cross-cultural divergences. Personality and Social Psychology Bulletin, 28, 578–587. Ho, D. Y. F. (1996). Filial piety and its psychological consequences. In M. H. Bond (Ed.), The handbook of Chinese psychology (pp. 155–165). Hong Kong: Oxford University Press. Jang, K. L., McCrae, R. R., Angleitner, A., Riemann, R., & Livesley, W. J. (1998). Heritability of facet-level traits in a cross-cultural twin sample: Support for a hierarchical model of personality. Journal of Personality and Social Psychology, 74, 1556–1565. Kanagawa, C., Cross, S. E., & Markus, H. R. (2001). Who am I?: The cultural psychology of the conceptual self. Personality and Social Psychology Bulletin, 27, 90–103. Kashima, E. S., & Hardie, E. A. (2000). The development and validation of the Relational, Individual, and Collective self-aspects (RIC) Scale. Asian Journal of Social Psychology, 3, 19–48. Kenny, D. A. (1994). Interpersonal perception: A social relations analysis. New York: Guilford. Kenny, D. A. (2007). SOREMO version (V.2) [Manual]. Retrieved . Kenny, D. A., Albright, L., Malloy, T. E., & Kashy, D. A. (1994). Consensus in interpersonal perception: Acquaintance and the Big Five. Psychological Bulletin, 116, 249–258. Kenny, D. A., & La Voie, L. (1984). The social relations model. In L. Berkowitz (Ed.). Advances in experimental social psychology (Vol. 18, pp. 142–182). San Diego: Academic. Kenny, D. A., & West, T. V. (2010). Similarity and agreement in self- and other perception: A meta-analysis. Personality and Social Psychology Review, 14, 196–213. http://dx.doi.org/10.1177/1088868309353414. Lecky, P. (1945). Self-consistency: A theory of personality. New York: Island Press. Leising, D. (2011). The consistency bias in judgments of one’s own interpersonal behavior: Two possible sources. Journal of Individual Differences, 32, 137–143. http://dx.doi.org/10.1027/1614-0001/a000046. Malloy, T. E., & Albright, L. (1990). Interpersonal perception in a social context. Journal of Personality and Social Psychology, 58, 419–428. Malloy, T. E., Albright, L., Díaz-Loving, R., Dong, Q., & Lee, Y.-T. (2004). Agreement in personality judgments within and between nonoverlapping social groups in collectivist cultures. Personality and Social Psychology Bulletin, 30, 106–117. http://dx.doi.org/10.1177/0146167203258863. Malloy, T. E., Albright, L., Kenny, D. A., Agatstein, F., & Winquist, L. (1997). Interpersonal perception and metaperception in non-overlapping social groups. Journal of Personality and Social Psychology, 72, 390–398. Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224–253. Markus, H. R., & Kitayama, S. (1998). The cultural psychology of personality. Journal of Cross-Cultural Psychology, 29, 63–87. McCrae, R. R. (2000). Trait psychology and the revival of personality and culture studies. American Behavioral Scientist, 44, 10–31. McCrae, R. R., Costa, P. T., Jr., Martin, T. A., Oryol, V. E., Rukavishnikov, A. A., Senin, I. G., et al. (2004). Consensual validation of personality traits across cultures.

89

Journal of Research in Personality, 38, 179–201. http://dx.doi.org/10.1016/S00926566(03)00056-4. Morling, B., & Lamoreaux, M. (2008). Measuring culture outside the head: A metaanalysis of individualism-collectivism in cultural products. Personality and Social Psychology Review, 12, 199–221. Morris, M. W., & Peng, K. (1994). Culture and cause: American and Chinese attributions for social and physical events. Journal of Personality and Social Psychology, 67, 949–971. Park, B., & Judd, C. M. (1989). Agreement on initial impressions: Differences due to perceivers, trait dimensions, and target behaviors. Journal of Personality and Social Psychology, 56, 493–505. Paulhus, D. L., & Reynolds, S. (1995). Enhancing target variance in personality impressions: Highlighting the person in person perception. Journal of Personality and Social Psychology, 69, 1233–1242. Peng, K., & Nisbett, R. E. (1999). Culture, dialectics, and reasoning about contradiction. American Psychologist, 54, 741–754. Sadler, P., & Woody, E. (2003). Is who you are who you’re talking to? Interpersonal style and complementarity in mixed-sex interactions. Journal of Personality and Social Psychology, 84, 80–96. http://dx.doi.org/10.1037/0022-3514.84.1.80. Saucier, G. (1994). Mini-markers: A brief version of Goldberg’s unipolar big-five markers. Journal of Personality Assessment, 63, 506–516. Schwartz, S. H. (2002). Country scores for Schwartz value dimensions. Unpublished raw data. Shweder, R. A. (1991). Thinking through culture: Expeditions in cultural psychology. Cambridge, MA: Harvard University Press. Singelis, T. M. (1994). The measurement of independent and interdependent selfconstruals. Personality and Social Psychology Bulletin, 20, 580–591. Spencer-Rodgers, J., Srivastava, S., Boucher, H. C., English, T., Paletz, S. B., Wang, L., & et al. (2010). The Dialectical Self Scale. University of California, Santa Barbara, Unpublished. Spencer-Rodgers, J., Williams, M. J., & Peng, K. (2010b). Cultural differences in expectations of change and tolerance for contradiction: A decade of empirical research. Personality and Social Psychology Review, 14, 296–312. http:// dx.doi.org/10.1177/1088868310362982. Spirrison, C. L., & Choi, S. (1998). Psychometric properties of a Korean version of the Revised NEO Personality Inventory. Psychological Reports, 83, 263–274. Suh, E. M. (2002). Culture, identity consistency, and subjective well-being. Journal of Personality and Social Psychology, 83, 1378–1391. Terracciano, A., Costa, P. T., Jr., & McCrae, R. R. (2006). Personality plasticity after age 30. Personality and Social Psychology, 32, 999–1009. Trapnell, P. D. (1994). Openness versus intellect: A lexical left turn. European Journal of Personality, 8, 273–290. van de Vijver, F., & Leung, K. (1997). Methods and data analysis for cross-cultural research. Thousand Oaks, CA: Sage. Watson, D., Hubbard, B., & Wiese, D. (2000). Self-other agreement in personality and affectivity: The role of acquaintanceship, trait visibility, and assumed similarity. Journal of Personality and Social Psychology, 78, 546–558. Wood, D., & Roberts, B. W. (2006). Cross-sectional and longitudinal tests of the Personality and Role Identity Structural Model (PRISM). Journal of Personality, 74, 779–809. http://dx.doi.org/10.1111/j.1467-6494.2006.00392.x. Yamaguchi, S. (1994). Collectivism among the Japanese: A perspective from the self. In U. Kim, H. C. Triandis, C. Kagitcibasi, S. C. Choi, & G. Yoon (Eds.), Individualism and collectivism: Theory, method, and applications (pp. 175–188). Thousand Oaks, CA: Sage. Yang, J., McCrae, R. R., Costa, P. T., Jr., Dai, X., Yao, S., Cai, T., et al. (1999). Crosscultural personality assessment in psychiatric populations: The NEO-PI-R in the People’s Republic of China. Psychological Assessment, 11, 359–368. Yik, M. S. M., Bond, M. H., & Paulhus, D. L. (1998). Do Chinese self-enhance or selfefface? It’s a matter of domain. Personality and Social Psychology Bulletin, 24, 399–406.