Prediction and Cross-Situational Consistency of Daily Behavior across Cultures: Testing Trait and Cultural Psychology Perspectives

Abstract

Trait and cultural psychology perspectives on the cross-situational consistency of behavior, and the predictive validity of traits, were tested in a daily process study in the United States (N = 68), an individualistic culture, and the Philippines (N = 80), a collectivistic culture. Participants completed the Revised NEO Personality Inventory (Costa & McCrae, 1992) and a measure of self-monitoring, then reported their daily behaviors and associated situational contexts for approximately 30 days. Consistent with trait perspectives, the Big Five traits predicted daily behaviors in both cultures, and relative (interindividual) consistency was observed across many, although not all, situational contexts. The frequency of various Big Five behaviors varied across relevant situational contexts in both cultures and, consistent with cultural psychology perspectives, there was a tendency for Filipinos to exhibit greater situational variability than Americans. Self-monitoring showed some ability to account for individual differences in situational variability in the American sample, but not the Filipino sample.

“…an implicit promise of trait theories is to account for consistency across a range of situations.” (Moskowitz, 1994, p. 921)

“In personological formulations, the person should at least be stable over time and across situations that are very similar” (Diener & Larsen, 1984, p. 872)

“…the functional value of consistency is less clear for East Asian selves (Heine, 2001, p. 886).

“Personality is less evident in collectivist cultures than it is in individualistic cultures, because the situation is such a powerful determinant of social behavior.” (Triandis, 1995, p. 74).

As illustrated above, trait and cultural psychologists offer a range of views regarding the prediction and consistency of trait-relevant behavior. In discussing these differences in emphases or expectations, it is useful to distinguish trait and cultural psychology perspectives. While acknowledging the role of situational influences, trait psychologists anticipate a degree of cross-situational consistency of behavior, and predictive validity of traits, in all cultures. In contrast, some cultural psychologists have predicted that behavior will be less consistent across situations and that traits will have less predictive value in collectivistic cultures, where behavior is thought to be more influenced by contextual factors such as roles and relationships (Markus & Kitayama, 1998; Suh, 2002; Triandis, 1995). Both trait and cultural psychology perspectives would be correct if a degree of consistency and predictive validity is evident in all cultures, but more so in individualistic cultures than in collectivistic cultures. The goal of the present study was to test and integrate trait and cultural psychology perspectives on consistency and predictive validity in an individualistic culture, the United States, and a collectivistic culture, the Philippines (Church, 1987; Hofstede, 2001).

Trait and Cultural Psychology Perspectives on Consistency

A core assumption of trait theory is the existence of relatively stable trait attributes of individuals that predict their behavior across time and situations (Johnson, 1997; Kenrick & Funder, 1988). For example, in their Five Factor Theory, McCrae and Costa (1996) proposed that the Big Five dimensions of Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness to Experience are universal and heritable dimensions that should predict relevant behavior in all cultures. Some cultural differences in the behavioral manifestations of traits are still consistent with a trait perspective. In addition, trait prediction of behavior is generally viewed as probabilistic rather than deterministic. Nonetheless, from a trait perspective, a degree of cross-situational consistency in trait-relevant behavior is expected in all cultures (Church, 2000).

Whereas trait psychologists generally endorse a universalistic perspective on the importance of traits across cultures, cultural psychologists emphasize the socially constructed nature of personality and some have questioned the universality and predictive value of traits across cultures (Markus & Kitayama, 1998; Shweder, 1991). For example, Markus and Kitayama (1998) have argued that different assumptions underlie conceptions of personality in individualistic and collectivistic cultures. In individualistic cultures, a person is viewed as an independent or autonomous entity with a distinctive set of traits or attributes that determine behavior. In contrast, in collectivistic cultures, a person is viewed as interdependent with others, and behavior is a consequence of being responsive to one’s social roles and relationships. Although Markus and Kitayama (1991) acknowledge the existence of internal attributes such as personality traits, these attributes are viewed as situation-specific, elusive and unreliable, and not very powerful in predicting behavior.

More recently, Markus and Kitayama (2003) also contrasted the different implicit models of agency that underlie action in individualistic and collectivistic (or at least East Asian) cultures. They argued that a disjoint (independent) model of agency is predominant in individualistic cultures, and actions follow from the expression of individuals’ preferences, intentions, and goals. In contrast, in collectivistic cultures, a conjoint (interdependent) model of agency is predominant, and actions are responsive to social obligations and expectations and to situational contexts. Markus and Kitayama (2003) explicitly linked these two models of agency to differences in cross-situational consistency. Specifically, less cross-situational consistency is anticipated in collectivistic cultures, where the conjoint (interdependent) model of agency is predominant. Indeed, some cultural psychologists have argued that in collectivistic cultures cross-situational consistency may be viewed as inflexible, immature, and unresponsive to social contexts (Choi & Choi, 2002; Markus & Kitayama, 1994; Suh, 2002).

As Oishi, Diener, Scollon, and Biswas-Diener (2004) have noted, it is important to distinguish between relative and absolute consistency, particularly in cross-cultural studies. Relative or interindividual consistency reflects the extent to which the rank order of individuals on a given trait or behavior is stable across different situations, without regard to the absolute level of scores, which may vary across situations. Relative consistency is typically quantified in terms of correlation coefficients, computed across individuals between two or more situations, and addresses whether there are trait-like individual differences in behavior across situations. In contrast, absolute or intraindividual consistency refers to whether individuals tend to exhibit the same amount or level of the trait or behavior across different situations and addresses the impact of situational effects on behavior. With the exception of Oishi et al. (2004), cultural psychologists have not distinguished between relative and absolute consistency in their theorizing about cultural differences in consistency. It would be possible, for example, for relative consistency to be moderately high in all cultures, consistent with trait perspectives, and, at the same time, absolute consistency could be lower in collectivistic cultures than in individualistic cultures as a result of stronger situational effects. In the present study, we examined both relative and absolute consistency.

Finally, cultural psychologists seem to imply that the behavior of individuals in collectivistic cultures is less traited or consistent in general, that is, for all or most traits (Markus & Kitayama, 1991; Triandis, 1995). Snyder’s (1974, 1987) original self-monitoring theory postulated just such individual differences in general traitedness versus situational determination of behavior, and thus may provide a framework for understanding individual or cultural differences in consistency. Snyder (1974) hypothesized that high self-monitoring individuals, who are concerned about the situational and interpersonal appropriateness of their behavior, would be relatively “trait-free” and would thus show considerable cross-situational variability in their behavior. In contrast, low self-monitoring individuals, who are less sensitive to situational cues and more guided by internal dispositions, would be relatively “traited” in their behavior and show greater behavioral consistency. Snyder (1987) said little about cultural differences, but did suggest that self-monitoring would be higher, on average, in Japan, a collectivistic culture, than in the United States. In the present study, we investigated whether self-monitoring can account, in part, for individual and cultural differences in cross-situational consistency.

Available Evidence

Despite early pessimism about the cross-situational consistency of behavior (Mischel & Peake, 1982), researchers have shown that consistency correlations in Western studies can be substantial when construct-valid behavioral indicators of traits are assessed, particularly if these behaviors are aggregated to increase reliability (Borkenau, Mauer, Riemann, Spinath, & Angleitner, 2004; Deluty, 1985; Diener & Larsen, 1984; Funder & Colvin, 1991; Hettema & Hol, 1998; Moskowitz, 1982, 1994; Small, Zeldin, & Savin-Williams, 1983). For example, in a laboratory setting, Funder and Colvin (1991) reported average consistency correlations for single behaviors that ranged from .20 to .37 depending on the similarity of the situations compared. Consistency correlations for factor scores based on aggregated behaviors were mostly in the .40 to .60 range. Importantly, moderate consistency correlations have also been found using observer ratings of behavior in naturalistic settings (Deluty, l985; Moskowitz, 1982; Small et al., 1983) and in daily process or experience sampling studies (Diener & Larsen, 1984; Moskowitz, l994). For example, Moskowitz (1994) had respondents report their dominant, submissive, agreeable, and quarrelsome behaviors in their daily social interactions over three weeks. Aggregated behaviors for each trait exhibited moderate consistency across most communal situations (i.e., with male and female close friends and casual acquaintances), but low to modest consistency across agentic situations (i.e., with male and female supervisors, co-workers, and supervisees), perhaps due to the greater influence of roles in the agentic situations.

There is also ample evidence in Western studies that personality trait measures can predict observer ratings of behavior in the laboratory (Borkenau et al., 2004; Funder, Furr, & Colvin, 2000) and in experience sampling studies in naturalistic settings (Côté & Moskowitz, 1998; McAdams & Constantian, 1983; Moskowitz, 1994; Moskowitz & Zuroff, 2005; Wu & Clark, 2003). For example, Wu and Clark (2003) found that trait measures of aggression, exhibitionism, and impulsivity showed good convergent and discriminant correlations with daily behaviors associated with these traits, aggregated across two weeks of daily behavior recordings. In the experience sampling studies cited here, trait measures have correlated about .20 to .50 with relevant aggregated behaviors.

These Western studies provide a standard of comparison for cross-cultural studies of cross-situational consistency and predictive validity. However, they do not tell us whether consistency and predictive validity are greater in individualistic cultures than in collectivistic cultures. Most relevant for the present study is a cross-cultural experience sampling study conducted by Oishi et al. (2004). These researchers examined relative and absolute consistency of affective experiences in four samples of university students: a multi-ethnic American sample in Illinois, Hispanic Americans in California, and students in Japan and India. Participants rated their positive affect (PA) and negative affect (NA) and noted the situational context (alone, with a friend, classmate/coworker, romantic partner, stranger, or family member) at five random moments each day for seven days. Oishi et al. found evidence of relative consistency in all four cultural groups, supporting the existence of trait-like individual differences in affective experience. At the same time, they found cultural differences in absolute consistency. Specifically, the Japanese, Hispanic, and Indian samples exhibited greater within-individual variability across situational contexts than the Illinois reference group for both PA and NA (with the exception of NA for the Indian sample). Oishi et al.’s results are consistent with our expectations based on an integration of trait and cultural psychology perspectives. However, the study only addressed the cross-situational consistency of affect, not behavior more generally.

Other cross-cultural studies provide only indirect evidence of cultural differences in cross-situational consistency or predictive validity. Suh (2002) found that trait ratings were less consistent across roles for Koreans than for Americans. To the extent that perceptions of cross-role trait consistency reflect actual behavioral consistency in these roles, this finding supports cultural psychology hypotheses regarding behavioral consistency. English and Chen (2007) found that Asian Americans exhibited less consistency in trait ratings than European Americans across relationship situations, but not across situational contexts that were based on location. However, in a similar study, Church, Anderson-Harumi et al. (2007) did not find a reliable pattern of greater cross-role consistency in trait ratings in individualistic cultures (United States, Australia) as compared to collectivistic cultures (Mexico, Philippines, Malaysia, and Japan), although the Japanese sample did tend to average lower than the other cultural groups. Finally, if we assume that raters in studies of interobserver agreement encounter target individuals in different situations, then cross-cultural studies of interobserver agreement are relevant. For example, Malloy, Albright, Díaz-Loving, Dong, and Lee (2004) found that Chinese, but not Mexicans, exhibited lower interobserver agreement in trait ratings than Americans. However, most studies have found comparable levels of interobserver agreement across cultures, which is more consistent with trait perspectives than cultural psychology perspectives (Church, Katigbak, del Prado, Valdez-Medina et al., 2006; Heine & Renshaw, 2002; McCrae et al., 2004; Spirrison & Choi, 1998; Yik, Bond, & Paulhus, 1998).

Although self-monitoring theory would seem to provide a promising framework for understanding individual and cultural differences in cross-situational consistency, several cross-cultural studies of self-monitoring have yielded unexpected results. Gudykunst and colleagues (Gudykunst et al., 1989; Gudykunst, Yang, & Nishida, 1987) and Goodwin and Soon (1994) found higher, not lower, self-monitoring scores in individualistic cultures than in collectivistic cultures. Furnham and Capon (1983) and Hosch and Marchioni (1986) found no significant differences in self-monitoring between individualistic and collectivistic cultures. Only Hamid (1994), who used the Lennox and Wolfe (1984) measure, not the Snyder measures (Snyder, 1974; Snyder & Gangestad, 1986), found cultural differences that were entirely consistent with theory: Hong Kong Chinese were higher in self-monitoring than New Zealanders. In addition, Church, Katigbak, del Prado, Ortiz et al. (2006), using an adaptation of the Lennox and Wolfe (1984) measure, found that mean comparisons of the United States with the Philippines, Malaysia, and Japan were somewhat consistent with expectations, but not those involving Australia and Mexico. Finally, Suh (2002, Study 1), in a within-culture analysis of Americans using Snyder’s (1974) scale, found that cross-role identity consistency was negatively associated with the other-directedness component of self-monitoring (r = −.21, p < .001), as would be expected, but also positively associated with the extraversion component of self-monitoring (r = .27, p < .001).

A number of explanations for these mixed cross-cultural findings with the self-monitoring construct are plausible. Between-culture comparisons with Likert-type scales are controversial and may be subject to problems associated with measurement inequivalence, response biases, and reference-group effects (Heine, Lehman, Peng, & Greenholtz, 2002; Johnson, Kulesa, Cho, & Shavitt., 2005; van de Vijver & Leung, 1997). An alternative explanation is that Snyder’s self-monitoring construct and measures have evolved away from their original focus on individual differences in the dispositional versus situational determination of behavior toward a conception that is highly correlated with extraversion. Several researchers have argued that extraversion is not relevant to the original conception of self-monitoring and should be excluded (Briggs & Cheek, 1988; Church, Katigbak, del Prado, Ortiz et al., 2006; John, Cheek, & Klohnen, 1996). In the present study, we examined the relationship between self-monitoring and consistency, using an adaptation of the Lennox and Wolfe (1984) measure, which better corresponds to the original conception of self-monitoring as involving modification of behavior to be situationally and interpersonally appropriate.

Finally, we were unable to locate any studies that examined cross-cultural differences in the ability of trait measures to predict daily behavior in naturalistic settings. Church, Katigbak et al. (2007) found that Big Five trait scores predicted relevant behaviors equally well, on average, in American and Filipino samples, but the behavior ratings were retrospective in nature, rather than “on-line” behaviors assessed through experience sampling. There is also empirical evidence that the expression of one’s internal attributes or preferences is more valued or important to Americans than to East Asians (Kim & Sherman, 2007; Markus & Kitayama, 2003). However, these studies enable only indirect inferences about cultural differences in cross-situational consistency or the predictive validity of traits. Clearly, there is a need for more studies that directly compare cross-situational consistency and the ability of traits to predict everyday behaviors across cultures.

Overview of the Present Study

In the present study, we compared cross-situational consistency and the predictive validity of traits in the United States and the Philippines using self-reports of daily behavior in naturalistic environments. Daily process studies can have greater ecological validity than laboratory experiments. For example, experimental studies may artificially reduce behavioral consistency by limiting the individual’s opportunity to manifest his or her personality in the selection of situations (Ickes, Snyder, & Garcia, 1997; Tennen, Affleck, & Armeli, 2005). Researchers have emphasized the importance of the following elements in studies of cross-situational consistency: (a) the assessment of behaviors that are construct-valid indicators of the traits to be studied; (b) aggregation of these behavioral indicators for greater reliability; and (c) appropriate delineation of the situations across which consistency will be examined (Epstein, 1983; Funder & Colvin, 1991; Jackson & Paunonen, 1985).

To address these issues, we used behaviors identified as valid indicators of the Big Five dimensions in both the United States and Philippines in a previous study (Church, Katigbak et al., 2007). We aggregated daily reports of these behaviors within relevant situational contexts. Defining the situational contexts across which behavioral consistency will be examined has been one of the most challenging issues in studies of cross-situational consistency. Many authors have called for theory-guided approaches (e.g., Funder & Colvin, 1991; Mischel & Shoda, 1995; Moskowitz, 1994) and some efforts have been made to construct situation taxonomies (Saucier, Bel-Bahar, & Fernandez, 2007; Ten Berge & De Raad, 1999, 2002; Van Heck, 1984). However, in the absence of established taxonomies with broad utility, most researchers have based their selection of situations on rational considerations (e.g., Emmons, Diener, & Larsen, 1986; McAdams & Constantian, 1983; Moskowitz, 1982; Murtha, Kanfer, & Ackerman, 1996). Our approach to the selection of the situational distinctions involved a review of literature for relevant situational distinctions, and rational considerations regarding which situational distinctions might be central in moderating behaviors associated with each of the Big Five dimensions.

To summarize, we sought to test and integrate trait and cultural psychology perspectives on cross-situational consistency and the predictive validity of traits. To do so, we conducted a daily process study, in which participants reported culturally-relevant behaviors associated with each Big Five dimension on a daily basis for about 30 days and also reported selected aspects of the situational context for each behavior. If both trait and cultural psychology perspectives are correct, the Big Five traits should predict relevant daily behaviors to at least a moderate extent in both the United States and the Philippines, supporting trait perspectives, but more so in the United States, supporting cultural psychology perspectives. Similarly, cross-situational consistency of trait-relevant behavior should be evident in both the United States and Philippines, but more so in the United States. Based on Oishi et al.’s (2004) findings, we expected more modest cultural differences for relative or interindividual consistency (i.e., the stable rank-ordering of individuals across situations) than for absolute or intraindividual consistency (i.e., within-individual variability in absolute scores across situations). We also examined whether individual and cultural differences in absolute consistency could be accounted for, in part, by self-monitoring differences.

Method

Sample

United States

A total of 80 European American students at Washington State University took part in the daily process study. Of these, 68 (13 men, 55 women) met our minimum criterion of 20 complete daily reports. Mean age of the final sample was 20.75 (SD = 2.93). The majority of students were education majors (56%), with the rest majoring in a variety of fields. All college year-levels were represented in the sample.

Philippine sample

A total of 82 students from De La Salle University in Manila took part in the daily process study. Two participants did not provide valid data for the Revised NEO Personality Inventory and were eliminated. Sixty-seven (84%) of the remaining 80 participants identified themselves as ethnic Filipinos; the others identified themselves as Chinese Filipinos (15%) or Spanish Filipinos (1%). All 80 participants (32 men, 48 women) met our minimum criterion of 20 complete daily reports. Mean age of the final Philippine sample was 18.31 (SD = 1.54). Mean age was lower than in the American sample (t[146] = 6.47, p < .01), because Filipino students start college at a younger age (there are no middle schools in the Philippine educational system). The majority of students were education majors (68%), with the rest majoring in a variety of fields. All college year-levels were represented in the sample.¹

Instruments

Revised NEO Personality Inventory (NEO-PI-R)

Costa and McCrae’s (1992) 240-item NEO-PI-R was used to measure the Big Five traits of Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. Items are rated using a 5-point scale that ranged from strongly disagree to strongly agree. In the Philippine sample, we used the Filipino (Tagalog) version (McCrae, Costa, del Pilar, Rolland, & Parker, 1998), which was previously translated using standard backtranslation methods. Previous studies have demonstrated the generalizability and validity of the NEO-PI-R five-factor model in the Philippine setting (Church, Katigbak et al., 2007; Katigbak, Church, & Akamine, 1996; Katigbak, Church, Guanzon-Lapeña, Carlota, & del Pilar, 2002; McCrae et al., 1998). In the present study, α reliability estimates for the five domain scores ranged from .85 to .93 in the United States and from .75 to .90 in the Philippines.

Self-monitoring measure

Church, Katigbak, del Prado, Ortiz et al. (2006) adapted Lennox and Wolfe’s (1984) measures of Self-Monitoring and Concern for Appropriateness and derived two inversely related subscales labeled Self-Monitoring and Autonomous Self-Expression. In the present study, we obtained a single self-monitoring score by reverse-keying the Autonomous Self-expression items. Thirty-eight items were administered and assessed the self-perceived tendency to modify or adapt self-presentation to be situationally appropriate (e.g., “In social situations, I alter my behavior if I feel that something else is called for”) versus the tendency to express one’s true self across situations (e.g., “My behavior with others reflects my real personality, not how others think I should behave”). Items were rated using a 6-point scale that ranged from strongly disagree to strongly agree. The English version was previously translated into Filipino (Tagalog) using the backtranslation method and good structural equivalence across the two cultures was demonstrated using confirmatory factor analyses (Church, Katigbak, del Prado, Ortiz et al., 2006). In the present sample, α reliability was .91 in the United States and .81 in the Philippines.

Daily Behavior Checklist (DBQ)

The DBQ lists 50 behaviors, 10 for each of the Big Five traits (see Appendix). At the end of each day, participants checked “yes” or “no” for each behavior to indicate whether or not they had performed the behavior that day. For behaviors they had performed, participants also checked boxes that indicated the situational context of the behaviors. In some cases, respondents checked more than one situational context, indicating, for example, that they had performed a particular behavior in more than one context that day (e.g., a particular extroverted behavior with both friends and family). To ensure that the behaviors were valid indicators of the respective Big Five dimensions in both cultures, we selected behaviors from a recent act-frequency study conducted in the United States and the Philippines (Church, Katigbak et al., 2007). Drawing on a variety of sources (e.g., Angleitner, Buss, & Demtröder, 1990; Botwin & Buss, 1989; Moskowitz, 1994; Wu & Clark, 2003), Church, Katigbak et al. (2007) identified 198 behaviors or acts that were likely indicators of the Big Five dimensions. Participants in that study filled out the NEO-PI-R and rated how often they had done each of the 198 behaviors during the past month using a five-point scale that ranged from never to very often. For the present study, we used the correlations of the NEO-PI-R domain scores with each behavior in that previous study to select 10 behaviors that were good indicators of each Big Five dimension in both cultures. In the American and Filipino samples, respectively, the mean correlations in absolute value relating each Big Five dimension to the selected behaviors were as follows: for Extraversion, .38 and .44; for Agreeableness, .33 and .34; for Conscientiousness, .33 and .42; for Neuroticism, .45 and .38; and for Openness to Experience, .31 and .31. These correlations indicate that the behaviors in the DBQ have similar levels of construct validity in the two cultures, an important requirement for studies of cross-situational consistency (Jackson & Paunonen, 1985).

Our selection of situational dimensions involved a review of literature and rational considerations. The left-most column of Table 1 shows the situational dimensions and categories used for each Big Five trait. Like many researchers (e.g., Diener & Larsen, 1984; McAdams & Constantian, 1983; Moskowitz, 1994; Oishi et al., 2004; Saucier et al., 2007), we deemed it important to include a relationship dimension for the affective and interpersonal Big Five traits of Neuroticism, Extraversion, and Agreeableness. Situational differences associated with status differentials were also considered important for Big Five Neuroticism and Agreeableness (see also Moskowitz, 1994), particularly because status differentials might have greater impact in cultures such as the Philippines, where deference to authority is valued (Church, 1987). A number of researchers view the task setting or location, which addresses the “what” and “where” of behavior, to be an important situational distinction (Emmons et al., 1986; Furr & Funder, 2004; Murtha et al., 1996; Saucier et al., 2007). For our student samples, we viewed this situational dimension as most relevant for the Big Five Openness to Experience and Conscientiousness dimensions. We also included an intellectual demands distinction for these two Big Five dimensions, recalling Shoda, Mischel, and Wright’s (1993) differentiation of situations based on the demands they make on the individual’s psychological competencies.

Table 1.

Mean Proportion of Days each Situational Category Provided the Context for Reported Big Five Behaviors

	United States		Philippines		ANOVA results
Situational categories	Mean	SD	Mean	SD	F(1, 146)	Partial η2
Neurotic behaviors
Relationship
Alone	.59	.22	.69	.21	8.89**	.06
Family	.13	.12	.12	.15	.02	.00
Romantic partner	.20	.23	.07	.16	16.22**	.10
Stranger	.03	.05	.01	.03	14.24**	.09
Acquaintance	.10	.10	.07	.12	3.29	.02
Friend	.31	.22	.23	.21	5.19	.03
Status
Low status	.02	.03	.01	.04	.15	.00
Same status	.45	.22	.34	.26	7.38**	.05
High status	.10	.11	.04	.10	12.60**	.08
Extraverted behaviors
Location
School	.45	.12	.69	.12	158.71**	.52
Work	.17	.17	.05	.14	20.18**	.12
Home	.83	.20	.70	.26	11.27**	.07
Recreation	.64	.24	.47	.23	18.04**	.11
Relationship
Alone	.28	.27	.18	.22	5.78	.04
Family	.37	.26	.62	.31	28.41**	.16
Romantic partner	.46	.41	.18	.28	24.30**	.14
Stranger	.31	.18	.18	.20	17.88**	.11
Acquaintance	.50	.24	.48	.30	.25	.00
Friend	.82	.13	.83	.13	.09	.00
Openness-to-experience behaviors
Location
School	.24	.14	.53	.19	104.61**	.42
Work	.07	.10	.03	.08	10.33**	.07
Home	.66	.23	.68	.25	.45	.00
Recreation	.36	.24	.32	.23	.92	.01
Intellectual demands
Difficult	.41	.23	.54	.31	9.01**	.06
Easy	.72	.23	.71	.26	.19	.00
Disagreeable behaviors
Relationship
Alone	.18	.18	.29	.27	8.19**	.05
Family	.09	.12	.15	.21	3.88	.03
Romantic partner	.14	.17	.06	.15	8.08**	.05
Stranger	.04	.09	.02	.10	1.15	.01
Acquaintance	.07	.11	.10	.17	1.55	.01
Friend	.28	.22	.31	.26	.88	.01
Status
Low status	.04	.07	.03	.12	.02	.00
Same status	.40	.23	.42	.28	.34	.00
High status	.08	.12	.06	.15	1.00	.01
Conscientious behaviors
Location
School	.36	.14	.56	.18	59.77**	.29
Work	.12	.13	.05	.13	11.13**	.07
Home	.71	.19	.80	.19	8.34**	.05
Recreation	.27	.22	.26	.21	.12	.00
Intellectual
Difficult	.38	.19	.57	.27	24.46**	.14
Easy	.76	.17	.74	.25	.24	.00

^**p < .01.

For each participant, average daily scores for each behavior item on the DBQ were obtained by summing the number of times the behavior was checked across all of the participant’s daily forms, then dividing by the number of daily forms completed. Total Big Five behavior scores were then computed by averaging these average daily scores across the 10 items (behaviors) associated with each Big Five trait. A few behaviors that defined the opposite pole of particular Big Five traits, were reverse-keyed (i.e., assigned a value of −1 rather than 1) in these computations (see Appendix). All of the selected behaviors in the Agreeableness domain were indicators of the negative pole of this domain (i.e., disagreeable behaviors). They were scored such that higher scores indicated greater Disagreeableness (i.e., they were not reverse-keyed). Disagreeable behaviors were used because the correlations with the NEO-PI-R Agreeableness scale were more equivalent across the two cultures for disagreeable than agreeable behaviors in the study used to select the behaviors (Church, Katigbak et al., 2007). The α reliabilities for the total Big Five behavior scores in the United States and Philippines, respectively, were as follows: for Neuroticism, .72 and .77; for Extraversion, .79 and .73; for Openness to Experience, .67 and .71; for Disagreeableness, .78 and .90; and for Conscientiousness, .61 and .75. Overall, the α values indicate that the behaviors associated with particular Big Five traits cohered fairly well.

Separate behavior scores were also computed for each situational category (e.g., extraverted behaviors at school, extraverted behaviors at work, etc.). Daily Big Five scores were first computed for each situational category by summing the relevant Big Five behaviors (items) reported in that situation for a given day. These daily situation scores for the respective Big Five traits were then averaged across all completed daily forms. For each participant, the resulting scores reflect the average number of behaviors for a particular Big Five dimension that the participant reported per day in a given situation.

Procedure

In both cultures, participants were recruited in regular classes and signed up for one of several group training sessions. During the training sessions, informed consent was obtained, and procedures for filling out and returning the Daily Behavior Questionnaires (DBQs) were explained. Students filled out their first daily form during the training session so that the research team could answer any questions that might arise. Participants were instructed to fill out one form at the end of each day or after all daily activities had been completed. They were told that it was better to skip a day, if necessary, than to fill out a form late for a previous day. Participants dated each form they completed. Participants were asked to drop off completed forms to the research laboratory or the designated drop box at least three times a week, although some participants did so less frequently. A member of the research team monitored the submission of forms and e-mailed or text-messaged participants who needed a reminder. Participants were informed that the daily behavior recording would continue for 30 days and that they would need to complete at least 25 DBQs to receive full compensation for study participation. American participants received $50 for completing the study, while Filipino participants received the peso equivalent of $30, consistent with the lower cost of living in the Philippines. As the final activity during the training session, participants filled out a short demographic form and the Revised NEO Personality Inventory. Some students were also able to fill out the self-monitoring measure during the approximately 1½ hour training session, while other participants were allowed to fill out the measure on their own time and turn it in to the researchers with the first submission of DBQ forms.

Results

Descriptive Statistics for Daily Behavior Questionnaires

The number of DBQ forms completed by participants ranged from 22 to 35 (M = 30) in the United States and from 26 to 33 (M = 30) in the Philippines. The number of calendar days participants needed to complete their forms ranged from 22 to 66 (M = 33) in the United States and from 30 to 65 (M = 47) in the Philippines. Thus, more days were skipped in the Philippine sample, but the behavior scores in the two cultural samples were based on the same average number of days. A MANOVA comparing Filipinos who took more versus fewer days to complete their forms, based on a median split of the number of calendar days for completion, revealed no significant differences in total Big Five behavior scores between the two groups (Wilks’ Λ = .90, F[5, 74] = 1.66, p = .15). A cross-tabulation relating cultural group to day of the week that the DBQs were completed revealed no cultural differences in the percentage of DBQ forms completed on any given day of the week (χ²[6] = 1.49, p = .96).

Table 1 shows the mean proportion of days for which each situational category provided the context for the reported Big Five behaviors. These mean proportion values provide some indication of how regularly the various situations were encountered or provided the context for the Big Five behaviors reported. For example, extraverted behaviors were reported in the school, home, and recreation settings fairly regularly (i.e., on about 45–83% of the days across the two cultures). In contrast, extraverted behaviors were reported relatively infrequently in work settings, probably, in part, because many of the students spend few, if any, hours working each day.

We computed analyses of variance to compare the two cultures’ mean proportions in Table 1. In 21 of 40 ANOVAs, statistically significant cultural differences were found, but most of the effect sizes (partial η² values) were modest (α was set at .01 to reduce experiment-wise error). Focusing on the largest and most consistent differences, Filipinos, as compared to Americans, mentioned the school setting more regularly as the context for extraverted, openness-to-experience, and conscientious behaviors. Filipinos also mentioned intellectually challenging situations—which may also emerge primarily in school settings—more regularly than Americans. In contrast, romantic partners provided a context for neurotic, extraverted, and disagreeable behaviors more regularly for Americans than Filipinos. Both cultural groups, especially Filipinos, mentioned the work setting infrequently. Overall, however, the two cultures were more similar than different in the regularity (i.e., proportion of days) that different situations were reported as contexts for the relevant Big Five behaviors. Indeed, the correlation between the mean proportion values for the two cultures, computed across all 40 situational categories in Table 1, was .90.

Trait Prediction of Daily Behavior across Cultures

Both trait and cultural psychology perspectives would be supported if the Big Five traits predicted daily behavior in both cultures, but more so in the United States than in the Philippines. Table 2 shows the Pearson correlations in each culture relating the NEO-PI-R domain scores and the total Big Five behavior scores obtained with the DBQ. Also shown are the adjusted multiple correlation values (Adj R) obtained when we regressed each Big Five behavior score onto the NEO-PI-R domain scores simultaneously. Consistent with trait perspectives, the Adj R values, which adjust for sample size and number of predictors, indicate that the Big Five traits as a group were able to predict daily behaviors to a moderate extent in both cultures. In addition, all but one of the convergent correlations relating a particular Big Five trait to the associated behavior score was statistically significant. The exception was the convergent correlation for the Agreeableness dimension in the American sample, which was marginally significant (p < .07). In contrast, there was no consistent support for the cultural psychology hypothesis that trait-behavior relationships are stronger in the individualistic culture than in the collectivistic culture. In particular, although the trait-behavior correlation for Extraversion was higher in the United States, the correlations for Agreeableness and Conscientiousness were actually higher in the Philippine sample.²

Table 2.

Pearson Correlations relating NEO-PI-R Domain Scores to Total Big Five Behavior Scores

	NEO domain scores
Behavior scores	N	E	O	A	C	Adj R
United States
Neurotic	.53**	−.06	.27*	−.09	−.22*	.51**
Extraverted	−.07	.45**	.02	.22*	.16	.40**
Open to experience	−.02	.14	.29**	−.04	−.06	.21
Disagreeable	.18	.24*	−.08	−.18	−.04	.40**
Conscientious	−.15	.29**	−.05	−.05	.33**	.40**
Philippines
Neurotic	.52**	−.02	.15	−.05	−.20*	.50**
Extraverted	−.08	.25*	.18	−.17	.15	.23
Open to experience	.03	.23*	.23*	.05	.18	.25
Disagreeable	.25*	.05	.11	−.45**	−.13	.47**
Conscientious	−.34**	.06	.04	.24*	.53**	.52**

^*p < .05.

^**p < .01.

There were also some statistically significant off-diagonal correlations in Table 2, although they were all smaller than the relevant diagonal correlations. As noted by McCrae and Costa (1996) in describing Five Factor Theory, behaviors can reflect more than one trait (e.g., reading a book can indicate an introverted need for solitude, as well as an Openness-related desire for intellectual stimulation). That is, trait theory does not predict that the off-diagonal correlations will be zero. For example, the one off-diagonal correlation that replicated across the two cultures—Conscientiousness trait scores modestly predicted (inversely) Neurotic behaviors—is probably sensible. In both cultures, students who were organized, self-disciplined, and dutiful (i.e., high in Conscientiousness) probably reduced their likelihood of experiencing daily stress or anxiety about schoolwork.

Relative Cross-Situational Consistency across Cultures

Both trait and cultural psychology perspectives would be supported if relative cross-situational consistency was evident in both cultures, but more so in the United States than in the Philippines. Table 3 shows, for each culture, the cross-situational consistency correlations between pairs of situational categories for each Big Five trait. For example, the significant consistency correlation relating neurotic behaviors in the alone and family contexts in the United States reveals that American participants who reported more neurotic behaviors when alone also tended to report more neurotic behaviors with family members (r = .30, p < .05).³ A few situational contexts were eliminated from these analyses because they were rarely mentioned in either culture (see low mean proportions in Table 1). In particular, we discarded the Status dimension, because so few participants in either culture reported neurotic or disagreeable behaviors with low or high status persons. In addition, we eliminated the stranger relationship category for neurotic and disagreeable behaviors, but retained it for extraverted behaviors.

Table 3.

Relative Consistency (Pearson Correlations) across Pairs of Situations in Two Cultures

Situational categories	United States	Philippines
Neurotic behavior
Relationship
alone-family	.30*	.16
alone-romantic partner	.28*	.03
alone-acquaintance	.19	.06
alone-friend	.36**	.05
family-romantic partner	.38**	.21
family-acquaintance	.21	.50**
family-friend	.21	.63**
romantic partner-acquaintance	.15	.20
romantic partner-friend	.04	.19
Acquaintance-friend	.47**	.49**
Extraverted behavior
Location
school-work	.38**	−.21
school-home	.50**	.49**
school-recreation	.41**	.18
work-home	.23	.05
work-recreation	.19	−.07
home-recreation	.26*	.16
Relationship
alone-acquaintance	.18	.09
alone-friend	−.03	−.01
alone-romantic partner	.03	−.13
alone-family	.08	.00
alone-stranger	.05	.01
acquaintance-friend	.69**	.31**
acquaintance-romantic partner	.04	.02
acquaintance-family	.22	.08
acquaintance-stranger	.82**	.39**
friend-romantic partner	−.10	.14
friend-family	.18	.41**
friend-stranger	.67**	.40**
romantic partner-family	.42**	.00
romantic partner-stranger	−.02	.02
family-stranger	.26*	.27*
Openness-to-experience behavior
Location
school-work	.41**	.22
school-home	.54**	.43**
school-recreation	.49**	.37**
work-home	.13	.10
work-recreation	.25*	.10
home-recreation	.40**	.28*
Intellectual demands
difficult-easy	.12	−.17
Disagreeable behavior
Relationship
alone-family	.42**	.17
alone-romantic partner	.33**	.06
alone-acquaintance	.20	.09
alone-friend	.23	.33**
family-romantic partner	.26*	.30**
family-acquaintance	.14	.52**
family-friend	.18	.58**
romantic partner-acquaintance	.11	.10
romantic partner-friend	−.12	.31**
acquaintance-friend	.38**	.56**
Conscientious behavior
Location
school-work	.18	.01
school-home	.65**	.70**
school-recreation	.37**	.43**
work-home	.19	.04
work-recreation	.27*	−.05
home-recreation	.56**	.34**
Intellectual demands
difficult-easy	−.07	−.15

Note. Consistency correlations that are significantly greater than their counterpart in the other culture are indicated in boldface (z tests on independent sample correlations; Hays, 1973, pp. 663–664).

^*p < .05.

^**p < .01.

There was evidence of relative consistency across many, but not the majority, of the situational contexts. In both cultures, many of the consistency correlations were substantial in size, consistent with trait perspectives. Consistency correlations that were significantly greater than their counterparts in the other culture are indicated in boldface in Table 3 (z tests on independent sample correlations; Hays, 1973, pp. 663–664). There were some indications of greater overall consistency in the American sample, as compared to the Filipino sample. In the American sample, 27 of 55 consistency correlations (49.0%) were statistically significant, compared to 21 of 55 (38.1%) in the Filipino sample. In addition, the overall mean of the 55 consistency correlations in Table 3 was somewhat larger in the American sample (mean r = .26), as compared to the Filipino sample (mean r = .21). On the other hand, of the 10 consistency correlations that were significantly greater in one or the other culture (shown in boldface in Table 3), half were higher in the American sample and half were higher in the Filipino sample. There were clearly some situational contexts across which relative consistency was greater in the Filipino sample. Thus, overall support for cultural psychology perspectives was somewhat mixed. We consider some of the more definitive cultural differences in the Discussion section.

Situational Effects across Cultures

The presence of relative or interindividual consistency across situations is not incompatible with the existence of situational effects, or cultural differences in these effects. To test for situational effects we conducted repeated-measures ANOVAs in each culture. In each analysis, gender was a between-subjects independent variable and Big Five scores for a set of situational categories constituted the within-subjects or repeated factor. For example, in one analysis, neurotic behavior scores in the following relationship categories were compared: when alone, with acquaintances, with friends, with romantic partners, and with family members. In these analyses, we focused on the two most pertinent questions for our purposes. First, is there evidence of situational effects in daily reports of Big Five behaviors? Situational effects would be revealed by statistically significant main effects for the repeated or within-subjects factor representing the situational categories. Second, are there cultural differences in these situational effects, or the degree of absolute consistency, and if so, were the situational effects stronger in the Philippines than in the United States? We answered this second question by comparing the effect sizes (partial ŋ² values) associated with the situational effects in each culture.

In the repeated-measures ANOVAs, there were no statistically significant gender × situation interaction effects. Main effects for gender were found only in the American sample. On average, American women, as compared to American men, reported more extraverted behaviors (across location contexts, F[1, 66] = 7.85, p < .01, partial ŋ² = .11; across relationship contexts, F[1, 66] = 6.25, p < .02, partial ŋ² = .09) and more conscientious behaviors (across location contexts, F[1, 66] = 6.38, p < .02, partial ŋ² = .09; across contexts varying in intellectual demands, F[1, 66] = 7.77, p < .01, partial ŋ² = .11). Importantly, in both cultures, all of the main effects for situational category (i.e., the repeated factor) were statistically significant (p < .001), demonstrating the presence of situational effects for each Big Five trait across the relevant situational categories.

Table 4 shows the effect sizes in each culture for the Big Five behavior scores across the relevant situational categories. Most of the effect sizes would be considered quite large (Cohen, 1988). Five of 8 effect sizes were larger in the Filipino sample than in the American sample, whereas two effect sizes were larger in the American sample. The effect sizes for neurotic behaviors across relationship contexts were about the same in the two cultures. Filipinos exhibited larger situational variability across relationship and location contexts, whereas Americans exhibited more situational variability across intellectually demanding versus easy task contexts. These results seem consistent with English and Chen’s (2007) view that Asians exhibit situational variability primarily across situations involving different interpersonal relationships. This interpretation might also apply to the greater Filipino variability across location contexts, if the salient distinctions between location contexts are also, in part, interpersonal in nature.

Table 4.

Strength of Situational Effects (Partial ŋ² Values) for Big Five Behaviors

Big Five behaviors/ Situational categories	United States	Philippines
Neurotic behavior
Relationship	.40	.39
Extraverted behavior
Location	.49	.59
Relationship	.36	.60
Openness behaviors
Location	.47	.55
Intellectual demands	.36	.14
Disagreeable behaviors
Relationship	.13	.25
Conscientious behaviors
Location	.44	.63
Intellectual demands	.31	.14

Note. Effect sizes (partial ŋ² values) were obtained from repeated-measures ANOVAs, with scores for a particular Big Five trait across the associated situational categories as within-subjects (repeated) factors (see text).

Overall, these analyses showed that situational differences in Big Five behavior scores were the norm in both cultures. That is, behaviors associated with each Big Five dimension were exhibited more frequently in some situational contexts than others. There was also some tendency for the situational effects to be stronger in the Philippines than in the United States, although this was not the case for all situational distinctions. Thus, at the group level, the cultural psychology prediction of stronger situational effects in collectivistic cultures, as compared to individualistic cultures, received some support.

Cultural Differences in Absolute Consistency at the Individual Level

The analyses of situational effects in the previous section did not address the cross-situational variability of individuals and whether the two cultures differ in this individual-level variability. To quantify cross-situational variability for each participant we computed the standard deviation of their behavior scores across different situational contexts for each Big Five trait (e.g., see Fleeson, 2001; Eid & Diener, 1999; Oishi et al., 2004). There were eight SD indices for each participant, one for each Big Five trait and its associated situational dimensions (e.g., for neurotic behaviors across different relationship categories) (see Table 5). In a MANOVA with culture and gender as independent variables and the eight SD indices as dependent variables, the main effect for gender was not statistically significant (Wilks Λ = .90, F [8, 137] = 1.85, p < .08), nor was the culture × gender interaction effect (Wilks Λ = .92, F[8, 137] = 1.52, p = .15). The main effect for culture was statistically significant (Wilks Λ = .79, F[8, 137] = 4.60, p < .001). Table 5 shows the means, standard deviations, and follow-up ANOVA F tests for the cultural comparisons. In all eight comparisons, the Filipino sample averaged higher variability than the American sample. For 3 of the 8 comparisons the cultural mean differences were statistically significant. Thus, there was a trend for Filipinos to exhibit greater cross-situational variability than Americans in the manifestation of Big Five behaviors, although the effect sizes were generally modest.

Table 5.

Cultural Means and Standard Deviations of Participants’ Cross-Situational Variability (SD) Indices

	United States		Philippines
Big Five behaviours
Situational dimension	M	SD	M	SD	F(1, 144)	η2
Neurotic behaviors
Relationship	.50	.27	.54	.39	.73	.01
Extraverted behaviors
Location	1.34	.46	1.39	.51	2.78	.02
Relationship	1.41	.40	1.48	.52	2.27	.02
Openness behaviors
Location	.58	.30	.78	.36	1.26**	.07
Intellectual demands	.70	.56	.95	.69	.79	.02
Disagreeable behaviors
Relationship	.26	.16	.33	.24	2.75	.02
Conscientious behaviors
Location	.58	.31	.93	.55	24.88**	.15
Intellectual demands	.70	.56	.95	.69	8.21**	.05

^**p < .01.

Self-monitoring Effects

Finally, we examined whether self-monitoring—the tendency to modify or adapt behavior to be situationally appropriate—could account for individual or cultural differences in cross-situational consistency. Our results indicated that self-monitoring could not mediate cultural differences in consistency, because the two cultures did not differ significantly in self-monitoring (Baron & Kenny, 1986). The Americans averaged 3.68 (SD = .58) and the Filipinos averaged 3.59 (SD = .33), t[146] = 1.11, p > .05.

Within the American sample, self-monitoring scores did predict individual differences in cross-situational consistency (i.e., the SD indices) for 3 of 8 trait-situation combinations. Specifically, self-monitoring predicted individual variability for both neurotic and disagreeable behaviors across relationship categories (for neurotic behavior, r = .47, p < .01; for disagreeable behavior, r = .37, p < .01). To a lesser extent, self-monitoring also predicted variability in openness-to-experience behaviors across different locations (r = .25, p < .05). In addition, in these three cases, self-monitoring provided unique prediction of cross-situational variability beyond that provided by Big Five trait levels, indicating that the relationships between self-monitoring and situational variability were not an artifact of mean trait levels (Baird, Le, & Lucas, 2006; Eid & Diener, 1999). When these three individual variability scores were predicted in hierarchical regression analyses by the NEO-PI-R domain scores (entered in Step 1) and self-monitoring (entered in Step 2), the changes in R² values in Step 2 were statistically significant for variability in neurotic behaviors across relationship categories (ΔR² = .06, p < .01; self-monitoring β-weight = .29, p < .01) and for variability in disagreeable behaviors across relationship categories (ΔR² = .12, p < .01; self-monitoring β-weight = .39, p < .01), and marginally significant for variability in openness-to-experience behaviors across locations (ΔR² = .05, p < .06; self-monitoring β-weight = .26, p < .06). In contrast, self-monitoring failed to predict the cross-situational variability of Filipino participants. None of the eight correlations relating self-monitoring scores to the individual variability (SD) indices were statistically significant in the Philippines.

Discussion

We sought to test and integrate trait and cultural psychology perspectives on the cross-situational consistency of trait-relevant behavior and the predictive validity of traits. Trait perspectives were supported by several findings. First, measures of the Big Five traits predicted relevant behaviors to a moderate extent in both the United States, an individualistic culture, and the Philippines, a collectivistic culture. The convergent trait-behavior correlations were similar in size to those reported in Western daily process or experience sampling studies (Côté & Moskowitz, 1998; McAdams & Constantian, 1983; Moskowitz, 1994; Moskowitz & Zuroff, 2005; Wu & Clark, 2003). Second, there was evidence of relative or interindividual consistency in daily reports of Big Five behaviors across many, although not the majority, of situational contexts. The size of some of the cross-situational consistency correlations were comparable to those reported in previous studies in both laboratory (Funder & Colvin, 2001) and naturalistic (Diener & Larsen, 1984; Moskowitz, 1994) settings. The present daily process study appears to be the first to extend such findings to a collectivistic culture. Oishi et al. (2004) conducted a similar study on the cross-situational consistency of affect, not behavior, across cultures. Our results are consistent with trait theory and the existence in all cultures of individual differences in trait attributes that predict the daily behavior of individuals across time and situations (Johnson, 1997; Kenrick & Funder, 1988).

Support for cultural psychology perspectives was more mixed. There was no consistent tendency for Big Five traits to predict daily behaviors better in the United States than in the Philippines. There were some indications of greater relative (interindividual) consistency in the American sample overall, but across some situational contexts relative consistency was greater in the Philippines. We found significant situational effects for the reported frequency of relevant Big Five behaviors, and more of these effects were larger in the Filipino sample than in the American sample. In addition, when we examined absolute consistency at the level of individuals with the SD indices, we found a trend for Filipinos to exhibit greater cross-situational variability than Americans. Finally, self-monitoring showed some ability to predict individual differences in consistency beyond that provided by Big Five trait levels, but only in the American sample.

Although our results were not as consistent as those of Oishi et al. (2004), there were some similarities in the pattern of results in the two studies, and the results of both studies confirmed the value of distinguishing between relative and absolute consistency. Oishi et al. (2004) found considerable relative consistency in the experience of affect across relationship situations in the United States, Japan, and India. Similarly, we found evidence of relative consistency of Big Five behaviors in the United States and Philippines, although not across all situational contexts. At the same time, Oishi et al. (2004) found considerable situational effects on absolute levels of affective experience, as well as larger within-individual variability for Japanese, Hispanics, and (for negative affect) Asian Indians, as compared to Americans in Illinois. Similarly, we found large situational effects on absolute levels of Big Five behaviors and some tendency for Filipinos to exhibit greater cross-situational variability than Americans at both the group and individual levels. Based on the two studies, we conclude that global traits predict daily behavior and affect and account for relative consistency in both individualistic and collectivistic cultures, but also that situational effects are prevalent and cross-situational variability may be somewhat stronger in collectivistic cultures.

Some of the cultural differences in the relative consistency correlations observed in the present study may reflect cultural differences in the nature of the respective situations or their affordance for various Big Five behaviors. For example, Filipinos generally exhibited greater relative consistency than Americans for pairings of the family situation with the friend or acquaintance situations. The family context in the Philippines, more so than in the United States, may resemble the interpersonal interaction contexts experienced with friends or acquaintances because of the generally larger and extended nature of Filipino families (Church, 1987). In contrast, Americans exhibited greater relative consistency than Filipinos for pairings of the stranger situation with the friend or acquaintance situations for extraverted behaviors. Filipinos, as compared to Americans, may view interactions with strangers as more distinct from interactions with friends and acquaintances, requiring more guarded or reserved (i.e., less extraverted) behavior. Triandis (1995), among others, has noted that the distinction between ingroup (e.g., friends) and outgroup (e.g., strangers) behavior may be stronger in collectivistic cultures than in individualistic cultures.

In either culture, a plausible explanation for some of the lower consistency correlations—for example, those involving intellectually challenging versus easy situations, romantic partners versus other relationships, or work settings versus other locations—could be the following: The respective situation pairs could be less similar to each other psychologically than situation pairs with higher consistency correlations. Some researchers have demonstrated a positive relationship between situational similarity and cross-situational consistency, although they have not assessed this relationship in naturalistic settings (Champagne & Pervin, 1987; Furr & Funder, 2004; Klirs & Revelle, 1986; Lord, 1982). An alternative explanation for the lower consistency correlations involving the romantic partner and work settings could be that some participants had few interactions in these contexts.

Self-monitoring could not serve as a mediator of between-culture differences in this study, because the American and Filipino samples did not differ significantly in average self-monitoring scores. Some cross-cultural psychologists discourage between-culture mean comparisons with Likert scales, which can be confounded by cultural differences in response styles and reference group effects (Heine et al., 2002; Johnson et al., 2005). Indeed, our within-culture analyses were more promising, at least in the American sample. In that sample, self-monitoring scores predicted individual differences in within-individual consistency across some situational categories, even after controlling for Big Five trait levels. Self-monitoring was a better predictor of individual variability across relationship contexts than across the other situational dimensions in the study, which were less interpersonal in nature (i.e., location; intellectually challenging vs. easy situations). This is sensible because self-monitoring refers primarily to modification of interpersonal behavior to be socially appropriate, to maintain others’ approval, and in response to interpersonal cues.

The failure of self-monitoring to predict individual variability in the Filipino sample does not appear to be due to lack of reliability (α = .81) or validity of the self-monitoring measure in the Philippine setting (e.g., see Church, Katigbak, del Prado, Ortiz, et al., 2006).⁴ Thus, it may be a valid finding that Filipinos’ self-perceptions of their tendency to modify behavior is unrelated to their actual situational variability. If, as predicted by cultural psychologists, self-monitoring is more normative in collectivistic cultures than in individualistic cultures, Filipinos who are variable in their actual behavior may not perceive themselves to be particularly high in self-monitoring. Although speculative, this interpretation would be consistent with the reference-group phenomenon described by Heine et al. (2002), who proposed that respondents rate themselves on Likert-type scales in comparison to cultural norms. This interpretation would also be consistent with the lower variability in self-monitoring scores found in the Philippine sample (SD = .33), as compared to the American sample (SD = .58).

Some limitations of the study should be noted. For one, our list of Big Five behaviors, although construct-valid in both cultures, could not be comprehensive. Also, we did not attempt to identify indigenous Filipino indicators of the Big Five dimensions because they might not be equally valid across the two cultures. We believe the behaviors sampled are reasonably representative of the respective Big Five dimensions, but future studies could attempt to replicate our study with different sets of daily behaviors.

The ecological validity of the daily process method was a strength of the study, as was the large number of days for which participants reported their behavior. There are trade-offs, however, in using daily behavior reports rather than sampling particular events (e.g., Moskowitz, 1994) or random moments throughout the day (e.g., Oishi et al., 2004). On the one hand, daily reports have the potential advantage of greater comprehensiveness. Participants provide a record of all of the designated behaviors they performed at least once over the course of each day, rather than a more limited sampling of behaviors during particular events or moments. Over approximately 30 days, this should give a reasonable estimate of the frequency that individuals performed the Big Five behaviors in particular situations. On the other hand, some recall inaccuracies in the daily reports are possible. In addition, it is conceivable that respondents sometimes rated their typical or usual daily behaviors, rather than their actual behaviors for the day. Similarly, because both the trait and behavior measure were based on self-report, similar self-appraisal processes or shared method variance could inflate the trait-behavior correlations. However, given the specific and concrete nature of the behaviors assessed and the short (daily) recall period, we do not believe these possibilities were serious concerns.

Another feature of our naturalistic study was that individuals probably differed to some extent in the situations they sought out or encountered during the approximately 30 days of the study. As a result, the number of Big Five behaviors exhibited in particular situations could reflect both personality traits and situational opportunities or affordances. To the extent that individuals selected situational contexts based on their personality traits (e.g., extraverts sought out friends and acquaintances more than introverts did), this can be viewed as a positive and valid feature of the data (Ickes et al., 1997). However, it also leaves open the possibility that some of the observed individual and cultural differences in consistency reflected, in part, differences in the situations participants encountered. We sought to address this issue to some extent in Table 1 by reporting the regularity, or mean proportion of days, that various situations were reported, which was generally quite similar across the two cultures. Given the limitations of our daily reporting procedure, it would be useful to conduct a similar study in which participants reported their behaviors and the associated situational contexts at random moments throughout the day. We believe it is unlikely that these limitations had a significant impact on our primary conclusions, however.

Finally, because of the intensive nature of daily process studies, we sampled only one individualistic and one collectivistic culture, which reduces the definitiveness of our findings. Also, in our desire to sample a less frequently studied collectivistic culture (the Philippines), we did not include any East Asian or Confucian cultures, where cultural psychology hypotheses have been most frequently tested and best supported in previous studies (Heine, 2001; Markus & Kitayama, 1991). In some of our previous research testing cultural psychology perspectives, we have found greater support in comparisons involving the United States and Japan than in comparisons involving alternative individualistic (i.e., Australia) and collectivistic (i.e., Mexico, Philippines, Malaysia) cultures (Church, Katigbak, del Prado, Ortiz et al., 2006; Church, Anderson-Harumi et al., 2007; del Prado et al., 2007; see also Malloy et al., 2004).

Indeed, some cultural psychologists have proposed an alternative to individualism-collectivism as an explanatory variable in studies of consistency. These researchers have attributed cultural differences between Western and East Asian cultures to East Asian or Confucian dialecticalism (Choi & Choi, 2002; English & Chen, 2007; Peng & Nisbett, 1999; Schimmack, Oishi, & Diener, 2002). Peng and Nisbett (1999) described dialecticalism as a broad cognitive tendency or system of thought characterized by acceptance of inconsistency or contradiction, expectations of complexity and change, and holistic thinking. To date, this interpretation has been applied primarily in studies that have found lower self-concept consistency (Choi & Choi, 2002; English & Chen, 2007) and greater tendencies to simultaneously report affects of opposite or inconsistent valence in East Asian, as compared to European American, samples (e.g., Bagozzi, Wong, & Yi, 1999; Kitayama, Markus, & Kurokawa, 2000; Schimmack et al., 2002; Scollon, Diener, Oishi, & Biswas-Diener, 2005). However, an interpretation in terms of East Asian or Confucian dialecticalism might also be relevant in cross-cultural studies of the cross-situational consistency of daily behavior.

Concluding Remarks

In summary, we found good support for trait perspectives and more mixed support for cultural psychology perspectives regarding the prediction and consistency of trait-relevant behavior. Considering the results of this and some of our previous studies—and recent theory and research on East Asian or Confucian dialecticalism (e.g., English & Chen, 2007; Peng & Nisbett, 1999)—it is possible that cultural psychology hypotheses regarding the traitedness of self-concepts, self-enhancement biases, and cross-situational consistency may be more definitively or strongly supported in comparisons of the United States with East Asian cultures (e.g., China, Japan, Korea) than in comparisons of individualistic and collectivistic cultures more generally. Additional studies in a broader range of cultures will be needed to draw firmer conclusions in this regard. In the meantime, an integration of trait and cultural psychology perspectives—currently the two dominant theoretical approaches or emphases in cross-cultural studies—retains heuristic value in designing studies, formulating hypotheses, and making theoretical refinements in the study of personality across cultures.

Appendix

Big Five Daily Behaviors in the Daily Behavior Questionnaire
Neurotic behaviors
Felt anxious about work that needed to be done; Experienced a lot of stress; Acted moody; Felt sad; Put myself down; Complained about a problem I was having; Broke down when a problem arose; Felt jealous of someone; Kept an even mood during some difficultya; Gave in to a bad habit when I was nervous.
Extraverted behaviors
Talked a lot; Hugged someone; Went out to socialize; Chatted with strangers; Took the lead in organizing a project or activity; Expressed my own opinion; Smiled and laughed with others; Felt cheerful and happy; Mixed well at a social function; Introduced myself to someone new.
Openness-to-Experience behaviors
Enjoyed some art; Read a play or novel; Thought about my emotional reactions to something; Experienced some intense feelings; Listened to or read with interest a news story about another country; Discussed politics; Listened with interest to someone whose values or beliefs differed from mine; Listened with interest to someone whose values or beliefs differed from mine; Tried out a new activity for the sake of doing something different; Discussed an issue from all points of view; Read poetry.
Disagreeable behaviorsb
Said/did something to hurt someone’s feelings; Criticized someone; Made a decision without consulting the others involved; Yelled at someone; Accused someone of talking behind my back; Got into an argument; Had doubts about someone’s honesty; Felt someone betrayed my trust; Got what I wanted by withholding the truth from someone; Made a rude comment about a person passing.
Conscientious behaviors
Checked out every detail on a task I completed; Put my clothes neatly away; Did not put away my things when I finished a project or an activitya; Did poorly on an assignment or exama; Did an important task well; Finished a task on time; Forgot about an appointmenta; Finished everything I planned to do today; Skipped class, work, or other scheduled activities on a whima; Reflected on the consequences of an action before going ahead with something.

Note. These behaviors comprised the Daily Behavior Questionnaire in both the United States and Philippine samples, but the instrument was administered in Filipino (Tagalog) translation in the Philippines.

^aReverse-keyed behaviors scored as −1 rather than 1.

^bAll of the selected behaviors in the Big Five Agreeableness domain were disagreeable behaviors, so we labeled the behavior domain as such.