Classify the qualities as relating to either collectivist or individualist societies.

In the density distributions approach, trait manifestations in everyday behavior are characterized in terms of the personality states (e.g., state extraversion, state agreeableness) that are reflected in enacted behaviors (Fleeson, 2001). Individual differences are characterized in terms of the distributional properties of these personality states over time. While all individuals exhibit considerable within-individual variability in these states over time (which can be represented by the standard deviation of their density distributions), an individual’s general trait level can be represented by the mean of their density distribution of personality states. One advantage of characterizing daily behaviors in terms of personality states is that the researcher does not need to identify and assess the specific behaviors that might reflect the trait of interest, as is done, for example, in the act frequency approach (Buss and Craik, 1983, Church et al., 2007). For example, rather than reporting the specific extraverted behaviors performed in a given situation (e.g., interacting with others, expressing one’s opinion), participants indicate their level of extraversion by rating their sociability, talkativeness, and so forth in the situation in much the same way they rate their general trait of extraversion. That is, behaviors are characterized in terms of momentary or short-term personality states that have much the same content as the associated traits (Fleeson, 2001).

There is substantial evidence in Western cultures that personality traits are manifested in corresponding behaviors in the laboratory (Borkenau et al., 2004, Funder et al., 2000) and in naturalistic settings (Church et al., 2008, Mehl et al., 2006, Moskowitz and Zuroff, 2005, Wu and Clark, 2003). Convergent correlations between traits and aggregated behaviors have ranged from about .20–.50 in these studies. Most relevant for the present investigation and the density distributions approach are studies that have examined the ability of traits to predict personality states in experience sampling studies. For example, in a Canadian sample, Heller, Komar, and Lee (2007) found good convergent and discriminant relationships between Big Five traits and personality states aggregated over 10 days of three reports per day. Fleeson and Gallagher (2009) summarized the results of 15 studies conducted in US samples. In a mega-analysis combining all of the data sets, the Big Five traits provided moderate prediction of single personality states (r range = .18–.37) and strong prediction of both the mean (r range = .42–.56) and maximum (r range = .34–.54) of the density distributions for the corresponding Big Five personality states. These results indicate that when behaviors are aggregated over time, trait-behavior correlations can exceed the .30–.40 level that has been proposed as the upper limit of personality prediction (Mischel, 1968, Ross and Nisbett, 1991). Fleeson and Gallagher also found that the maximum state level, but not the median, mode, and minimum of the density distributions correlated significantly with the corresponding trait when the mean state level was controlled. Based on these results, Fleeson and Gallagher concluded that “trait standing, as assessed by questionnaire, implicates primarily individuals’ mean trait manifestation but also provides independent information about individuals’ maximum trait manifestation in behavior” (p. 1106).

Although limited in number, available studies in North American samples provide cogent support for the density distributions approach to traits and provide a standard of comparison for cross-cultural studies.1 However, only cross-cultural studies can tell us whether traits predict personality states or daily behaviors better in some cultures than others, a question that we return to later. Consistent with trait perspectives, we expected personality traits to predict corresponding personality states to at least a moderate degree in all cultures.

Although the relationship between traits and affect is well-established in Western studies (Costa and McCrae, 1980, DeNeve and Cooper, 1998, Watson and Clark, 1997), much of the research has involved one-time global reports rather than reports of daily affect in experience sampling studies. Drawing on the density distributions approach, Wilt et al. (2012) proposed a dynamic mediation model of trait-affect relationships, in which the link between trait extraversion and positive affect is due, at least in part, to the tendency of trait extraverts to enact more extraverted behaviors (i.e., experience more extraversion states) in their daily lives, which in turn leads to more positive affect states. Wilt et al. contrasted their model with prominent temperamental or constitutional explanations of trait-affect relationships (e.g., Depue and Collins, 1999, Gross et al., 1998, Rosenberg, 1998), which link individual differences in positive affect to stable individual differences in neurobiological substrates, and seem to imply that the extravertedness of one’s daily behavior is irrelevant to the amount of positive affect experienced.

As applied to extraversion and positive affect (PA), the dynamic mediation model predicts that “(a) individuals with higher standings on measures of trait extraversion tend to enact more extraverted states, (b) extraverted states cause PA states, and (c) the accumulation of PA states leads to a higher overall standing on trait measures of PA” (p. 1209). The first link in the model is consistent with the density distribution approach to traits (Fleeson, 2001, Fleeson and Gallagher, 2009). The second link is supported by US studies in which extraversion states have predicted experienced positive affect (Fleeson et al., 2002, Heller et al., 2007, Wilt et al., 2012). Although a reverse direction of causation from PA to extraversion states is possible, a causal path from extraversion to PA is supported by studies that have found increases in positive or negative affect after experimental manipulation of state extraversion or neuroticism (e.g., by instructing participants to act extraverted or introverted in group interactions) (Fleeson et al., 2002, McNiel and Fleeson, 2006, McNiel et al., 2010, Zelenski et al., 2013). As a consequence of these proposed causal links, the dynamic mediation model predicts that the relationship between trait extraversion and trait PA will be mediated by aggregated extraversion states and aggregated positive affect states. Indeed, Wilt et al. found support for the full mediation model in a series of studies in both naturalistic and experimental settings. The results provided evidence for the role of personality states as a mechanism underlying trait-affect relationships, at least for trait extraversion and PA.

In the present study, we did not assess trait PA so we could not test the final link in the model, but we were able to test the cross-cultural generalizability of the remaining links. Consistent with the dynamic mediation model, we tested the following hypotheses:

Finally, Wilt et al. (2012) focused on extraversion and PA in their test of the dynamic mediation model, but noted that the model might also be applied with alternative traits and affect states. For example, in addition to the well-established relationships between Extraversion and PA and between Emotional Stability and negative affect (NA), other Big Five dimensions—particularly Agreeableness and Conscientiousness—have been shown to correlate with global indicators of subjective well-being (DeNeve and Cooper, 1998, McCrae and Costa, 1991). McCrae and Costa suggested that Agreeableness and Conscientiousness might exert an instrumental influence on subjective well-being because of the daily life circumstances that agreeable and conscientious people create for themselves, such as more positive interpersonal and achievement-related experiences. The relationship between Openness to Experience and subjective well-being is less consistent (DeNeve & Cooper, 1998), but it is possible that enacting more creative or intellectually stimulating behaviors (i.e., Openness to Experience states) could also foster more PA in one’s daily life. Thus, the dynamic mediation model may be applicable to additional traits and affects. Because we assessed all of the Big Five traits and personality states, and both PA and NA, we had the opportunity to extend tests of the dynamic mediation model to additional trait-state relationships, if warranted by the results.

Cultural psychologists emphasize the deeply entwined nature of culture and personality (Heine, 2001, Markus and Kitayama, 1998). Most relevant for the present study is the cultural psychology prediction that behavior and affect will be less determined by traits in non-Western cultures (Markus and Kitayama, 1998, Triandis, 1995). One theoretical perspective links the reduced “traitedness” of behavior and affect to the cultural dimension of individualism–collectivism and associated differences in independent versus interdependent self-construals (Markus & Kitayama, 1991). In individualistic cultures, the self is viewed as independent, unique, and relatively stable, leading to greater motives or tendencies to express one’s internal attributes or traits in behavior. In contrast, people in collectivistic cultures are thought to view the self as relatively interdependent, socially embedded, and malleable (Heine et al., 2001), leading to behavior that is more determined by the expectations and demands of social roles and relationships (Kanagawa et al., 2001, Suh, 2002). From this perspective, we would expect traits to be stronger predictors of corresponding personality states in individualistic cultures than in collectivistic cultures.

Two additional cultural dimensions, dialecticism and tightness, are also potentially relevant. Dialecticism has been defined as a system of thought rooted in Eastern philosophical traditions, and associated with greater acceptance of contradiction, expectations of cognitive and behavioral change, and holistic (vs. analytical) thinking (Peng and Nisbett, 1999, Spencer-Rodgers et al., 2010). Because of their greater dialecticism, people in East Asian cultures are expected to exhibit less traitedness or consistency in their self-concepts, behaviors, and affects across roles and situations (Church et al., 2008, Church et al., 2012, English and Chen, 2007, English and Chen, 2011, Oishi et al., 2004, Suh, 2002). Finally, Gelfand, Nishii, and Raver (2006) defined cultural tightness (vs. looseness) as “the strength of social norms and the degree of sanctioning within societies” (p. 1226). The greater impact of situational constraints in tight cultures, as compared to loose cultures, would be expected to result in weaker prediction of daily behavior and affect from traits (Gelfand et al., 2011).

Only a few cross-cultural studies have related traits to daily behavior and the results have been mixed. In an experience sampling study, Church, Katigbak, et al. (2008) did not find consistently stronger trait-behavior relationships in the United States, an individualistic culture, than in the Philippines, a collectivistic culture. Ching et al. (2013) found that correlations relating Big Five traits to associated personality states were as high, on average, in three Asian samples (China, Malaysia, and Japan) as in American and Mexican samples. Nezlek, Schütz, Schröder-Abé, and Smith (2011) found cultural differences between Americans and Germans in how the Big Five traits related to the quality and quantity of social interactions in a 14-day experience sampling study. The authors attributed some of the differences to the greater formality and structure in German society, a description that recalls the concept of cultural tightness.

Several cross-cultural studies have replicated the Western finding that extraversion and neuroticism are good predictors of positive and negative affect, respectively (e.g., Allik and Realo, 1997, Rodríguez and Church, 2003), although the relationships may be weaker in collectivistic cultures than in individualistic cultures (Lucas, Diener, Grob, Suh, & Shao 2000). However, these findings were based on one-time global assessments rather than experience sampling methods. Several cross-cultural studies that did examine daily affect in experience sampling studies focused on the relationship between positive and negative affect (e.g., Perunovic et al., 2007, Scollon et al., 2005), within-individual variability in affect (Church et al., 2013, Oishi et al., 2004), or cultural differences in levels of well-being (Oishi, 2002), rather than the prediction of daily affect from personality traits or states. Potentially relevant, however, are studies that have reported cultural differences in the impact of daily events on well-being or affect (e.g., Nezlek et al., 2011, Oishi et al., 2007). Because daily events presumably elicit personality states, these results suggest that cultures may differ in the nature and strength of the relationships between personality states and affects. In summary, although existing cross-cultural evidence is limited and mixed, we drew on cultural psychology theory to formulate the following hypothesis:

Drawing on previous research (Church et al., 2012, Díaz-Loving and Draguns, 1999, Gelfand et al., 2011, Hofstede, 2001), we sampled five cultures (the US, Venezuela, Philippines, China, and Japan) that were expected to vary along the cultural dimensions of individualism–collectivism, dialecticism, and tightness. Hofstede ranked 53 countries and regions on individualism based on a cross-national study of values. The US ranked 1st, Japan 22nd, the Philippines 31st, and Venezuela 50th. China was not included but Taiwan (43rd) and Singapore (40th) were ranked as collectivistic. Dialecticism has been linked to Asian countries, particularly those in East Asia such as China and Japan (Peng and Nisbett, 1999, Spencer-Rodgers et al., 2004). Finally, in Gelfand et al.’s (2011) study of 33-countries, China and Japan were high in tightness and the US and especially Venezuela were relatively loose (the Philippines was not included in the study). Church, Alvarez et al. found that Chinese and Japanese averaged higher in cultural tightness than Filipinos and Americans, who, in turn, averaged higher than Venezuelans. It would be difficult, if not impossible, to differentiate various combinations of these cultural dimensions in a small set of cultures. However, we hoped to use measures of the cultural dimensions to confirm our classification of the cultures and to determine whether individualism–collectivism, dialecticism, or tightness would best account for any cultural differences in the ability of the Big Five traits to predict personality states and affect.

Participants in each culture first rated their traits in general and completed measures of individualism–collectivism, dialecticism, and tightness. Participants then rated their personality states and their PA and NA three times a day for 20 days. General traits and personality states were both assessed using the Big Five dimensions, enabling a more straightforward evaluation of the degree of convergence between traits and personality states (Fleeson and Wilt, 2010, Fleeson et al., 2002). Based on trait theory, we expected the density distribution and dynamic mediation models to generalize well across cultures, as stated in Hypotheses 1–4. At the same time, if cultural psychology perspectives are valid, we should also find some cultural differences in the strength of trait-state relationships, as stated in Hypothesis 5.