Cross-cultural evidence for the fundamental features of extraversion

Summary

This research investigated the fundamental nature of extraversion through four studies testing whether extraversion facets are linked by reward sensitivity or sociability. The authors developed a new Social Interaction Scale to measure preference for social activity independent of reward-seeking behavior.

Using structural equation modeling across diverse samples including participants from 39 nations (total N = 6,469), results consistently showed that only facets reflecting reward sensitivity (affiliation, ascendance, venturesome) loaded on a higher order extraversion factor, which correlated strongly with pleasant affect (r = .62 to .82).

The Social Interaction Scale did not load on the extraversion factor and showed no correlation with pleasant affect, suggesting that extraverts’ sociability may be a consequence of reward sensitivity rather than the core defining feature of the trait.

Key Takeaways

• Reward sensitivity, not sociability, forms the core of extraversion: Only facets that reflect sensitivity to rewarding situations (affiliation, ascendance, venturesome) loaded on the higher order extraversion factor, while a pure measure of social interaction preference did not.

• Extraversion correlates strongly with pleasant affect across cultures: The relationship between the higher order extraversion factor and pleasant affect was robust (r = .59 to .82), providing support for reward sensitivity models that posit pleasant affect as a direct outcome of the behavioral activation system.

• The extraversion structure generalizes across cultures with some moderation: While the basic structure replicated across 39 nations and between individualistic and collectivistic cultures, the extraversion-pleasant affect correlation was significantly weaker in collectivistic cultures (r = .59) compared to individualistic cultures (r = .77), suggesting cultural factors may moderate how reward sensitivity manifests in social behavior.

Research Context and Background

Extraversion has been central to personality psychology since the work of Jung (1921/1971) and James (1907), appearing in most modern taxonomic approaches and major personality inventories. However, fundamental questions about its defining characteristics remain unresolved. The lay conception equates extraversion with sociability, supported by McCrae and Costa’s (1987) finding that adjectives like “sociable,” “fun-loving,” “affectionate,” “friendly,” and “talkative” loaded most strongly on extraversion factors.

Watson and Clark (1997) identified six facets consistently included in extraversion models: venturesome (excitement seeking), affiliation (warmth and gregariousness), positive affectivity (joy and enthusiasm), energy (feeling lively and active), ascendance (dominance or exhibitionism), and ambition (achievement orientation). Depue and Collins (1999) proposed a more succinct three-component model: affiliation (enjoying close interpersonal bonds), agency (social dominance, assertiveness, exhibitionism), and impulsivity (which they argued should not be included).

A major puzzle in the extraversion literature is its strong, consistent correlation with pleasant affect. Early research assumed extraversion caused pleasant affect through increased social activity, as social situations tend to increase positive emotions. However, Pavot, Diener, and Fujita (1990) found that extraverts reported more pleasant affect even when alone and did not spend more time in social situations than introverts, challenging simple sociability-based explanations.

Gray’s (1970) Behavioral Activation System (BAS) theory offered an alternative framework. According to this model, individual differences in reward sensitivity create differences in approach behavior and pleasant affect. Depue and Collins (1999) elaborated this into a comprehensive theory of extraversion as reflecting a strong positive incentive motivational system. The challenge for reward sensitivity models is explaining why extraverts appear more sociable if the underlying mechanism is general reward sensitivity rather than specific preference for social interaction.

Research Question/Hypothesis

Primary Research Question: What is the fundamental feature that links the various facets of extraversion—sociability or reward sensitivity?

Competing Hypotheses:

1. Sociability-as-Core Model: The variance common to all extraversion facets reflects a tendency to enjoy and prefer social interactions. All facets (including a pure social interaction measure) should load on a single higher order extraversion factor, which should correlate moderately with pleasant affect (as an indirect outcome mediated by social activity).

2. Reward-Sensitivity-as-Core Model (authors’ prediction): Only facets reflecting sensitivity to rewards (affiliation, ascendance, venturesome) should load on the higher order extraversion factor. A pure measure of social interaction preference (independent of reward-seeking) should not load on this factor. The extraversion factor should correlate strongly with pleasant affect (as a direct outcome of the reward sensitivity system). Social interaction should not correlate with pleasant affect.

Secondary Aims:

• Test the generalizability of the extraversion structure across 39 nations
• Examine whether individualism-collectivism moderates the extraversion-pleasant affect relationship
• Validate findings using established extraversion scales (NEO-PI-R, International Personality Item Pool)

Methodology

Study Design

The research employed a cross-sectional, correlational design using structural equation modeling (specifically, confirmatory factor analysis with latent variables) across four studies. Studies 1 and 2 tested the basic model structure, Study 2 examined cross-cultural generalizability using means and covariance structure analysis (MACS), Study 3 replicated findings with established scales (NEO-PI-R), and Study 4 addressed potential methodological concerns about the Social Interaction Scale using revised items.

Participants/Subjects/Materials

Validation Sample (for scale development): 121 U.S. undergraduate psychology students

Study 1: 443 U.S. college students from two large Midwestern universities (52% male, 48% female; 94% ages 18-25); 404 with complete data used in analyses

Study 2: 6,469 college students from 39 countries across all inhabited continents (2,468 men, 3,923 women, 78 not reporting gender; 83% ages 18-25); 5,842 with complete data used in analyses. Countries included represented diverse cultural contexts from Argentina to Zimbabwe, with individualism-collectivism ratings obtained.

Study 3: 158 students in a semester-long subjective well-being course; 134 with complete data

Study 4: Validation sample of 68 undergraduates; structural modeling sample of 142 upper-level students in subjective well-being course; 131 with complete data

Procedures and Data Collection

Scale Development: The authors created a new 31-item Extraversion Scale with four facets:

1. Affiliation (11 items): Warmth and enjoying close relationships (e.g., “You enjoy talking with your friends,” “You are a very friendly person”)

2. Ascendance (6 items): Dominance and assertiveness (e.g., “You are a leader of others,” “You are self-confident”)

3. Venturesome (6 items): Excitement seeking (e.g., “You prefer to be with people who are exciting rather than quiet”)

4. Social Interaction (8 items, expanded to 17 in Study 4): Novel scale designed to assess preference for social activity independent of reward-seeking, using three approaches: (a) reverse-scored enjoyment of nonsocial situations, (b) preference for social over equally enjoyable nonsocial situations, (c) participation in social activities without describing specific activities. Items emphasized choosing social situations even when attractive nonsocial alternatives existed (e.g., “You always prefer being with others to spending time alone,” “You enjoy being alone” [reverse-scored]).

The scale was derived from an initial 42-item version through exploratory and confirmatory factor analyses, retaining items that loaded consistently across samples and methods.

Pleasant Affect Measurement:

• PANAS Positive Affect Scale (Watson, Clark, & Tellegen, 1988)
• Frequency ratings for four positive emotions (affection, joy, contentment, pride)
• Intensity ratings for the same four emotions
• Study 3 used three administrations of the PANAS across a semester

Validation Measures:

• NEO-PI-R Extraversion scale and facets (Costa & McCrae, 1992)
• Multidimensional Personality Questionnaire Positive Emotionality scales (Tellegen & Waller, 1994)
• Eysenck Personality Questionnaire Extraversion scale (Eysenck & Eysenck, 1975)
• International Personality Item Pool Extraversion facets (Goldberg, 1999)
• Self-reported social activity (number of activities, number of friends seen, percentage of time in social activities)

Translation: For the international sample, back-translation was conducted for Chinese, Japanese, Korean, and Spanish versions. These translations were used across multiple countries speaking these languages.

Data Collection: Participants completed questionnaires in university settings, typically receiving course credit. Study 3 participants completed measures multiple times across a semester.

Statistical Analysis

Primary Analytic Method: Structural equation modeling using AMOS software (Arbuckle, 1999). To reduce model complexity given limited indicators per construct, items within each facet were grouped into three parcels based on item-total correlations, creating three indicators per latent facet.

Model Evaluation Criteria (following Little, 1997):

• Chi-square statistic (though acknowledged as overly sensitive with large samples)
• Chi-square to degrees of freedom ratio (target: close to 2)
• Nonnormed Fit Index (NNFI; target: > .90)
• Incremental Fit Index (IFI; target: > .90)
• Root-Mean-Square Error of Approximation (RMSEA; target: ≤ .05)

Baseline Model: For relative fit indexes (NNFI, IFI), an independence model with indicator means equal to the grand mean was used rather than the default (means = 0), providing more conservative estimates. All scores were centered around the grand mean across combined U.S. and international samples.

Cross-Cultural Analysis (Study 2): Means and Covariance Structure (MACS) analysis tested sequential constraints:

1. Identical structure with freely estimated parameters across groups
2. Factor loadings constrained to equality
3. Indicator means constrained to equality
4. Structural model parameters (variances, factor loadings) constrained to equality
5. Specific correlations of theoretical interest constrained to equality

Model Comparison: For nested models, chi-square difference tests assessed whether constraints significantly worsened fit. Following the modeling approach, changes in NNFI and IFI < .05 indicated trivial information loss from constraints, supporting parsimony.

Cultural Categorization: Nations were rated on individualism-collectivism by Hofstede (1980) and Triandis (Suh et al., 1998) using 1-10 scales (interrater correlation = .78 for 26 overlapping nations). Nations rated ≤ 3 were classified as collectivistic; those rated ≥ 6 as individualistic.

Reliability: Cronbach’s alpha coefficients were calculated for all scales across studies.

Software: AMOS 4 for structural equation modeling; standard statistical software for correlations, descriptive statistics, and reliability analyses.

Results

Primary Findings

Scale Validation: The new Extraversion Scale demonstrated good convergent validity, correlating .55-.58 with established extraversion measures (NEO-PI-R, MPQ, EPQ) in the validation sample. Facet subscales showed appropriate discriminant validity, correlating most strongly with conceptually similar NEO-PI-R facets. Critically, the Social Interaction Scale correlated .45 (p < .001) with self-reported social activity but only -.05 to .06 (ns) with pleasant affect across studies, successfully separating sociability from positive emotionality.

Study 1 - U.S. Sample (N = 404):

Sociability-as-Core Model (all four facets loading on extraversion):

• Acceptable fit: χ²(85) = 228.74, p < .001, χ²/df = 2.69, NNFI = .94, IFI = .95, RMSEA = .07
• Critical finding: Social Interaction loading on extraversion was only .03 (ns, critical ratio = .52), failing to share variance with other facets despite clear social content

Reward-Sensitivity-as-Core Model (only affiliation, ascendance, venturesome loading on extraversion):

• Good fit: χ²(85) = 190.18, p < .001, χ²/df = 2.24, NNFI = .95, IFI = .96, RMSEA = .06
• Standardized factor loadings: Affiliation = .72, Ascendance = .71, Venturesome = .22 (all significant)
• Extraversion-pleasant affect correlation: r = .82 (very strong)
• Social Interaction-pleasant affect correlation: constrained to 0 (estimated at -.05, ns when freed)
• Social Interaction-venturesome correlation: r = .37 (accounting for shared social variance)

Study 2a - International Generalizability (N = 5,842 from 39 nations, plus 404 U.S.):

Test of Uniqueness: Sequential constraint testing showed:

• Identical structure fit well across U.S. and international samples
• Factor loadings could be constrained to equality with minimal fit decrease
• Indicator means could be constrained with acceptable fit
• Structural parameters (variances, factor loadings) could be constrained: χ²(204) = 1,929.46, NNFI = .93, IFI = .94, RMSEA = .04
• Extraversion-pleasant affect correlation differed significantly between samples: Δχ²(1) = 4.66, p < .05
  • U.S. sample: r = .80
  • International sample: r = .71
• Pattern of factor loadings similar to Study 1: Affiliation = .73, Ascendance = .60, Venturesome = .33
• Social Interaction again did not load on extraversion factor

Study 2b - Individualism-Collectivism Comparison (N = 404 per group, randomly sampled):

Three-Group Comparison (U.S., Individualist nations, Collectivist nations):

• Identical structure fit well with free parameters: χ²(255) = 556.56, NNFI = .94, IFI = .95, RMSEA = .03
• Factor loadings constrained successfully across groups
• Means could NOT be constrained across all three groups (substantial fit decrease)
• Means could be constrained only between U.S. and individualist samples, indicating significant mean differences
• Structural parameters constrained successfully: χ²(308) = 845.26, NNFI = .91, IFI = .92, RMSEA = .04
• Extraversion-pleasant affect correlation varied significantly across all three groups: Δχ²(2) = 8.43, p < .05
  • U.S. sample: r = .77
  • Individualist sample: r = .77 (not significantly different from U.S.)
  • Collectivist sample: r = .59 (significantly lower)
• Factor loadings: Affiliation = .72, Ascendance = .72, Venturesome = .30

Secondary Findings

Study 3 - Replication with Established Scales (N = 134):

Using new facet scales:

• Excellent fit: χ²(13) = 16.35, ns, χ²/df = 1.26, NNFI = .98, IFI = .99, RMSEA = .04
• Factor loadings: Affiliation = .74, Ascendance = .59, Venturesome = .23
• Extraversion-pleasant affect: r = .62
• Social Interaction-extraversion: constrained to 0
• Social Interaction-venturesome: r = .55

Using NEO-PI-R facet scales (Warmth, Assertiveness, Excitement Seeking):

• Excellent fit: χ²(13) = 15.54, ns, χ²/df = 1.20, NNFI = .99, IFI = .99, RMSEA = .04
• Factor loadings: Warmth = .62, Assertiveness = .52, Excitement Seeking = .44
• Extraversion-pleasant affect: r = .66 (very similar to new scales)
• Results demonstrate findings are not artifacts of the new Extraversion Scale

Study 4 - Revised Social Interaction Scale (N = 131):

To address concerns about reverse-keyed items and implicit contrasts in original Social Interaction Scale, 9 new positively-keyed items were developed focusing on time spent with others vs. alone. The 17-item composite scale (original 8 + new 9 items):

• Correlated .62 with original scale (uncorrected), .94 when corrected for unreliability
• Showed similar pattern: correlated .46 with gregariousness but .05 (ns) with pleasant affect
• Correlated .45 with actual social activity index

Using new facet scales with 17-item Social Interaction Scale:

• Excellent fit: χ²(5) = 4.99, ns, χ²/df = 1.00, NNFI = 1.00, IFI = 1.00, RMSEA = .00
• Factor loadings: Affiliation = .58, Ascendance = .58, Venturesome = .31
• Extraversion-pleasant affect: r = .67

Using International Personality Item Pool facet scales:

• Good fit: χ²(5) = 7.94, ns, χ²/df = 1.59, NNFI = .95, IFI = .98, RMSEA = .07
• Factor loadings: Friendliness = .74, Assertiveness = .74, Excitement Seeking = .40
• Extraversion-pleasant affect: r = .62

Results confirmed findings were not due to item keying or measurement artifacts.

Reliability: All facet scales demonstrated acceptable to good internal consistency across studies:

• Affiliation: α = .75 to .90
• Ascendance: α = .76 to .83
• Venturesome: α = .58 to .80
• Social Interaction: α = .75 to .87

Cross-Study Consistency: The pattern of a strong extraversion-pleasant affect correlation (.59 to .82 across all studies and samples) and non-significant Social Interaction-pleasant affect correlation (-.06 to .06) was remarkably consistent across all four studies, multiple samples, different measures, and diverse cultural contexts.

Discussion and Interpretation

Main Conclusions

The authors conclude that reward sensitivity, not sociability, forms the fundamental core feature of extraversion. This conclusion rests on several converging lines of evidence:

1. Structural Evidence: Only facets reflecting reward sensitivity (affiliation, ascendance, venturesome) consistently loaded on the higher order extraversion factor across all studies, samples, and measurement instruments. The Social Interaction Scale, despite clear social content and correlations with other sociability measures, shared no variance with the common extraversion factor.

2. Correlational Pattern: The higher order extraversion factor showed very strong correlations with pleasant affect (.59 to .82), consistent with Depue and Collins’ (1999) theory that pleasant affect is a direct outcome of the behavioral activation system. In contrast, Social Interaction consistently failed to correlate with pleasant affect, suggesting that preference for social interaction per se is independent of the positive emotionality that characterizes extraverts.

3. Theoretical Integration: The findings support Watson and Clark’s (1997) argument that positive emotionality forms the “glue” holding extraversion facets together, rather than sociability being the core feature. The authors argue that extraverts’ apparent sociability is better understood as a manifestation of their reward sensitivity, given that social situations tend to be especially rewarding for most people.

Comparison with Previous Work

The findings challenge the folk conception and some empirical interpretations (e.g., McCrae & Costa, 1987) that equate extraversion with sociability. While McCrae and Costa found that sociable, fun-loving, affectionate, friendly, and talkative were the highest-loading adjectives on extraversion, the current research suggests this may reflect the confounding of sociability with reward-seeking in most existing measures.

The results align with and extend several lines of previous research:

• Watson and Clark (1997): Confirmed their finding that extraversion facets correlate more strongly with positive affect than with each other, and that positive emotionality may be the core feature.

• Pavot et al. (1990): The strong extraversion-pleasant affect correlation even when controlling for social activity supports their finding that extraverts experience more pleasant affect even when alone, contradicting simple sociability-based explanations.

• Cunningham (1988a, 1988b): The pattern whereby reward-seeking facets (but not pure sociability) load on extraversion parallels Cunningham’s experimental findings that pleasant moods increase preference for rewarding social situations but not non-rewarding social situations.

• Depue and Collins (1999): Provides strong empirical support for their theoretical model of extraversion as reflecting individual differences in the behavioral activation system, with pleasant affect as a direct manifestation of this system’s activity.

The cross-cultural findings extend beyond previous structural studies (e.g., McCrae & Costa, 1997) by examining not just whether the structure replicates but also whether the nomological network (specifically the extraversion-pleasant affect relationship) generalizes and how it may be moderated by cultural context.

Mechanisms and Explanations

The authors propose a two-factor model to explain the relationship between extraversion and sociability:

Factor 1: Internal Reward Sensitivity - Extraverts possess a more sensitive behavioral activation system (BAS), which:

• Increases motivation to approach potentially rewarding stimuli
• Produces stronger pleasant affect when rewards are encountered
• Enhances energy, desire, and self-efficacy when engaging goal-directed behavior
• Operates through dopaminergic pathways as described by Depue and Collins (1999)

Factor 2: External Rewardingness of Social Situations - Social situations, particularly those involving warmth and close interpersonal bonds:

• Are inherently rewarding for most humans due to fundamental needs to belong (Baumeister & Leary, 1995)
• Provide more opportunities for rewards than many non-social situations
• Elevate pleasant affect even for introverts (Pavot et al., 1990)

Resulting Pattern: When reward-sensitive individuals (extraverts) encounter rewarding situations (social contexts), they experience stronger pleasant affect and greater motivation to approach these situations, manifesting as apparent sociability. However, this sociability is a by-product rather than a primary feature—extraverts should be equally drawn to any rewarding stimulus, social or otherwise.

Cultural Moderation: The weaker extraversion-pleasant affect correlation in collectivistic cultures (.59 vs. .77) suggests that cultural factors may influence Factor 2—the rewardingness of social situations. In collectivistic cultures:

• Social behaviors may be more strongly motivated by norms, duties, and obligations rather than pleasure
• Pleasant emotions may play a smaller role in guiding behavior (Suh et al., 1998)
• The link between reward sensitivity and social approach behavior may be attenuated

This cultural moderation supports the model by demonstrating that when the external rewarding properties of social situations change (or their influence on behavior changes), the strength of the extraversion-pleasant affect relationship changes accordingly, even though the basic structure (which facets load on extraversion) remains constant.

Clinical/Practical Significance

Understanding Individual Differences in Well-being: The strong, consistent extraversion-pleasant affect relationship (.59 to .82 across diverse samples) has important implications for understanding individual differences in subjective well-being. Rather than advising introverts simply to “be more social” to increase happiness, interventions might more effectively focus on helping individuals identify and engage with personally rewarding activities, which may or may not be social in nature.

Personality Assessment: The findings suggest that extraversion scales that heavily emphasize sociability (preference for being with others) may be less valid than those that emphasize enjoyment of rewarding activities and positive emotionality. Assessment developers should ensure their extraversion facets capture reward sensitivity rather than merely social preference.

Cross-Cultural Applications: The generalizability of the basic structure across 39 nations demonstrates that extraversion as reward sensitivity is a cross-culturally valid construct, supporting its use in international research and practice. However, the moderating effect of individualism-collectivism on the extraversion-pleasant affect relationship suggests that the manifestation and consequences of extraversion may vary across cultural contexts. In collectivistic cultures, extraversion may be less strongly associated with subjective well-being.

Person-Situation Fit Research: The findings challenge simplistic person-situation fit hypotheses that assume extraverts are happier in social situations merely because they are social. Instead, researchers should examine whether situations are rewarding rather than simply whether they are social when predicting extravert-introvert differences in affect and behavior.

Therapeutic Implications: For clients seeking to increase positive emotions, therapists might focus on identifying individually rewarding activities rather than assuming social activities will be universally beneficial. Understanding that apparent social withdrawal in some individuals may not reflect social anxiety but rather differences in what activities are rewarding could prevent misdiagnosis.