Abstract
This investigation evaluates the structure and correlates of lower order traits related to approach, specifically, facets of extraversion and behavioral activation system (BAS) sensitivity. A 3-factor structure of approach was derived in community and clinical samples: assertiveness, enthusiasm, and sensation seeking. All factors were positively associated with Openness/Intellect scores. Enthusiasm and assertiveness were both negatively associated with Neuroticism scores, but were distinguished by associations with Agreeableness and Conscientiousness. Sensation seeking was negatively associated with Conscientiousness scores. The 3 factors demonstrated a unique profile of association with components of impulsivity. Enthusiasm and assertiveness were negatively related to psychopathological symptoms, whereas sensation seeking was largely independent of psychopathology. Results suggest that approach is associated with 3 subfactors, which differ in their pattern or magnitude of associations with other variables, thus underscoring the importance of distinguishing among them. Further, results support the construct validity of the Assertiveness and Enthusiasm aspect scales of the Big Five Aspect Scales to assess traits at this level of the personality hierarchy.
Notes
As Reward Responsiveness loaded with Drive on a latent variable labeled agency in research by Wacker, Mueller, Hennig, and Stemmler (Citation2012), cross-loadings were added from Reward Responsiveness to assertiveness in Model 3. In both samples, the change in fit was negligible (Δχ2 ns, ΔCFI <.01) and the factor loadings were not significant (–.04 in Sample 2 and.04 in Sample 3).
The time period between questionnaire administrations might have had numerous effects, including the attenuation of correlations among scales across the instruments relative to correlations among scales within the same instrument. It is important to note that such a time gap could, in fact, influence the factor structure of the scales as a whole. To explore this possibility, we evaluated the configural invariance of Model 3 across Sample 2 and 3 in a “stacked” confirmatory factor analysis, in which these models are estimated simultaneously. The fit of this model was acceptable, χ 2(76) = 418.89, p <.01, CFI =.92, RMSEA =.06, supporting the structural stability of Model 3 across a sample administered all measures concurrently and a sample administered measures across a longer time period. Evidence did not support the metric invariance of Model 3 across Samples 2 and 3, however, χ 2(87) = 628.86, p <.01, CFI =.88, RMSEA =.07, Δχ 2(11) = 209.87, p <.01, ΔCFI >.01, suggesting that factor loadings did in fact differ across samples—which might reflect this alternate method of questionnaire administration or numerous demographic and clinical differences between samples.