The structure of positive and negative automatic cognition

Confirmatory factor analysis (CFA) was used to test hypotheses about the structure of positive and negative automatic cognitions, using the responses of 304 college students to separate questionnaires assessing positive (ATQ-P; Ingram & Wisnicki, 1988) and negative (ATQ-N; HolIon & Kendall, 1980) automatic thoughts. For each instrument, an oblique version of the four-factor model originally proposed by the developers (ATQ-N: Personal Maladjustment, Negative Expectations, Low Self-esteem, Giving Up/Hopelessness; ATQ-P: Positive Daily Functioning, Positive Self-evaluation, Other's Evaluations of Self, Positive Future Expectations) provided a marginally acceptable fit to the data (median relative-fit indices = .87 and .89, respectively). Whereas the structure of positive automatic cognition was invariant with respect to gender, women's negative automatic cognitions were more complexly interrelated than men's, and different cognitions were diagnostic of negative automatic thinking for males and females. Within each ATQ model, the four factors were highly intercorrelated (median standardised phi s = .87 and .82, respectively) and largely reflected a single, dominant second-order factor. Supporting this conclusion, a hierarchical model that posits an overarching ''super'' factor underlying the four first-order factors fits the data of each instrument better than did the popular total score model (i.e. one first-order factor). Analyses of convergent and discriminant validity using measures of personality and emotion provided mixed support for the separate four-factor models, with stronger evidence of discriminant validity for the model composed of two, correlated second-order factors (standardised phi = -.53). When considered together, positive and negative automatic thoughts are most accurately conceptualised as separate, negatively correlated, higher-order domains of cognitive experience, each composed of multiple, interrelated subdimensions.