Hostile attribution bias and angry rumination: A longitudinal study of undergraduate students

Abstract

Angry rumination and hostile attribution bias are important cognitive factors of aggression. Although prior theoretical models of aggression suggest that aggressive cognitive factors may influence each other, there are no studies examining the longitudinal relationship between angry rumination and hostile attribution bias. The present study used cross-lagged structural equation modeling to explore the longitudinal mutual relationship between hostile attribution bias and angry rumination; 941 undergraduate students (38.5% male) completed questionnaires assessing the variables at two time points. The results indicate that hostile attribution bias showed a small but statistically significant effect on angry rumination 6 months later, and angry rumination showed a quite small but marginally significant effect on hostile attribution bias across time. The present study supports the idea that hostile attribution bias influences angry rumination, and argue that the relationship between angry rumination and hostile attribution bias may be mutual. Additionally, the results suggest that there may be a causal relation of different aggression-related cognitive factors.

Fig 1. Standardized structural model of longitudinal cross-lagged panel analysis on hostile attribution bias and angry rumination across two waves.

Note, ABQ = hostile attribution; AR = angry rumination; a1 = angry rumination parcel 1 of time 1; a2 = angry rumination parcel 2 of time 1; a3 = angry rumination parcel 3 of time 1; a4 = angry rumination parcel 4 of time 1; b1 = angry rumination parcel 1 of time 2; b2 = angry rumination parcel 2 of time 2; b3 = angry rumination parcel 3 of time 2; b4 = angry rumination parcel 4 of time 2; c1 = hostile attribution bias parcel 1 of time1; c2 = hostile attribution bias parcel 2 of time1; c3 = hostile attribution bias parcel 3 of time1; c4 = hostile attribution bias parcel 4 of time1; d1 = hostile attribution bias parcel 1 of time2; d2 = hostile attribution bias parcel 2 of time2; d3 = hostile attribution bias parcel 3 of time2; d4 = hostile attribution bias parcel 4 of time2. All the reported parameters are standardized. ***p < 0.001.