Attitudes "cause" behaviors: A structural equation analysis
Abstract
Confirmatory maximum likelihood estimation of linear structural equation models with latent variables was employed to evaluate the causal predominance of attitudes over behaviors. Two wave, 2 variable (2W2V) crosslagged structural models were developed using attitude and behavior panel data from 158 college students on studying, exercise, and dating. Additional causal factors that have been shown to have predictive utility in the context of attitude and behavior models were added to the 2W2V models to determine the impact of the specification of other relevant factors on the cross-lag parameter estimates. Although attitudes consistently had a significant direct effort on subsequent behavior in the 2W2V models, this pattern did not hold in the expanded models. Results suggest that the specification of added factors clarifies the direct as well as indirect impact of attitudes on behavior for varying content domains and enables a more complete assessment of the generality of the nature of attitude–behavior relations. (32 ref) (APA PsycInfo Database Record (c) 2016 APA, all rights reserved)