COGNITIVE BIAS COVARIES WITH ALCOHOL-CONSUMPTION

Cognitive models of behavior led to the prediction that ambiguous stim uli would be interpreted as alcohol-relevant in heavier drinkers. Wome n were asked to define words that could be interpreted as alcohol-rele vant or not (e.g., shot). Two measures served as indicators of a laten t construct of cognitive bias: the number of words construed to be rel ated to alcohol, and the position of the first word in the series that was defined as alcohol-relevant. Average quantity, average frequency, and maximum number of drinks imbibed in a single episode served as in dicators of a latent construct of alcohol consumption. These two const ructs covaried significantly (.44) in a two-factor latent variable ana lysis that fit the data better than alternative models. These findings support an alcohol-related interpretive bias and suggest hypotheses c oncerning the types of interpretive preferences that may contribute to alcohol consumption.