ABSTRACT
According to Ajzen and Fishbein, close relations between attitude and behaviour can be expected only if both measures agree in their degree of specification. Their review of attitude, behaviour studies supported this reasoning: the reported correlations between attitude and behaviour are indeed larger to the extent that the specification of both measures was similar. The correlation between attitudes and behaviours largely varies across content domains. This variation is nicely illustrated in a meta-analytic review by Eckes and Six. In some domains, such as altruistic behaviour or family planning, correlations were low to moderate, whereas in other areas, such as using drugs, the predictive power of attitudes was substantial. Another notable finding of this quantitative review was that the findings varied considerably within domains as well. The attitude-behaviour relation is high if both concepts are measured at the same level of specificity. General attitude measures poorly predict single behaviours, but are good predictors of aggregate measures of behaviour.
