Prediction versus explanation in the measurement of values

Although economists and sociologists have often concluded that values and other internal states have little causal weight in determining behaviour, there is some evidence that the tide is turning in this respect. This article contributes to the recent revival of interest in subjectivity by comparing two different kinds of survey methods that can be used to measure values in general, and the value of civic mindedness in particular. The explicit approach -widely used in current empirical research - derives value measures on the basis of direct questions asked in nationally representative surveys such as the General Social Survey and the National Election Studies. The factorial approach imputes values indirectly from answers to vignettes. In this study, the explicit approach is revealed to have greater predictive validity but substantially lower construct validity than the factorial approach. This finding highlights the distinction between prediction and explanation in social research, and casts doubt on the adequacy of revealed preference theory.