The role of general and specific attitudes in predicting travel behavior - A fatal dilemma?

Attitudes are increasingly used in travel behavior research to help explain and predict travel behavior. In such studies, empirical correlations between attitudes and behavior are routinely interpreted as causal effects, which paves the way for policy interventions aimed at changing attitudes and thereby, ultimately, behavior. This paper contributes to a recent and growing body of work which points at the shaky foundations underlying this attitude-behavior conceptualization. In contrast to previous work in this direction, we distinguish between general attitudes and specific attitudes and we study their potential and limitations in explaining and predicting travel behavior. We build and empirically confirm a set of hypotheses which argues that neither of these two types of attitudes is capable of providing empirical evidence for a causal effect of attitudes on behavior. General attitudes, which have the advantage of being relatively exogenous to the behavior being studied, only have a weak empirical association with specific travel behaviors. Specific attitudes towards these travel behaviors overcome this problem as they are much more strongly correlated with behavior, but this comes at the cost of a severe loss in exogeneity; in other words, the causal relation from specific behavior to specific attitudes is considerably stronger than the opposite effect. In combination, our findings suggest that it is very difficult, if not impossible, to identify and measure attitudinal variables that satisfy two necessary criteria for causal inference: empirical association and exogeneity. Implications for travel behavior researchers and transport policy makers are likely to be far-reaching.