Aspects of the interrelationships of attitudes and behaviour as illustrated by a longitudinal study of British adults: 3. Variation in individuals' attitudes over time and a cross-temporal ecological fallacy

In most models of the links between attitudes and behaviour it is assumed (implicitly if not explicitly) that people have stable predispositions to act in particular ways. This assumption has rarely been tested in studies of British voting behaviour which show, as in the first two papers of this series, strong links between measured attitudes and party choice when a longitudinal data set is used. Investigations of the respondents' attitudes over time show substantial inconsistency, however, which suggests a cross-temporal ecological fallacy and raises serious questions regarding the traditionally employed models of voting behaviour.