Key points are not available for this paper at this time.
A structural equation model for attitude-behavior relationships is presented which conceptualizes attitude scales as congeneric measurements. The model represents a re-parameterization of an earlier one (see Alwin, 1973). A re-analysis of the Tittle-Hill data on political participation is performed using the model. The results support the view that measurement error in verbal attitudes and true attitude change are important sources of bias which should be taken into account in attitude-behavior studies. This analysis reveals some specification errors in the form of the model estimated previously and suggests some possible errors in the specification of the equations relating attitude constructs over time. Some problems with the present formulation are discussed.
Duane F. Alwin (Wed,) studied this question.