Evidence that changes in social cognitions predict changes in self-reported driver behavior: Causal analyses of two-wave panel data.

Previous research on the theory of planned behavior (TPB) is characterized by cross-sectional tests of the model's proposed causal relationships. In the absence of effective experimental techniques for changing the TPB's cognitive antecedents, the present research aimed to provide a stronger non-experimental test of the model, using causal analyses of two-wave panel data. Two studies of driver behavior were conducted in which naturally occurring within-participant changes in TPB constructs were measured over time, and used to predict corresponding within-participant changes in both intentions and behavior. A two-wave panel design was used in both studies. Study 1 had a one-month gap between baseline and follow-up. At both waves, a convenience sample comprising predominantly university students (N=135) completed questionnaire measures of all TPB cognitions and behavior (compliance with speed limits in urban areas). Cross-lagged multiple regressions and bootstrapping procedures for testing multiple mediators supported all of the relationships proposed by the TPB. These findings were extended in study 2 using a large, non-student sample of speed limit offenders (N=1149), a six-month gap between baseline and follow-up, and a larger number of cognitive antecedents. Participants completed postal questionnaires at both waves to measure all cognitions proposed by the two-component TPB, along with moral norm, anticipated regret, self-identity and speeding on urban roads, country roads, and fast dual carriageways or motorways. Changes in instrumental and affective attitude, descriptive norm, self-efficacy, moral norm, anticipated regret and self-identity predicted changes in intention to speed. Changes in intention and self-efficacy predicted behavior-change. Injunctive norm and perceived controllability did not predict intention or behavior-change. Additionally, direct (unhypothesized) relationships with behavior were found for affective attitude, descriptive norm and anticipated regret. The implications of the findings for theory and the development of effective behavior-change interventions are discussed.