Reliability generalization of the alpha coefficient: the driving cognitions questionnaire meta-analysis

Anxiety and fear while driving contribute to risky driving behavior and involvement in traffic accidents. The Driving Cognitions Questionnaire (DCQ) is widely used to measure maladaptive cognitions related to driving anxiety, but there has been no systematic synthesis of its reliability across studies. This study aims to evaluate the internal reliability of the DCQ using the reliability generalization (RG) approach. A meta-analysis was conducted on Cronbach's alpha coefficients for the three DCQ subscales (Panic, Accident, and Social). The literature search followed the PRISMA guidelines, and the analysis was conducted according to the REGEMA guidelines. A total of 16 independent studies were analyzed using a random effects model, with alpha transformation using the Bonett approach. Heterogeneity, publication bias, and meta-regression analyses were performed to test for variations in reliability between studies. The results show that the population reliability estimates are in the good to very good category, with alphas of 0.884 (Panic), 0.870 (Accident), and 0.854 (Social). Very high levels of heterogeneity were found in all subscales (I 2 > 90%). Meta-regression identified age and male gender proportion as significant moderators on the Social subscale, while no significant moderators were found on the Panic and Accident subscales. These findings indicate that the DCQ is a reliable instrument at the population level, but its reliability is contextual. This study emphasizes the importance of reporting reliability based on cumulative evidence in traffic psychology research.