This paper presents results of a meta-analysis of the driver improvement literature. A total of 187 candidate studies were identified and, based on methodological criteria, 35 were considered to be of sufficient quality to warrant inclusion in the meta-analysis. These studies provided information on the crash and traffic violation effects of 106 individual intervention programs. The treatment programs consisted of interventions commonly used by states in attempting to prevent crash and violation repeaters, ranging from warning letters to license suspension or revocation. Specifically excluded from consideration were treatments and sanctions triggered by driving under the influence offenses. Analysis of standardized effect sizes across all types of interventions and research designs indicated that driver improvement programs were associated with small but significant positive effect sizes for both subsequent crashes (dw = 0.03) and subsequent traffic violations (dw = 0.06). Expressed as percentage change in post one-year means and proportions, the standardized effect sizes represent about a 6 percent reduction in crashes and an 8 percent reduction in violations. For license suspension and revocation actions, the most effective approaches studied, the estimated percentage crash and violation reductions were 17 and 21 percent, respectively. On the other hand, the distribution of educational or informational material had no significant effect on subsequent rates for either crashes or violations. Court-initiated traffic violator school programs were not found to reduce subsequent crash risk, though there was a small reduction in traffic violation risk. Distributions of treatment effect sizes for both outcome measures typically were heterogeneous even after controlling (stratifying) for the univariate and two-way effects of numerous potential moderators. Although the heterogeneity complicates interpretation, the majority of the treatment clusters remained associated with statistically significant reductions in postintervention crash and violation rates. Results of a secondary analysis indicated a small but statistically significant correlation between the crash and traffic violation effect sizes of the 106 interventions (r = 0.30). Thus, there is evidence that treatments that reduce violations also tend to reduce crashes, but the relationship is too weak to permit a reliable estimate of the magnitude of crash reduction from knowledge of a treatment’s effect on violations. (Author/publisher)
Abstract