Toward a Dynamic System of Driver Improvement Program Evaluation

Publisher Website
Google Scholar
Abstract
A number of prior driver improvement research studies are reviewed. The evidence indicates that traffic violations are reduced, at least temporarily, by a variety of driver improvement techniques. The evidence for accident reduction is more equivocal, although a few relatively well-designed studies have reported statistically significant accident reduction. It is argued that the statistical and psychological aspects of accidents prohibit high correlations and large treatment effects. The methodological shortcomings of past driver improvement research are discussed, and the research design characteristics of an optimum evaluation system are outlined. An evaluation system currently under construction in California fulfills most of these optimum characteristics, such as built-in experimental replication, randomized treatment assignment, timely on-line computer-generated effectiveness measures, high statistical power, and cost benefit modeling. One of the major methodological issues in past evaluation efforts is the handling of drivers who do not show up for, or who do not complete, the treatment. The need for considering the entire treatment program (shows and no-shows combined) as the primary unit of evaluation is stressed.

This publication has 8 references indexed in Scilit: