The post-contact driving records of 13,594 male negligent operators (NOs) were predicted using three data sources: prior driving record, NOs' questionnaire responses, and interview information supplied by Driver Improvement Analysts (DIAs). NOs attending group meetings and individual hearings filled out two questionnaires--one which elicited factual data and another which requested NOs to report their feelings. After each hearing, DIA made a prediction about NO's probability of improvement and completed a questionnaire. Equations predicting post- contact accidents and convictions were constructed based on stepwise, multiple regression analyses using half of the sample. Of the accident prediction equations, only the one based solely on prior driver record variables successfully cross-validated. Equations using variables from all three data sources predicted convictions in the cross-validation sample, and two out of the three conviction equations also predicted cross-validation accidents. No significant improvement in accuracy of prediction was made by "tailoring" equations to different contact groups in the construct sample. DIAs, in general, could not predict accidents although there was slight evidence that a few DIAs could. In contrast, most DIAs were able to predict convictions to a limited degree. The results concerning DIA prediction were not cross-validated. Pertinent research from clinical psychology and driver behavior disciplines was reviewed.

  • Corporate Authors:

    California Department of Motor Vehicles

    Research and Statistics Section
    Sacramento, CA  United States 
  • Authors:
    • Marsh, W C
    • Hubert, D M
  • Publication Date: 1974-1

Media Info

  • Features: Appendices; References; Tables;
  • Pagination: 136 p.

Subject/Index Terms

Filing Info

  • Accession Number: 00082739
  • Record Type: Publication
  • Report/Paper Numbers: #50 Final Rpt.
  • Contract Numbers: B0141
  • Created Date: Mar 26 1975 12:00AM