Techniques for Predicting High-Risk Drivers for Alcohol Countermeasures. Volume 1, Technical Report

Lacey, John H.; Stewart, J. Richard; Council, Forrest M. · 1979 · ROSA P / United States. National Highway Traffic Safety Administration

archive: archived pipeline: cataloged verified

Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)

Summary

This 1979 technical report, conducted by the University of North Carolina Highway Safety Research Center for the National Highway Traffic Safety Administration, addresses the challenge of identifying drivers at high risk of alcohol-related (A/R) crash involvement prior to an incident. The study was motivated by the high cost of broad-based countermeasure programs, such as the Alcohol Safety Action Project, and the need for a more focused approach to target individuals who could benefit from specific interventions. The primary research questions were whether high-risk individuals could be identified before a crash occurred and whether effective countermeasures could be matched to these groups. The researchers identified six high-risk driver groups through a literature review and preliminary analysis of North Carolina accident data: males aged 16–20, males aged 21–24, persons with previous DUI convictions, persons with three or more moving violations, persons recently divorced, and persons recently released from prison. A seventh group, a one-tenth sample of the general driving population, served as a control. Using data available through December 1974 from state motor vehicle, accident, divorce, and prison records, the team developed predictive models for each group using the GENCAT categorical data analysis technique. These models utilized stepwise variable selection to identify subgroups within each category with the highest predicted probability of A/R crash involvement in 1975. The models were validated through concurrent testing on a reserved one-third sample of the 1975 data and prospective testing using 1976 crash data. The results indicated that the models effectively identified subgroups with the highest risk of A/R crash involvement. For instance, the highest-risk subgroup within the DUI category had a predicted A/R crash involvement proportion of 0.07701, representing a risk 21 times greater than the general driving population. The prospective analyses confirmed that the models maintained predictive accuracy for subsequent years. Despite the statistical success of the models, the report concludes that their practical utility is limited by the lack of demonstrably effective countermeasure programs. An addendum notes that only 8% of drivers identified as "most likely" to be involved in an A/R crash actually experience such a crash within twelve months. Consequently, even a fully effective countermeasure would prevent only about eight crashes per 100 high-risk drivers targeted. The authors recommend that these predictive models be used primarily to facilitate well-designed evaluations of A/R crash reduction countermeasures rather than for immediate widespread implementation, due to the modest potential impact on overall crash rates.

Key finding

Predictive models based on driver history and demographic data effectively identified subgroups with significantly elevated risks of alcohol-related crash involvement, with the highest risk subgroup (young males with prior DUI convictions) exhibiting a predicted involvement rate 21 times greater than the general driving population.

Methodology

dataset

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via bulk_ingest_rosap on 2026-05-23 (6 acquisition events logged).

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• induced exposure
• telematics crash prediction
• sex gender
• incidence prevalence
• novice drivers
• demographic disparities

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: crash risk outcomes
• Methodological Resource: dataset resource
• Theoretical Contribution: computational model