CRASH PREDICTION MODELS FOR OLDER DRIVERS: A PANEL DATA ANALYSIS APPROACH

The graying of America is resulting in a larger proportion of older individuals in the population. Recent transportation surveys show that an increasing number of older individuals are licensed to drive and that they drive more than their same age cohort a decade ago. These trends necessitate increased study of their potential highway safety problems. Considerable progress has been made on understanding older drivers' safety issues. Nonetheless, research has been rather limited and the findings inconclusive. One of the methodological limitations is the lack of considering temporal order between events (i.e., the time between onset of medical condition, symptom, and crash). Without time-series data, researchers have often linked a "snap-shot" of medical conditions and driving patterns to more than one year of crash data, hoping to accumulate enough data on crashes. The interpretation of the results from these studies is difficult in that one cannot explicitly attribute the increase in highway crash rates to medical conditions and/or physical limitations. This paper uses a panel data analysis approach to identify factors that place older drivers at greater crash risk. Our results show that factors that place female drivers at greater crash risk are different from those influencing male drivers. More risk factors were found to be significant in affecting older men's involvement in crashes than older women. When the analysis controlled for the amount of driving, women who live alone or who experience back pain were found to have a higher crash risk. Similarly, men who are employed, score low on word-recall tests, have a history of glaucoma, or use antidepressant drugs were found to have a higher crash risk. The most influential risk factors in men were the amount of miles driven, and use of antidepressants.