CARE: An Automobile Crash Data Analysis Tool

T he University of Alabama, with National Highway Transportation Safety Administration funding for traffic accident problem identification and evaluation, has developed the Critical Analysis Reporting Environment software system to analyze automobile crash data. CARE’s developers designed it for use by transportation safety engineers and policymakers, rather than trained statisticians. To date, eight different states have applied CARE to crash data. The tool provides decision makers with the software environment they need to make effective decisions involving the development and enforcement of strategies to reduce crashes, without having to rely on statisticians and analysts to interpret that information. CARE provides an intuitive interface that transportation safety engineers can use to obtain information from the responses to a fixed set of multiplechoice questions on automobile crash forms that law enforcement officials fill out. As such, CARE’s analysis domain is restricted to categorical data, represented by nominal or ordinal values such as type of crash, driver gender, and the county where a given crash occurred. Even for noncategorical, quantitative information such as driver age or number of vehicles involved, CARE groups the values into ranges labeled with nominal values such as ages 25-34, 35-45, and so on. Periodic batch updates at regular, prespecified intervals insert crash data into a data set suitable for CARE processing. Thus, no transactional updates are applied to the data stored in CARE. The restriction of the data to categorical values has implications for both the types of analysis conducted and the storage implementation of CARE-compatible data sets. Analysis features a heavy emphasis on counting. Our user community frequently poses questions like the following: