Calculating DWI/DWAI recidivism with limited data: using state driver license file for drinking and driving research

One major research issue in drinking-driving is the volume of DWI/DWAI recidivism in a political unit during a given period of time. This article addresses a methodological issue: How can limited data from the official driver license file be used to calculate drinking-driving recidivism rates? New York State maintains one of the most comprehensive driver license files in the nation, but a dynamic process purges records on the file that are more than ten years old. The magnitude of the recidivism rate calculated from this file, thus, is influenced by the number of data points included: the more years of data included, the higher the rate. We used OLS to examine the impact of the dimension of the data on the recidivism rate and mathematically extended the file to the point where the impact of the data dimension is minimum. We, then, calculated the New York State DWI/DWAI recidivism with an "extended dimension."