Legal control of drunken driving: A time series study of California data

This study uses a Box-Jenkins type of interrupted time series to evaluate the effectiveness of the new drunken driving laws introduced in California in January 1982 in the light of evidence from relevant data for the first four years following its enactment. Monthly data covering the period January 1978-December 1985 are used to estimate univariate and transfer function models of the time series of numbers of persons killed in auto accidents - alcohol-related as well as all accidents. Apart from a dummy variable to indicate the legislative change, other inputs included in the transfer functions are vehicle miles driven per month and number of Saturdays in each month. The fitted models show a significant decline in the death rates from auto accidents following the enactment of the new laws. Moreover, control effects appear to be persistent rather than transitory.