The effect of Massachusetts' reduced drinking age on auto accidents is examined by employing an interrupted time series analysis of monthly accident data covering the period January, 1969, through September 1973. These raw data were adjusted using monthly mileage and seasonal indices or where possible, a control group not affected by the drinking law. Correlograms of the adjusted series were computed to check for remaining systematic bias. Finally, the average accident rates for the adjusted, well-behaved series before and after the March 1973 change were compared using standard t-tests. Accident rates among 18–20 year olds did increase significantly—about 40% for involvement in fatalities. Nevertheless, the results are consistent with the hypothesis that, as a result of the reduced drinking age, 18–20 year old driving-after-drinking behavior has become comparable to that of older drivers.
Abstract