A comparison of regression and ARIMA models for assessing program effects: An application to the mandated highway speed limit reduction of 1974

Time series analysis is a technique that has been utilized by econometricians & others for examining the relationship between events & time, particularly for forecasting purposes. More recent work has focused on time series analysis as a method of evaluating the effects of an exogenous event on a series. The major disadvantage of the interrupted time series design over a simple pre-post comparison is that the form of the change is taken into account. Two alternative models for analyzing such data are examined here: regression & ARIMA (Auto Regressive Integrated Moving Averages). An example of their application is demonstrated using data on highway deaths in NC occurring before & after the national reduction in speed limits instituted in 1974. Conclusions are drawn about the comparative usefulness of these two techniques for program evaluation. 3 Figures, 12 References. HA.