This article describes the multistage sampling system employed in the 1996 national roadside survey and compares it to the sampling methods employed in the two prior surveys in 1973 and 1986. Also described are the data collection procedures at the selected sites, the breath-test devices used to collect blood alcohol concentration (BAC) data, and the methods used to impute BAC values where breath-test measures were not obtained. Overall, almost twice as many (6,298) in 1996 compared to 3,698 in 1973 and 3,043 in 1986) drivers were interviewed in the most recent national survey as in the previous efforts. The procedures implemented in the three surveys are sufficiently similar to permit comparison of theses surveys conducted at 10-year intervals. (A)
Samenvatting