This paper discusses how to determine the prevalence of drinking drivers' under- and over-estimates of their blood alcohol content (BAC). Using data collected as part of the 1990 Minnesota, USA, roadside survey of night drivers, it examines how far drivers above the legal limit are more likely than drivers below the limit to underestimate their BAC. For two weeks in September 1990, voluntary interviews were conducted with 2857 drivers between 10pm and 3am in 16 Minnesota communities. They were asked if they had consumed any alcohol in the last 12h; those who had were asked to estimate their BAC at the time. All drivers were tested on a passive BAC sensor, and, if that revealed BAC of at least 0.02%, also give a breath test. 50% of the drinking drivers were classified as underestimators of BAC, 32% as overestimators, and 19% as estimators of BAC within 0.01 of its actual value. Overestimators of BAC usually had BAC below 0.05%, and underestimators usually had BAC over 0.05%. 94% of drivers, with BAC over the statutory 0.10% limit, and all drivers with BAC over 0.15%, underestimated their BAC. The results suggest that drink-driving might be reduced by informing drivers of the dangers of relying on their own estimates of BAC.
Abstract