behaviors customers may use to avoid DWI and riding with intoxicated drivers. Tables
1±3 present variables examined for this article. The timeframe for DWI behaviors
examined was the last 12 months.
A generalized form of the Quantity±Frequency±Variability index (QFV; Cahalan et
al., 1969) was used to measure the extent of alcohol use. For this measure, respondents
were asked questions about their usual patterns of drinking, regardless of beverage type.
The QFV measure places respondents into one of the four following drinking habit
categories: infrequent, light, moderate or heavy drinkers. The CAGE questionnaire was
used as a measure of alcohol-use problems (Mayfield et al., 1974). Respondents were
defined as problem drinkers if they endorsed two or more of the four items on the
CAGE measure.
Estimated measures of BAC were computed using a refinement of Widmark's (1932)
formula that incorporated the amount of body water (based on weight) for males and
females and the range of metabolic rates found in the general population (NHTSA,
1994). Respondents reported how many drinks they usually had when drinking outside
the home in the last 12 months and the period of time they usually took to consume the
drinks. Answers to these items, along with information collected about respondents'
gender and body weight, were used to compute their estimated BACs when drinking
outside the home.
IN SITU BREATH TESTING
To provide additional data on profiles of SR users, project staff rode with one of
our participating cab drivers on weekend nights when approximately half of SRs are
provided (50.8%). These assessments were conducted between November of 1997