Validity of synthetic estimates of problem-drinker prevalence.

For the purpose of evaluating their validity, survey-based synthetic estimates of problem-drinker prevalence in California's counties were correlated with social indicators of alcohol abuse. Prevalence estimates were predictive of automobile accidents and drunk-driving arrests but not of other problems. Data on drinking practices from a statewide survey provide more valid estimates than national survey data. Taking into account empirical and other factors impinging upon the validity of this prevalence formula, it is concluded that synthetic estimates provide a defensible but limited basis for estimating needs for alcoholism services in small areas.