Are Estimates of the Concentration of Alcohol Consumption Affected by Undercoverage? Evidence from Five Pooled U.S. Surveys

Abstract

Conclusions concerning the properties of alcohol consumption distributions are often based on surveys that underestimate the amount consumed by sampled populations. Relationships between underreporting of alcohol consumption and the concentration of heavy drinking could present problems for several important research findings. Using data from five US telephone surveys, we compared state-level survey estimates of alcohol consumption and concentration with published estimates based on sales and taxes. Demographic factors predicting undercoverage were assessed in a multilevel analysis. Concentration was not significantly related to coverage, but it was related to rates of any drinking. State-level undercoverage was predicted by the proportion of males, younger age, political conservatism, Hispanic ethnicity, and the proportion of ethanol sold as spirits. Additional research is needed to determine the consequences ofthe relationships between these factors and undercoverage, but national estimates ofconsumption distribution properties are unlikely to be affected greatly by low coverage rates.