Reliability of retrospective self-reports of alcohol consumption among women: data from a U.S. national sample.

Drinking histories (retrospective self-reports) can be a valuable resource for time-ordered analyses of causes and consequences of drinking. However, there is a scarcity of data on the reliability of drinking histories from general population samples. We report here on the reliability and consistency of reported ages of onset and typical drinking frequencies, quantities and volume, from drinking histories provided in 1981 and 1986 by national samples of women drinkers with and without drinking problems. Statistical reliability was generally modest, yet large percentages of women gave exactly the same reports 5 years apart. Reliability was apparently reduced by limited response options, and was lower among younger drinkers, whose drinking was more changeable between 1981 and 1986. We discuss ways to improve reliability and to make best use of drinking histories.