Do alcoholics give valid self-reports?

Self-reports on drinking among alcoholics (100 men inpatients) were compared with descriptions of their consumption given by collaterals (one friend or relative each) at 10 points during an 18-month follow-up study. The correlations between the two were only moderate; barely one-half of the variance in the alcoholics' self-reports corresponded to the collaterals' assessments. Patients underestimated collaterals' descriptions about three times as often as they overestimated them, but their over- and underestimations appeared to be of roughly equal size. The relationships between alcoholics' and collaterals' reports tended to be curvilinear. Among subjects whom the collaterals had described as abstinent or controlled drinkers, patients' and collaterals' assessments were similar but patients' descriptions grossly underestimated collaterals' reports when uncontrolled consumption was reported by the latter. The results support a moratorium on the use of patients' self-reports in follow-up studies on alcohol consumption.