Individual cigarette usage: self-reported data as a function of respondent-reported data.

The interview method was used to collect data from living respondent-subject pairs regarding cigarette usage of the subject; the nonsmoker subject was explicitly not studied. Several different measures of usage were tested. The mean correlation coefficient between subject-reported data and the corresponding respondent-reported data was 0.86; that is, at least 70% of the variability in subject-reported data can be accounted for from information obtained from an appropriate living respondent closely associated with the subject. Selected regression equations are presented for estimating cigarette usage of a subject from data provided by a respondent.