Adequacy of survey data collected from substitute respondents

This study examined interview data on personal habits (smoking, drinking and dietary histories) when collected from two sources. A sample of 300 pairs of subjects (mostly husbands and wives) was obtained from an ongoing health survey in Hawaii, and both members of each pair were interviewed separately about the habits of the husband. Care was taken that the members of each pair had no opportunity to communicate with each other between the start and completion of both interviews. The results showed remarkably good agreement in overall mean values between the two groups of respondents. In general, approximately 75% of the pairs agreed within acceptable limits on most items. There was no clear association of the degree of agreement within pairs with a variety of characteristics of the surrogate group, including age, race, level of education, family income, and duration of residence of proportion of meals eaten together with the subject. These findings are encouraging for the use of surrogate respondents in studies of personal habits where direct interviews are not always feasible, or where the use of such substitutes can improve the reppresentativeness and size of the sample.