Interviewing older adults: the accuracy of factual information

Using data from a probability sample of adults in a metropolitan area, the accuracy of survey measures was assessed by comparisons to administrative records, census counts, and maps. Some types of information, including birth date and make and year of car, were reported accurately by most respondents, whereas frequent errors were detected for other types of information, including voting behavior, house value, and characteristics of neighbors. Comparisons were made with respect to the relative accuracy of the reports of respondents in three age groups: younger than 60, 60 through 69, and 70 or older. For many measures, no age differences were detected, and for those where age differences were observed, the older respondents were sometimes more accurate than the younger respondents. The potential consequences of measurement error are serious. Evidence to date, however, does not indicate consistently that these problems are any more serious for older respondents than for any other age group.