The Effects of Nonresponse and Late Response on a Survey of Physician Attitudes

We examined the extent of nonresponse bias in a population of physicians (N = 408) mailed a cancer attitudes questionnaire by comparing respondents' and nonrespondents' demographic and medical practice characteristics. In addition, we assessed the usefulness of pursuing a higher response rate through a followup mailing by comparing characteristics and responses of early (first mailing) and late (second mailing) respondents. Although the overall response rate increased from 35% to 58% as a result of the follow-up mailing, nonresponse bias was still present. U.S. medical school graduates and recently licensed physicians were more likely to return questionnaires than were other physicians. Characteristics and attitudes of respondents remained similar over the follow-up mailing effort. Follow-up mailings are worthwhile only if nonresponse bias is reduced. Researchers must focus their attention on increasing the participation of all types of physicians at all points in the data collection process.