Abstract
As part of a national mail survey of family physicians, we tested the individual and combined effects on initial response rates of a small nonmonetary incentive and an invitation to withdraw from the survey (2 X 2 fully crossedfactorial design). The effectiveness and efficiency of threefollowup contacts based on Dillman's Total Design Method were also examined in a timeseries design. Neither the incentive, a sticker, or the withdrawal provision alone nor the two treatments in combination with one another affected initial or final response significantly. The withdrawalprovision appeared to save a small amount in follow-up costs, although it was also associated with a decrease in the rates offinal response that approached signif icance. The threefollow-up mailings (postcard reminder, second questionnaire, and third questionnaire sent by certified mail) increased the initial response of 40.6% by 6.0, 11.3, and 15.0 percentage points respectively. Nonlabor costs associated with the postcard were $2.01 per returned questionnaire; for the secondfollow-up they were $7.34; and for the certified mailing, $6.25.