Physician Response to a Mailed Survey an Experiment in Timing of Payment

Abstract This article reports the results of an experiment in obtaining physician response to a mailed questionnaire. Each physician was eligible for a payment of 20. A randomly se-lected half received the payment with their initial question-naire and cover letter; half were told they would receive their payment after they completed and returned the question-naire. The same mail and telephone foliowup procedures were used for both groups. Overall, prepayment had signifi-cant positive effects on response rates. This paper examines these effects in terms of response rates for various special-ties, field efficiencies, cost, and representativeness of the sample. The ability to make unbiased inferences about populations based on sample survey data depends on having complete information about all selected sample units or establishing that the nonrespondents do not differ from respondents in any important way (Fowler, 1984). Since it is difficult to rule out biased nonresponse, it is desirable to obtain high response rates. This can be especially challenging with some populations. Physi-cians, for example, are widely believed to be a difficult population from which to collect survey data (Sudman, 1985). Physicians are frequently