Recruiting survey respondents to mailed surveys: controlled trials of incentives and prompts

Abstract

Three controlled experiments tested the efficacy of 1) a postcard or telephone prompt, 2) a lottery, 3) monetary incentives, and 4) questionnaire length to recruit adult survey respondents to a random sample of residences in San Diego, California, during 1986-1988. In experiment 1, the group randomly assigned to receive a telephone call prenotification plus the lottery incentive responded 26-66% more frequently than did controls (p = 0.02) after a single mailing. The postcard plus lottery was 17-54% more effective than no intervention with controls (p = 0.05). A second mailing of the survey weakened these effects. In experiment 2, the group randomly assigned to receive a two-page survey with the lottery announcement responded 69% more frequently after one mailing (p = 0.03) and 53% more frequently after a second mailing (p = 0.04) than the group that received an eight-page survey without the lottery incentive. The shorter form alone or the lottery alone did not increase response rates significantly relative to the long form without a lottery. In experiment 3, a monetary incentive of $5.00 contingent on response to the second mailing of the survey increased the rate of response from initial "nonresponders" by 100% relative to control who received no incentive (p = 0.03) and 75% over those who received $1.00 not contingent on response (p = 0.04). Little sampling bias and no reactivity was attributable to the recruitment procedures.