OBJECTIVES: This study assessed efforts to increase response rates to a mailed physician survey and examined whether, as a result, nonresponse bias was reduced. METHODS: Randomly selected physicians and geneticists were mailed a questionnaire concerning genetics knowledge and attitudes. In the final but not the pilot survey, a $25 incentive and intensive follow-up were used to increase the response rate. RESULTS: The response rate from physicians in the final survey was 64.8% (n = 1140), compared with 19.6% in the pilot test (n = 69). Sample representatives in sociodemographic and practice characteristics was improved by follow-up. Respondents recruited with more difficulty did not differ on the principal outcome variable, genetics knowledge, except on one subscore. Pilot study and final survey respondents did not differ in knowledge. CONCLUSIONS: Although the effect of increased response rates on the principal outcome variable in this study was minimal, this may not be the case for other studies. Every effort should be made to attain as high a response rate as is practical and to establish that respondents are representative of the population being sampled.
OBJECTIVES: This study assessed efforts to increase response rates to a mailed physician survey and examined whether, as a result, nonresponse bias was reduced. METHODS: Randomly selected physicians and geneticists were mailed a questionnaire concerning genetics knowledge and attitudes. In the final but not the pilot survey, a $25 incentive and intensive follow-up were used to increase the response rate. RESULTS: The response rate from physicians in the final survey was 64.8% (n = 1140), compared with 19.6% in the pilot test (n = 69). Sample representatives in sociodemographic and practice characteristics was improved by follow-up. Respondents recruited with more difficulty did not differ on the principal outcome variable, genetics knowledge, except on one subscore. Pilot study and final survey respondents did not differ in knowledge. CONCLUSIONS: Although the effect of increased response rates on the principal outcome variable in this study was minimal, this may not be the case for other studies. Every effort should be made to attain as high a response rate as is practical and to establish that respondents are representative of the population being sampled.