Nonresponse Bias in a Mail Survey of Physicians
McFarlane, E. S., Olmsted, M., Murphy, J. J., & Hill, C. A. (2006, May). Nonresponse Bias in a Mail Survey of Physicians. Presented at American Association for Public Opinion Research 61st Annual Conference, Philadelphia, PA.
In the absence of additional information, response rates are often used alone as a proxy measure for survey quality. However, total survey error is comprised of many factors, including both sampling and nonsampling errors; nonsampling errors may arise from responders and nonresponders alike (Biemer and Lyberg, 2003). Despite low response rates common in physician surveys, it is unclear to what extent response bias exists among studies with physicians. The effects of nonresponse bias in a brief mail survey of physicians were examined. In a national sample of board-certified physicians, a short survey was mailed asking physicians to nominate the five best hospitals in their specialty regardless or cost or location. Up to three follow-ups were mailed to nonresponders to gain participation. The final response rate was 47.3%. As more than half of the sample were nonresponders, there was potential for bias if nonresponders differed significantly from responders. Willingness to respond, measured by overall response and timing of response, was analyzed with respect to several demographic variables including gender, region, specialty, urbanicity, and survey length. Next, respondent outcome measures were analyzed with respect to the demographic variables and timing of response. The outcome measures of interest were (1) nominating a top hospital in their specialty, (2) nominating two top hospitals in their specialty, and (3) nominating only hospitals in their region. This paper will describe the factors associated with nonresponse in physician surveys, the direction of nonresponse bias, and the relationship between response bias and response rates.