Authors’ Response to “Captive Samples Are Not the Answer to Survey Response Rates”

We were pleased to have received Dr Ringwald’s letter in response to our article “Optimizing Survey Response Rates in Graduate Medical Education Research Studies.” ¹ We agree that the participants we reported on in our paper were likely more motivated than others, due to the application and selection processes used in these studies. We do, however, believe that more can and should be done by survey researchers to create a meaningful evaluation culture. That culture, which we think would improve response rates, includes raising the importance of contributing to survey research as a component needed to advance the science of education.

Currently, no such culture exists, and educational researchers often expect to get poor response rates rather than exploring what they can do to enhance those rates. What could we do to add value to survey participants? We have anecdotally found that residency directors and other survey participants in the health professions value receiving confidential results that compare findings from their programs or disciplines to the aggregate of all participants. By sharing these, they actually see findings presented in ways they care about and can use to improve or otherwise enhance their programs. Why would potential participants complete a survey that appears to enter their responses into a black hole that never provides anything meaningful back to them? This is the type of relationship we believe has value.

It is true that some studies have shown that response rates are not always inversely correlated with nonresponse bias, but this claim cannot be universally made because it depends on the study population and the specific topic being surveyed. ^{2, 3} We believe that simply having and communicating a response rate goal is helpful, and conveying that goal, and our progress toward it, to potential respondents is also helpful.

We did not indicate in our paper that our findings were generalizable to the whole community involved in family medicine residency training. Rather, we cited that as a limitation. ¹ We believe that having a high response rate is much better than basing findings on low response rates, which often happens. We agree that adopting standardized guidelines for developing, analyzing, and reporting survey research data would help, but we stand firm on our statements that survey researchers also have a role in improving response rates.