Methodology and Sample Representativeness of the CERA 2026 Clerkship Director Survey

The Council of Academic Family Medicine (CAFM) Educational Research Alliance (CERA) aims to provide infrastructure to support family medicine educational research that is both rigorous and generalizable.¹ CERA’s goals are to increase collaboration, provide mentorship, and offer a data clearinghouse for CAFM members to increase scholarly output.²

CERA has conducted an annual survey of clerkship directors (CDs) since 2012.³ Results of the CERA CD survey series have been disseminated as 56 published papers and 57 conference presentations as of the time of this writing.^4,5 This paper presents the methods and demographic results of the 2026 CD Survey to assess any potential response bias between the sample and the sampling frame. In addition, this paper highlights the cognitive interviewing process that the survey director has been using before finalizing the survey questions. Cognitive interviewing involves asking a colleague to read each question aloud and describe how they would answer the question.⁶ The goal of cognitive interviews is to identify issues with question design that might result in challenges with responding accurately to the question, which would impact the conclusions drawn from the results of the survey.

The methodology of the CERA CD survey has been described in detail previously.⁷ CAFM members were invited to propose survey questions for inclusion in the CD survey from January 2026 to February 2026. Nine submissions were received and underwent peer review; five were accepted based on peer-review rankings. Approved projects were assigned a CERA mentor to help refine the survey questions. The resulting questions were pilot-tested through a cognitive interview with an academic physician who was not part of the survey sample, conducted by the survey directors (K.S. and A.K.), and beta-tested through the SurveyMonkey platform (Symphony Technology Group, Menlo Park, CA).

SurveyMonkey distributed invitations to 184 individuals through an institutional email. Twenty-eight emails could not be delivered and were resent via the survey director’s personal email. Ten CDs who had previously opted out of SurveyMonkey were emailed directly; nine provided replacement contacts, who were then invited. Three undeliverable addresses and four “not the CD” responses without replacements were removed, yielding a final sample of 187 recipients.

The survey was open from June nine to July 17, 2026. Nonresponders and partial respondents were sent five weekly reminders to respond, with a final sixth reminder on the last day the survey was open.

We used the Society of Teachers in Family Medicine (STFM) membership database to generate demographics from the CD sampling frame. This database contained demographics for 155 CDs (151 in the United States and four in Canada). The remaining 32 email addresses did not have an STFM record, so base demographics could not be obtained for those individuals. Therefore, demographic information is available for 82% of the sampling frame. Individuals who responded to the CD survey were included in the sample even if their demographics were not available from the STFM demographic survey. Existing STFM demographic information was not updated with responses from the CERA CD survey.

The proportions of the location, gender, race/ethnicity, and self-identified URIM status of the sampling frame vs. the sample were compared via χ² tests and Fisher’s exact test when indicated. The project was approved by the American Academy of Family Physicians’ Institutional Review Board in June 2026.

For all proposed modules, research teams revised all original questions during the peer review and CERA mentorship process (Table 1). The revised questions were submitted for cognitive interviewing and pilot testing, which resulted in 46% (23/50) of the questions being further revised before the final survey was disseminated. Table 2 lists accepted topics for the 2026 CD Survey.

Surveys were returned. Two respondents only answered the initial question, “Are you the clerkship director?” and were counted as nonresponses. Therefore, 69 surveys had complete data, which resulted in an overall response rate of 36.9% (69/187). The proportions did not differ significantly by geographic location, gender, race/ethnicity, or URiM status between the sampling frame and the sample (Table 3).

The yearly CD survey provides insight into potentially interesting topics for undergraduate medical educators. A strength of the CERA survey is the ability to target a national audience of CDs in a structured format.^7,8 This year’s CD survey response rate of 36.9% is significantly lower than in prior years, and demographic analysis indicates that the sample is still representative of US and Canadian CDs in the sampling frame across all assessed demographic areas.

The historical response rate for the clerkship director survey has ranged between 48.8% and 83.1%. Over the past few survey cycles, CERA has been making efforts to identify and include family medicine clerkship directors from osteopathic medical schools. The denominator for the survey of 187 potential respondents was the highest ever for the survey (previous surveys ranged from 129 to 174 potential respondents). Despite the increased potential respondents, the numerator of responses was the lowest it has been at 69 responses (previous surveys ranged from 78 to 123 responses). Survey administrators kept this survey open for one week longer than normal. There was no significant survey abandonment for those who did respond.

The CERA CD survey questions underwent significant revisions from the original question submissions. The cognitive interviewing process further refined 46% of the questions for clarity, standardization, and specificity, resulting in notable question modifications. Other CERA surveys should consider adopting the cognitive interview process in their methodologies.

There are several limitations to note. CD demographic data are limited to those provided by STFM and are all self-reported. Self-reported demographic data for STFM faculty members have a nonresponse rate below 20%. Additionally, although demographic data on CDs are informative, they do not necessarily reflect the characteristics of the students or institutions they serve, and it remains unclear whether CDs are demographically representative of their local contexts.

Taken together, the 2026 CERA CD survey demonstrates demographic representativeness relative to STFM-reported CD data, providing a sound basis for generalizing results to this population.

The authors acknowledge Ray Biggs, Rhiannon Johnson, and April Davies for their assistance in coordinating the technical aspects of survey creation and dissemination.

The authors are members of the CERA Steering Committee but have no financial disclosures.