Secondary Logo

Journal Logo

Scholarly Perspectives

Avoiding the Virtual Pitfall: Identifying and Mitigating Biases in Graduate Medical Education Videoconference Interviews

Marbin, Jyothi MD1; Hutchinson, Y-Vonne JD2; Schaeffer, Sarah MD, MPH3

Author Information
doi: 10.1097/ACM.0000000000003914
  • Free

Abstract

In light of the COVID-19 pandemic, the Association of American Medical Colleges (AAMC) “strongly encourages medical school and teaching hospital faculty to conduct all interviews with potential students, residents, and faculty in a virtual setting.” 1 While video conference interviews (VCIs) can be both convenient and cost effective, they may also exacerbate existing biases and introduce new sources of bias into the recruitment process. Graduate medical education (GME) programs committed to equity and diversity in recruitment must be aware of these biases and implement strategies to mitigate them.

Recognizing the importance of a diverse physician workforce, the Accreditation Council for Graduate Medical Education requires that GME programs “engage in practices that focus on mission-driven, ongoing, systematic recruitment and retention of a diverse and inclusive workforce.” 2 Implicit bias, or the unconscious and automatic preference for a social group, 3 interferes with the recruitment of a diverse class of residents and fellows. For example, in-group bias—in which people prefer those who share similar experiences—has been shown to play a role in residency selection. 4 Biases against physical attributes, such as candidates who are “facially unattractive” and/or obese, can influence the selection process as well. 5 One study found that applicants who “smiled, wore glasses, and wore jackets in their photographs were more likely to match” 6 into residency.

To mitigate such bias, GME programs have implemented strategies including structured interview questions, hiding applicant photographs during the review process, and involving a diverse group of faculty in recruitment. 7,8 Some programs have adopted a more holistic review of applications, 9,10 which deemphasizes academic markers that can reflect structural bias, such as United States Medical Licensing Examination scores 11 and membership in medical honor societies. 12

Using VCIs for GME interviews has the potential to introduce new sources of bias and amplify existing biases. Unfortunately, there is a paucity of literature—and thus little guidance—on mitigating bias in VCIs. The speed at which the transition is being made to VCIs, the magnitude of the change, the lack of research around bias and VCIs, and the underlying stress on all involved related to the COVID-19 pandemic put programs at risk of implementing virtual interviews without fully considering the role bias may play in VCIs. In an effort to support equitable implementation of VCIs, we highlight biases that may be exacerbated by VCIs and share strategies to mitigate them. These strategies are guided by one member of our author team (Y.H.), who has many years of professional expertise helping organizations identify and reduce the impact of bias, and by both other authors (J.M. and S.S.), who have led institutional efforts related to diversity and recruitment.

VCIs Can Promote Diversity and Connection

VCIs can be beneficial for both applicants and programs. Several studies have explored the feasibility of VCIs for residency and fellowship programs and have found them to be cost saving, timesaving, and generally acceptable to both faculty and applicants. 13–16 VCIs can reduce the cost of applicant recruitment for programs, in some cases by over 50%. 17 The reduced financial burden on applicants allows them to apply to more programs and to schedule interviews more easily as travel is not necessary. 13,17,18 This may allow for more socioeconomic diversity among applicants.

Video conferences may also allow applicants to interact with a diverse group of faculty and residents during the interview process through participating in virtual affinity groups. Applicants may have easier access to faculty and trainees with similar experiences (e.g., those who are first-generation professionals, underrepresented in medicine, parents, or working with disabilities). These affinity groups can help applicants build a sense of community with like-minded individuals within the program.

VCIs Can Introduce New Sources of Bias

Relying on technology may introduce new sources of bias into the interview process.

“Digital redlining,” or decreased access to broadband internet in marginalized communities, is linked to existing structural racial and/or socioeconomic disparities. 19,20 Digital redlining may limit access to broadband internet for some applicants. In addition, applicants may not be able to afford broadband connections or high-quality cameras, microphones, or computers. Low internet bandwidth and hardware or software problems can lead to glitches in video conferencing that hamper image and sound quality, leading to dysfluent communication. Communication delays of just over 1 second can cause conversation partners to be perceived as less extroverted, conscientious, and attentive. 21 These glitches can affect interview scores; even when told to disregard audiovisual (AV) quality, interviewers still gave higher scores to applicants with higher AV fluency. 22 Poor internet connections or audio quality may further disadvantage applicants who speak English with an accent 23 or for whom English is a second language. VCIs may therefore disadvantage applicants who lack access to technology or have poor internet connections.

The use of video cameras for interviews introduces another potential source of bias. Although there are fewer data published in this area, what we do know should give us pause. The concept of the “coded gaze” describes how bias can be embedded in machine learning. For example, by using nondiverse datasets to teach machines to recognize human faces, developers normalize whiteness, which can affect how people with darker skin appear on video. 24 One study found that facial analysis software had an error rate of 0.8% for light-skinned men but ranged from 20.0% to 34.7% for dark-skinned women. 25 If a camera is not programmed for optimal facial recognition, the subtleties of expression and nonverbal communication may be lost. Some cameras are calibrated to recognize white skin as default and require additional corrective lighting for people with darker skin. 26 In the absence of published studies, anecdotal experience and stories from social media reveal an extra burden on people of color to ensure they have the proper lighting to show up well on camera. This puts additional stress on an already high-stakes interview experience. As we increase our reliance on technology, we should be mindful that computer algorithms underlying technology are subject to the same racial biases we see in American society. 27

Time zone differences pose a practical challenge in VCIs and can disadvantage some applicants. Interviewing applicants in different time zones may lead to confusion in interview times. International applicants may be forced to interview during unusual hours to accommodate program interview schedules.

Video Interviews May Amplify Implicit Biases

Videoconference communication requires an increased cognitive load; this increased cognitive load can make interviewers lean more heavily on implicit biases when assigning interview scores to applicants.

Cognitive load is the “mental activity imposed on working memory.” 28 The phenomenon known as “Zoom fatigue,” or the exhaustion we experience after videoconference meetings, is due in part to the increased cognitive load of communicating over videoconference. 29–32 There are a number of factors adding to the burden of information processing in videoconferences. Eye gaze and gestures function differently during in-person interactions than they do in virtual settings. 33 In normal conversation, we maintain and break eye contact periodically; in videoconferences, we focus on the screen for longer periods of time to signify interest and attention. 34 Facial expressions, body language, and tone can all be interpreted differently over videoconference. 35 Because our visual frame to observe others is limited to their shoulders and face, we lose many of the nonverbal communication cues we normally use; in the absence of these, we do more mental work to elicit meaning from conversation. Video conferences also give us a constant window into our own performance; we are not used to watching ourselves as we speak, and our attention to our own appearance as we talk and gesture can take up mental bandwidth. 36 The combination of all of these factors increases our cognitive load during video communication.

When our cognitive load is increased, we automatically rely on implicit associations to help us process information. 29,37 When cognitive load is high, interviewers may unconsciously resort to social stereotypes, including appearance and voice, to make decisions. They may emphasize superficial factors—such as hometown, hobbies, or alma maters—to connect with applicants, thus falling into the trap of the “in-group” or “affinity” bias. Perceived social similarities may influence how interviewers score applicants. 4

Implicit bias may also be triggered in the setting of high cognitive load as interviewers are exposed to previously unavailable environmental cues about applicants. Applicants may have child- or elder-care responsibilities and may not have access to a quiet, private space in which to conduct interviews. Interviewers may make assumptions about applicants based on this window into their living situations and may fill in the gaps with assumptions rooted in bias as they assign interview scores.

Applicants can also be subject to bias during interviews. Interviewers sharing aspects of their personal lives (verbally or via images in the background of their video frame) may think they are making themselves approachable but may instead alienate applicants, particularly if the “shared” experiences reflect privileges that have historically been limited to a subset of individuals. Such nonverbal cues can send implicit—and explicit—messages to applicants about belonging and inclusion. Table 1 provides examples of how other types of biases can show up in VCIs.

Table 1
Table 1:
Potential Biases in Graduate Medical Education Interviews and Examples in VCIs

Mitigating Bias in VCIs

Programs seeking to implement equitable interview practices in VCIs should follow existing best practices for mitigating bias in interviews, including using standardized interview questions, 10,38 increasing the involvement of diverse faculty in interview process, 8 and avoiding assessment of vague, subjective qualities like “fit.” 39

There are a number of tip sheets produced by organizations including the AAMC 40–42 that offer applicants, interviewers, and programs technology tips, recommendations for backgrounds, advice for interacting in an interview setting, and so forth. We refer applicants, interviewers, and programs to these tip sheets for their helpful suggestions.

Our recommendations in Table 2 focus specifically on mitigating bias in VCIs. Because of the dearth of research in this area, some of our recommendations are based on anecdotal experience. Given the power dynamics and differential resources between interviewees and interviewers, we put the burden of action on programs and not on individual applicants. Examples of program-level strategies include making interviewers aware of technology-related inequity, offering back-up plans if technology fails, and advocating for medical schools to provide quiet, protected spaces with strong internet connections and high-quality hardware for applicant interviews. Additionally, programs can help interviewers become aware of the relationship between cognitive load and implicit associations and can offer solutions including hiding the image of speaker’s own face to reduce distractions, avoiding multitasking, and standardizing virtual backgrounds.

Table 2
Table 2:
Suggested Strategies for Graduate Medical Education Programs to Mitigate Bias in VCIs

We acknowledge our own discomfort in recommending behavioral changes to address structural inequities. At a systems level, programs and medical societies should advocate for upstream solutions to structural problems such as digital redlining. 20 This shared activation and investment can ultimately help create a more equitable system.

Concluding Remarks

With increasing use of VCIs during the COVID-19 pandemic, a change in the way we conduct interviews is upon us. While VCIs offer benefits to both applicants and programs, we must also remain mindful of the additional biases that can come with them. Additional research is needed in this area to better understand how bias appears in VCIs and to develop additional strategies to mitigate bias. For now, by acknowledging these biases and making intentional efforts to mitigate them, we can continue to work toward equity and inclusion in recruitment.

References

1. Association of American Medical Colleges.. AAMC Resources for Residency Program Selection. https://www.aamc.org/what-we-do/mission-areas/medical-education/conducting-interviews-during-coronavirus-pandemic#program. Accessed December 15, 2020
2. Accreditation Council of Graduate Medical Education. Common Program Requirements (Residency) Sections I-V: Table of Implementation Dates. https://www.acgme.org/Portals/0/PFAssets/ProgramRequirements/CPRResidencyImplementationTable.pdf. Accessed December 15, 2020
3. Haider AH, Sexton J, Sriram N, et al. Association of unconscious race and social class bias with vignette-based clinical assessments by medical students. JAMA. 2011; 306:942–951
4. Bass A, Wu C, Schaefer JP, Wright B, McLaughlin K. In-group bias in residency selection. Med Teach. 2013; 35:747–751
5. Maxfield CM, Thorpe MP, Desser TS, et al. Bias in radiology resident selection: Do we discriminate against the obese and unattractive? Acad Med. 2019; 94:1774–1780
6. Corcimaru A, Morrell MC, Morrell DS. Do looks matter? The role of the Electronic Residency Application Service photograph in dermatology residency selection. Dermatol Online J. 2018; 2413030/qt5qc988jz
7. Wusu MH, Tepperberg S, Weinberg JM, Saper RB. Matching our mission: A strategic plan to create a diverse family medicine residency. Fam Med. 2019; 51:31–36
8. Tunson J, Boatright D, Oberfoell S, et al. Increasing resident diversity in an emergency medicine residency program: A pilot intervention with three principal strategies. Acad Med. 2016; 91:958–961
9. Association of American Medical Colleges.. Roadmap to Excellence: Key Concepts for Evaluating the Impact of Medical School Holistic Admissions. Washington, DC: Association of American Medical Colleges, 2013
10. Aibana O, Swails JL, Flores RJ, Love L. Bridging the gap: Holistic review to increase diversity in graduate medical education. Acad Med. 2019; 94:1137–1141
11. Edmond MB, Deschenes JL, Eckler M, Wenzel RP. Racial bias in using USMLE step 1 scores to grant internal medicine residency interviews. Acad Med. 2001; 76:1253–1256
12. Boatright D, Ross D, O’Connor P, Moore E, Nunez-Smith M. Racial disparities in medical student membership in the Alpha Omega Alpha Honor Society. JAMA Intern Med. 2017; 177:659–665
13. Shah SK, Arora S, Skipper B, Kalishman S, Timm TC, Smith AY. Randomized evaluation of a web based interview process for urology resident selection. J Urol. 2012; 187:1380–1384
14. Daram SR, Wu R, Tang SJ. Interview from anywhere: feasibility and utility of web-based videoconference interviews in the gastroenterology fellowship selection process. Am J Gastroenterol. 2014; 109:155–159
15. Pasadhika S, Altenbernd T, Ober RR, Harvey EM, Miller JM. Residency interview video conferencing. Ophthalmology. 2012; 119:426–426.e5
16. Hariton E, Bortoletto P, Ayogu N. Residency interviews in the 21st century. J Grad Med Educ. 2016; 8:322–324
17. Pourmand A, Lee H, Fair M, Maloney K, Caggiula A. Feasibility and usability of tele-interview for medical residency interview. West J Emerg Med. 2018; 19:80–86
18. Temple E, Lagzdins M. Streamlining the residency interview process using web-based teleconferencing. Am J Health Syst Pharm. 2014; 71:697–701
19. Hoffman D, Novak T, Schlosser A. The evolution of the digital divide: How gaps in internet access may impact electronic commerce. J Comput Mediat Commun. 2000; 5:JCMC534.
20. Jackson D. Digital Redlining: How Major American Communication Companies Are Controlling Who Gets Broadband Access or Not. Atlanta Black Star. https://atlantablackstar.com/2017/04/06/digital-redlining-major-american-communication-companies-controlling-gets-broadband-access-not. Published April 2017 Accessed December 15, 2020
21. Schoenenberg K, Raake A, Koeppe J. Why are you so slow? Misattribution of transmission delay to attributes of the conversation partner at the far-end. Int J Hum Comput Stud. 2015; 72:477–487
22. Fiechter JL, Fealing C, Gerrard R, Kornell N. Audiovisual quality impacts assessments of job candidates in video interviews: Evidence for an AV quality bias. Cogn Res Princ Implic. 2018; 3:47
23. Lippi-Green R. English with an accent: Language, ideology, and discrimination in the United States. Lang Probl Lang Plan. 1999; 23:209–210
24. Bermudez-Silverman P. The coded gaze: Algorithmic bias? What is it and why should I care? https://medium.com/africana-feminisms/the-coded-gaze-algorithmic-bias-what-is-it-and-why-should-i-care-51a416dbc3f3. Published May 2018 Accessed December 15, 2020
25. Buolamwini J. Gender shades: Intersectional accuracy disparities in commercial gender classification. Conference on fairness, accountability and transparency. Proc Mach Learn Res. 2018; 81:77–91
26. Lewis S. The racial bias built into photography. New York Times. https://www.nytimes.com/2019/04/25/lens/sarah-lewis-racial-bias-photography.html. Published April 2019 Accessed December 15, 2020
27. Heilweil R. Why algorithms can be racist and sexist. Voxhttps://www.vox.com/recode/2020/2/18/21121286/algorithms-bias-discrimination-facial-recognition-transparency. Published February 2020 Accessed December 15, 2020
28. Burgess DJ. Are providers more likely to contribute to healthcare disparities under high levels of cognitive load? How features of the healthcare setting may lead to biases in medical decision making. Med Decis Making. 2010; 30:246–257
29. Sander L, Bauman O. Zoom fatigue is real—Here’s why video calls are so draining. Ideas.Ted.com. https://ideas.ted.com/zoom-fatigue-is-real-heres-why-video-calls-are-so-draining. Published May 2020 Accessed December 15, 2020
30. Bailenson J. Why Zoom meetings can exhaust us. Wall Street Journal. https://www.wsj.com/articles/why-zoom-meetings-can-exhaust-us-11585953336. Published April 2020 Accessed December 15, 2020
31. Kadaba L. “Zoom fatigue” is real. Why video meetings strain your brain and how to fix it. Philadelphia Inquirer. https://www.inquirer.com/health/coronavirus/zoom-fatigue-science-brain-how-to-combat-20200520.html. Published May 2020 Accessed December 15, 2020
32. Fosslien L, Duffy M. How to combat Zoom fatigue. Harvard Business Review. https://hbr.org/2020/04/how-to-combat-zoom-fatigue. Published April 29, 2020 Accessed December 15, 2020
33. Fussell SR, Benimoff NI. Social and cognitive processes in interpersonal communication: Implications for advanced telecommunications technologies. Hum Factors. 1995; 37:228–250
34. Ramzi L. Zoom doom: I feel so uncomfortable making eye contact in work video calls. Vogue.com. https://www.vogue.com/article/zoom-video-chat-eye-contact. Published May 2020 Accessed December 15, 2020
35. Jing M. The reason Zoom calls drain your energy. Remote control. BBC. https://www.bbc.com/worklife/article/20200421-why-zoom-video-chats-are-so-exhausting. Published May 2020 Accessed December 15, 2020
36. Geall L. Have you got Zoom fatigue? Why you’re finding video calls so exhausting. Stylist.com. https://www.stylist.co.uk/life/zoom-fatigue-video-call-virtual-drinks-exhaustion-tiring/376846. Published April 2020 Accessed December 15, 2020
37. van Ryn M, Saha S. Exploring unconscious bias in disparities research and medical education. JAMA. 2011; 306:995–996
38. Bohnet I. How to take the bias out of interviews. Harvard Business Review. https://hbr.org/2016/04/how-to-take-the-bias-out-of-interviews. Published April 2016 Accessed December 15, 2020
39. Shappell E, Schnapp B. The F word: How “fit” threatens the validity of resident recruitment. J Grad Med Educ. 2019; 11:635–636
40. Association of American Medical Colleges.. Virtual Interviews: Tips for Medical School Applicants. https://www.aamc.org/system/files/2020-05/Virtual_Interview_Tips_for_Medical_School_Applicants_05142020.pdf. Accessed December 15, 2020
41. Association of American Medical Colleges.. Virtual Interviews: Tips for Program Directors. https://www.aamc.org/system/files/2020-05/Virtual_Interview_Tips_for_Program_Directors_05142020.pdf. Accessed December 15, 2020
42. Association of American Medical Colleges.. Virtual Interviews: Tips for Interviewers. https://www.aamc.org/system/files/2020-05/Virtual_Interview_Tips_for_Interviewers_05142020_0.pdf. Accessed December 15, 2020

References cited only in the tables

43. Williams J. The 5 biases pushing women out of STEM. Harvard Business Review. https://hbr.org/2015/03/the-5-biases-pushing-women-out-of-stem. Published March 2015 Accessed December 15, 2020
    44. Harasym PH, Woloschuk W, Mandin H, Brundin-Mather R. Reliability and validity of interviewers’ judgments of medical school candidates. Acad Med. 1996; 71suppl 1S40–S42
    45. Jhangiani R, Tarry H. Principles of Social Psychology. 1st International ed. Victoria, BC: BCcampus, 2014
      46. Bendick M, Nunes A. Developing the research basis for controlling bias in hiring. J Soc Issues. 2012; 238–262
      47. Dunham Y. Mere membership. Trends Cogn Sci. 2018; 22:780–793
      48. Eagleson G, Waldersee R, Simmons R. Leadership behaviour similarity as a basis of selection into a management team. Br J Soc Psychol. 2000; 39(pt 2):301–308
        Copyright © 2021 by the Association of American Medical Colleges