Secondary Logo

Journal Logo

Advanced Search
Research Reports

Are Female Applicants Rated Higher Than Males on the Multiple Mini-Interview? Findings From the University of Calgary

Ross, Marshall MD; Walker, Ian MD; Cooke, Lara MD, MSc; Raman, Maitreyi MD, MSc; Ravani, Pietro MD, PhD; Coderre, Sylvain MD, MSc; McLaughlin, Kevin MBChB, PhD

Author Information

M. Ross is a resident, Department of Emergency Medicine, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada.

I. Walker is clinical associate professor, Department of Emergency Medicine, and director of admissions, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada.

L. Cooke is associate professor, Department of Clinical Neurosciences, and associate dean of continuing medical education, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada.

M. Raman is clinical associate professor, Department of Medicine, and associate director of admissions, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada.

P. Ravani is professor, Department of Medicine, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada.

S. Coderre is professor, Department of Medicine, and associate dean of undergraduate medical education, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada.

K. McLaughlin is professor, Department of Medicine, and assistant dean of undergraduate medical education, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada.

Funding/Support: None reported.

Other disclosures: None reported.

Ethical approval: This study was granted ethical approval by the Conjoint Health Research Ethics Board at the University of Calgary.

Previous presentations: The findings of this study were presented at the Canadian Conference of Medical Education, Vancouver, British Columbia, Canada, April 26, 2015.

Correspondence should be addressed to Kevin McLaughlin, Office of Undergraduate Medical Education, Cumming School of Medicine, University of Calgary, Health Sciences Centre, 3330 Hospital Dr. NW, Calgary, Alberta, Canada T2N 4N1; telephone: (403) 220-4252; e-mail: [email protected].

doi: 10.1097/ACM.0000000000001466
  • Free

Abstract

Since approximately 95% of students entering medical school will ultimately graduate, the medical school admissions process is considered by many as our best (and perhaps only) opportunity to select the “ideal” candidates for a career in medicine.1,2 For applicants, the interview process may be a life-defining moment, a fact that weighs heavily on those of us involved in the selection process as we try to predict whether a given applicant is more likely to mature into a CanMEDS-proficient physician than his or her peers.3 Such predictions are notoriously difficult and inaccurate,4,5 but because society needs physicians—and there are many more applicants to medical school than positions available—we need a selection process and must strive to make this as reliable, valid, and “fair” as possible.

In addition to a short-listing process that rates indicators of prior academic performance and nonacademic attributes, most medical schools also use some form of interview to assess other attributes that may be relevant for future physicians. Historically, this involved a selection committee conducting a single interview of each candidate, and there was often significant variation—both within and between centers—in the makeup of committees, the content discussed, the degree of structure of the interviews, and the process of selecting the best candidates.6,7 And, although this approach was convenient, concerns arose about the fairness of this process given the limited scope of a single interview and the proportion of variance in scores attributable to the interviewers.8,9 To address some of the limitations of the traditional interview format, in particular issues of context-specificity of performance and interviewer-related biases, the multiple mini-interview (MMI) was introduced at McMaster University.10 Based on the format of the objective structured clinical examination, the MMI exposes candidates to a variety of contexts and examiners and is designed to provide a structured assessment of a range of attributes relevant to being a physician. The MMI format has been recognized as a major innovation in medical education and, by being adopted in many other centers, has revolutionized the admissions process.11–14

The inclusion of the MMI to the medical school admissions process has resulted in a selection process that appears to be more reliable,10,14–16 more acceptable to applicants,13,14 less rater dependent,10 and more predictive of future performance than the traditional interview format.14,15,17,18 Despite ongoing concerns regarding the validity and dimensionality of the MMI,14,19 these impressive findings have fomented the incorporation of the MMI into the selection process in many medical schools in North America, Europe, and Australia.11–13 Yet every change carries the possibility of unintended outcomes, such as changing the demographic characteristics of the medical profession. For example, because females are typically rated higher than males in communication as soon as this skill can be reliably assessed,20 one might predict that females would outperform males whenever we assess the performance of applicants on tasks that involve communication.21,22 When the MMI was first piloted at McMaster there was no significant association between applicant gender and MMI ratings, although the sample size in this initial study was relatively small (n = 117).10 Subsequent studies also reported no difference in ratings of female and male applicants,23–26 although, once again, the sample size in these studies was small, with studies often including fewer than 30 male applicants.24–26 When introducing the MMI at the University of Calgary Cumming School of Medicine, Brownell and colleagues11 reported that, in the opinion of both applicants and interviewers, “the MMI was free of gender bias,” and this conclusion was also drawn in subsequent studies.15,16,25 Interestingly, in a more recent study from a single center in Scotland, a 10-station MMI was reported as reliable, and the process was “acceptable to all parties” (more than 90% of candidates and raters agreed or strongly agreed with a fairness statement).13 The authors of this study did not comment on gender differences in MMI scores, but provided data that allowed male and female scores to be compared. In 2010, 280 of 477 (58.7%) candidates interviewed at that center were female, and female candidates were rated significantly higher than male candidates on the MMI (109.1 [SD 11.1] vs. 105.3 [SD 13.1], P < .0007, Cohen d = 0.31).13

Here we describe two studies in which we compared MMI ratings for female and male applicants to our medical school, and explored the impact of MMI ratings on the ranking of applicants. Our first study, a cumulative meta-analysis, was designed to allow us to study gender differences in MMI ratings with increased statistical power and precision.27,28 In our second study, we compared MMI ratings for female and male medical school applicants in a single year using multivariate analysis in which we adjusted for other variables that are associated with MMI ratings. We then performed sensitivity analysis to explore the impact of differential weighting assigned to MMI ratings on the odds of a female applicant being included in the top 150 ranked applicants to our medical school (our typical intake each year is 150 students). We predicted that if there were gender differences in MMI ratings, then females would be rated higher than male applicants, and increasing the weight assigned to MMI ratings would increase the odds of a female applicant being included in the top 150 ranked applicants.

Study 1

Method

Participants.

Our participants were all applicants (n = 3,079) to the Cumming School of Medicine, University of Calgary, who completed the MMI in years 2010 to 2014. This study was granted ethical approval by the Conjoint Health Research Ethics Board at the University of Calgary.

Materials.

The data used in this study were applicants’ MMI ratings for years 2010 to 2014, which were collected prospectively as part of our admissions selection process. Over this five-year period, the number of stations on our MMI ranged from 9 to 12.

Procedures.

For years 2010 to 2014, we noted how many female and male applicants completed the MMI and the mean and SD for MMI ratings for females and males. We analyzed our data as a cumulative meta-analysis because this allowed us to study the additive effect of each additional year of MMI data on our findings, thus allowing us to identify the time point at which we had sufficient data to infer gender differences in MMI ratings.

Statistical analysis.

We converted means and SDs into standardized mean differences and used Hedges g for our effect size calculation.28 We then performed cumulative meta-analysis using the random-effect model, using STATA version 11 statistical software (STATA Corporation, College Station, Texas).

Results of Study 1

The number of applicants (and percent female) completing the MMI for years 2010 to 2014 was 640 (47.5%), 633 (50%), 641 (52.6%), 639 (55.2%), and 526 (54%), respectively. The cumulative standardized mean difference between female and male applicants’ ratings is shown in Figure 1.

F1
Figure 1:
Cumulative meta-analysis showing the effect of adding each additional year of data on the standardized mean difference and 95% CI of MMI ratings for 3,079 female and male applicants, from a study of gender differences in MMI ratings, University of Calgary Cumming School of Medicine, 2010–2014. Results to the right of the vertical axis (standardized mean difference = 0) favor females, and results to the left favor males.Abbreviations: MMI indicates multiple mini-interview; SMD, standardized mean difference.

Discussion of Study 1

Our findings suggest that at our medical school, females are rated higher than males on the MMI. The change in the standardized mean difference over time implies that there were sufficient data to draw this conclusion in 2012 and that the addition of subsequent data simply confirms this finding with greater precision. There are, however, limitations to this type of univariate analysis—in particular, the fact that no other explanatory variables—that could confound or modify the effect of gender—were considered. To address these limitations, we performed our second study.

Study 2

Method

Participants.

Our participants were 1,254 individuals who applied to the Cumming School of Medicine, University of Calgary, in 2014, of whom 526 completed our MMI. This study was granted ethical approval by the Conjoint Health Research Ethics Board at the University of Calgary.

Materials.

The data used for this study were collected prospectively as part of our admissions selection process. These data included ratings of applicants’ files and subsequent MMI ratings. Applicants’ file data included their curriculum vitae, academic record (undergraduate grade point average [GPA]), performance on an aptitude test (Medical College Admission Test [MCAT]), and letters of reference. We also noted the gender and age of applicants, and for each MMI station we noted the gender of the interviewer and whether the interviewer was a trainee. For each of the 12 stations on the MMI there was a single attribute that we intended to assess, and the complete list of target attributes was as follows: conflict management/communication; attitude towards learning; responsiveness to feedback; decision-making ability/data interpretation; research ethics; communication skills; cultural competency; empathy; visual observation and creative thinking; resource management/project planning; honesty and integrity; and disclosure of error. Performance on the MMI was rated using a combination of a checklist and a global rating scale, which then contributed equally to the final rating for a given station.

Procedures.

Each applicant’s file was reviewed by 4 independent reviewers, from a panel of 60, who provided subjective scores for seven domains that were intended to represent the seven CanMEDS competencies.3 For example, academic record was rated as part of the Expert Role, and leadership experience was considered when rating the Manager Role. The file review scores were then compiled using the following weighting: 10% for each of the domains representing the seven CanMEDS competencies, 20% for GPA, and 10% for the verbal reasoning component of the MCAT. We ranked applicants based on mean file review scores and invited the top 526 applicants to attend our MMI. Our MMI stations were seven minutes long and were rated by a single interviewer.

Statistical analysis.

To compare MMI ratings for female and male applicants, we used an independent-sample t test with Cohen d as our measure of effect size. We performed multiple linear regression to study the association between applicant gender and MMI ratings after adjustment for other potential explanatory variables: applicant’s age, GPA, MCAT score (for verbal reasoning, physical sciences, and biological sciences), interviewer’s gender, and whether the interviewer was a trainee (medical student or resident). We included interactions between applicant and interviewer variables in our regression model and performed backward elimination to remove nonsignificant explanatory variables, beginning with interaction terms.

To explore the impact of the weighting given to the MMI rating on the odds of female applicants being offered a medical school position, we performed a sensitivity analysis where we combined file review scores and MMI ratings for each applicant. We considered six scenarios where MMI rating contributed 0%, 10%, 20%, 30%, 40%, or 50% to the final applicant score, and for each scenario we calculated the odds ratio by comparing the odds of a female applicant being in the top 150 ranked applicant versus those of a female being in the original 1,254 applicants. We used STATA version 11 statistical software (STATA corporation, College Station, Texas) for our statistical analyses.

Results of Study 2

Of 1,254 individuals who applied to the Cumming School of Medicine, University of Calgary, in 2014, 650 (51.8%) were female. Mean age of applicants was 24.4 (3.89) years, and 3.6% of applicants self-identified as Aboriginal. Of the 526 applicants invited for MMI following file review, 284 (54%) were female, and the gender distribution of applicants selected for MMI interview was not significantly different from that of the original applicant cohort (odds ratio of an applicant being female was 1.10; 95% CI [0.89, 1.37], P = .34).

The mean MMI rating for female and male applicants was 6.60 (SD 1.75) and 6.34 (SD 1.88), respectively (P < .01, d = 0.14). There were no interactions in our regression model, including no interaction between gender of applicant and interviewer (P = .94), but applicant’s gender, applicant’s age, interviewer’s status (trainee vs. nontrainee), and MCAT scores for verbal reasoning and biological science were associated with MMI rating. After adjusting for other variables, there was still a significant positive association between being a female applicant and MMI rating (regression coefficient 0.23, 95% CI [0.14, 0.33], P < .001). These data are shown in Table 1.

T1
Table 1:
Variables Associated With 1,254 Applicants’ MMI Ratings, From a Study of Gender Differences in MMI Ratings, University of Calgary Cumming School of Medicine, 2010–2014a

In our sensitivity analysis, we found that the gender breakdown of the top 150 applicants based solely on file review scores (i.e., when MMI rating did not contribute to overall applicant score) was not significantly different from that of the original application cohort (odds ratio of an applicant being female was 1.32; 95% CI [0.92, 1.91], P = .11). When varying the weight given to MMI rating, we found that whenever the MMI rating contributed 10% or more of the overall score, the odds of a female applicant being ranked in the top 150 students were significantly higher than for male applicants. These data are shown in Figure 2.

F2
Figure 2:
The impact of weighting assigned to MMI scores on the odds ratio (and 95% CI) of a female applicant being ranked in the top 150 applicants, out of 1,254 total applicants, from a study of gender differences in MMI ratings, University of Calgary Cumming School of Medicine, 2010–2014. Results to the right of the vertical axis (odds ratio = 1.0) favor females, and results to the left favor males.Abbreviation: MMI indicates multiple mini-interview.

Discussion of Studies 1 and 2

The results of our cumulative meta-analysis suggest that at our medical school, females are consistently rated higher than male applicants, and in our second study we found that assigning ≥ 10% weight to MMI ratings resulted in increased odds of female applicants being ranked in the top 150 applicants.

If females are rated higher than males on the MMI, then we should try to understand why. We would propose two possible explanations that have divergent implications for the validity of the MMI. The first is that females are more likely to demonstrate the attributes that the MMI is intended to capture. In general, when compared with males, females typically demonstrate better communication skills,20–22 are rated higher on certain aspects of critical thinking (such as open-mindedness and maturity),18,29 and make more ethical decisions.30,31 Because we expect females to outperform males on these tasks, one could argue that the fact that females are rated higher than males on the MMI is a source of validity for the MMI.32 However, the alternative explanation is that females are rated higher on the MMI because we expect them to be better at communication, critical thinking, and ethical decision making. In this scenario, females are rated higher as a result of observer expectancy, a type of bias that would reduce the validity of MMI ratings.

Observer expectancy or confirmation bias (also referred to as “self-fulfilling prophecy”33) refers to a process where preexisting beliefs subconsciously influence interpretation of data in a way that supports these preexisting beliefs,34,35 and there are many examples of observer expectancy bias in the psychology literature.36 For example, when asked to rate the performance of males and females on a muscular endurance task, both male and female college students tended to overestimate the performance of males and underestimate the performance of females,37 whereas when asked to rate the amount of smiling of males and females on video clips, psychology students overestimated smiling of females relative to males.38 These gender-based stereotypes are generated automatically and can be triggered by subtle manipulations, such as altering the voice output of computers or perceived gender of virtual humans.39,40 Being largely subconscious, observer expectancy may be difficult to detect and equally difficult to suppress—especially when observers are asked to rate ill-defined constructs based upon limited exposure, such as an eight-minute MMI station.41–43

Should we be concerned if the inclusion of the MMI (or any other source of data) creates a gender imbalance in medical school admissions? The answer to this question depends on the perspective taken, and we would propose patient-centered outcomes as the most meaningful way to judge the impact of any change to the medical school admissions process. If changes to the selection process produce a workforce that delivers higher-quality health care, a gender imbalance may be acceptable—and may even be necessary to improve health care delivery. But an admissions process that creates a gender imbalance with equal or worse health care outcomes is unacceptable. Previous studies have suggested that female physicians typically demonstrate greater empathy and use more positive statements than males when interacting with their patients21,44 and that this type of communication is associated with patients providing better historical data, reporting enhanced satisfaction and psychosocial health, and using less health care resources.45–48 However, these types of data do not provide direct evidence in support of any type of admissions process.

So, how can we demonstrate the impact of the selection process on health care outcomes? Interventions designed to improve health care outcomes are being implemented continuously, so a pre/post comparison of selection processes would be confounded by other interventions. A randomized comparison of different selection processes with longitudinal follow-up would be ideal, but this type of multicenter study may not be acceptable or feasible for many medical schools. Clearly, we face significant challenges in demonstrating the impact of the admissions process on outcomes that are both measurable and meaningful.5 Without knowing the impact of the MMI on health care outcomes, we cannot say whether gender differences in MMI ratings are acceptable or unacceptable—but we do know that the origins of gender imbalance in medical school admissions can be traced back much further than the history of the MMI. In the United States, males have consistently outnumbered female graduates.49 Since 1993–1994 (10 years before the description of the MMI), in Canada female students have outnumbered males in all but 2 years—after being consistently in the minority in the preceding 22 years.50 The reason for these gender imbalances and their impact on health care delivery are unknown, but are also worthy of further exploration.

There are some important limitations to our studies that we should highlight. First, this was a single-center undertaking, so our findings do not necessarily imply that females will be rated higher than males in other MMI stations at other institutions. As is the case with all observational studies, our study design limits us to reporting associations and then generating hypotheses on the cause of these associations. Further studies are needed to confirm and explain the findings of our studies—in particular, whether gender difference in ratings are due to observed versus expected differences in performance. Finally, the finding of these studies—gender differences in MMI ratings and the odds of being offered a position in medical school—is an unintended outcome of the MMI, and the significance of this outcome is unclear without consideration of intended outcomes. Further studies should explore the impact of the MMI-driven selection process on meaningful outcomes, such as health care delivery, to establish whether a gender imbalance related to MMI ratings is acceptable.

Conclusions

Replacing the traditional interview format with the MMI has improved many aspects of the medical school admissions process, including the reliability and validity of candidate selection. But the findings from our studies suggest that at our center, females are rated higher than male applicants and that the inclusion of MMI ratings in the overall scores for candidates increases the odds of a female applicant being offered a position in our medical school. We feel that this is an important finding that needs first to be confirmed and then explained. In particular, we need to know if we are selecting female applicants because during the MMI we observe better communication, critical thinking, and ethical decision-making skills, or if we rate female applicants higher because of alternative explanations that need to be further elucidated.

Acknowledgments: The authors would like to thank the administrative staff from the Medical School Admissions office for their ongoing support in the admissions process and quality improvement initiatives.

References

1. Maher BM, Hynes H, Sweeney C, et al.Medical school attrition—Beyond the statistics a ten year retrospective study. BMC Med Educ. 2013;13:13.
2. Schwartzstein RMGetting the right medical students—Nature versus nurture. N Engl J Med. 2015;372:1586–1587.
3. Frank JRThe CanMEDS 2005 Physician Competency Framework. 2005.Ottawa, Ontario, Canada: Royal College of Physicians and Surgeons of Canada;
4. Ferguson E, James D, Madeley LFactors associated with success in medical school: Systematic review of the literature. BMJ. 2002;324:952–957.
5. McLaughlin KNavigating in the fog of medical school admissions data. Adv Health Sci Educ Theory Pract. 2012;17:627–629.
6. Edwards JC, Johnson EK, Molidor JBThe interview in the admission process. Acad Med. 1990;65:167–177.
7. Edwards JC, Johnson EK, Molidor JBAdmission to medical school: International perspectives. Adv Health Sci Educ Theory Pract. 1996;1:3–16.
8. Elam CL, Johnson MMAn analysis of admission committee voting patterns. Acad Med. 1997;72(10 suppl 1):S72–S75.
9. Harasym PH, Woloschuk W, Mandin H, Brundin-Mather RReliability and validity of interviewers’ judgments of medical school candidates. Acad Med. 1996;71(1 suppl):S40–S42.
10. Eva KW, Rosenfeld J, Reiter HI, Norman GRAn admissions OSCE: The multiple mini-interview. Med Educ. 2004;38:314–326.
11. Brownell K, Lockyer J, Collin T, Lemay JFIntroduction of the multiple mini interview into the admissions process at the University of Calgary: Acceptability and feasibility. Med Teach. 2007;29:394–396.
12. Harris S, Owen CDiscerning quality: Using the multiple mini-interview in student selection for the Australian National University Medical School. Med Educ. 2007;41:234–241.
13. Dowell J, Lynch B, Till H, Kumwenda B, Husbands AThe multiple mini-interview in the U.K. context: 3 years of experience at Dundee. Med Teach. 2012;34:297–304.
14. Patterson F, Knight A, Dowell J, Nicholson S, Cousans F, Cleland JHow effective are selection methods in medical education? A systematic review. Med Educ. 2016;50:36–60.
15. Pau A, Jeevaratnam K, Chen YS, Fall AA, Khoo C, Nadarajah VDThe Multiple Mini-Interview (MMI) for student selection in health professions training—A systematic review. Med Teach. 2013;35:1027–1041.
16. Uijtdehaage S, Doyle L, Parker NEnhancing the reliability of the multiple mini-interview for selecting prospective health care leaders. Acad Med. 2011;86:1032–1039.
17. Reiter HI, Eva KW, Rosenfeld J, Norman GRMultiple mini-interviews predict clerkship and licensing examination performance. Med Educ. 2007;41:378–384.
18. Goho J, Blackman AThe effectiveness of academic admission interviews: An exploratory meta-analysis. Med Teach. 2006;28:335–340.
19. Rees EL, Hawarden AW, Dent G, Hays R, Bates J, Hassell ABEvidence regarding the utility of multiple mini-interview (MMI) for selection to undergraduate health programs: A BEME systematic review: BEME guide no. 37. Med Teach. 2016;38:443–455.
20. Berglund E, Eriksson M, Westerlund MCommunicative skills in relation to gender, birth order, childcare and socioeconomic status in 18-month-old children. Scand J Psychol. 2005;46:485–491.
21. Roter DL, Hall JA, Aoki YPhysician gender effects in medical communication: A meta-analytic review. JAMA. 2002;288:756–764.
22. Bylund CL, Makoul GEmpathic communication and gender in the physician–patient encounter. Patient Educ Couns. 2002;48:207–216.
23. Lemay JF, Lockyer JM, Collin VT, Brownell AKAssessment of non-cognitive traits through the admissions multiple mini-interview. Med Educ. 2007;41:573–579.
24. Hofmeister M, Lockyer J, Crutcher RThe multiple mini-interview for selection of international medical graduates into family medicine residency education. Med Educ. 2009;43:573–579.
25. Hofmeister M, Lockyer J, Crutcher RThe acceptability of the multiple mini interview for resident selection. Fam Med. 2008;40:734–740.
26. Hecker K, Donnon T, Fuentealba C, et al.Assessment of applicants to the veterinary curriculum using a multiple mini-interview method. J Vet Med Educ. 2009;36:166–173.
27. Garg AX, Hackam D, Tonelli MSystematic review and meta-analysis: When one study is just not enough. Clin J Am Soc Nephrol. 2008;3:253–260.
28. Egger M, Davey Smith G, Altman DGSystematic Reviews in Health Care: Meta-Analysis in Context. 2001.London, UK: BMJ Publishing Group;
29. Facione PA, Giancarlo CA, Facione NC, Gainen JThe disposition toward critical thinking. J Gen Educ. 1995;44:1–25.
30. Walsh CM, Hardy RCDispositional differences in critical thinking related to gender and academic major. J Nurs Educ. 1999;38:149–155.
31. Glover SH, Bumpus MA, Sharp GF, Munchus GAGender differences in ethical decision making. Women Manage Rev. 2002;17:217–227.
32. Cook DA, Brydges R, Ginsburg S, Hatala RA contemporary approach to validity arguments: A practical guide to Kane’s framework. Med Educ. 2015;49:560–575.
33. Singhapakdi A, Vitell SJ, Franke GRAntecedents, consequences, and mediating effects of perceived moral intensity and personal moral philosophies. J Acad Mark Sci. 1999;27:19–36.
34. Merton RKSocial Theory and Social Structure. 1968.New York, NY: Free Press;
35. Darley JM, Fazio RHExpectancy confirmation processes arising in the social interaction sequence. Am Psychol. 1980;35:867–881.
36. Rosenthal R, Rubin DBInterpersonal expectancy effects: The first 345 studies. Behav Brain Sci. 1978;1:377–415.
37. Brawley LR, Landers DM, Miller L, Kearns KFSex bias in evaluating motor performance. J Sport Psychol 1979;1:15–24.
38. Briton NJ, Hall JAGender-based expectancies and observer judgments of smiling. J Nonverbal Behav. 1995;19:49–65.
39. Nass C, Moon Y, Green NAre machines gender neutral? Gender-stereotypic responses to computers with voices. J Appl Soc Psychol. 1997;27:864–876.
40. Kulms P, Kramer NC, Gratch J, Kang SHIt’s in their eyes: A study on female and male virtual humans’ gaze. In: Proceedings of the 10th International Conference on Intelligent Virtual Agents. 2011.Berlin, Germany: Springer-Verlag;
41. Soon CS, Brass M, Heinze HJ, Haynes JDUnconscious determinants of free decisions in the human brain. Nat Neurosci. 2008;11:543–545.
42. Raymond JSexist attitudes: Most of us are biased. Nature. 2013;495:33–34.
43. Devine PGStereotypes and prejudice: Their automatic and controlled components. J Pers Soc Psychol. 1989;56:5–18.
44. Bylund CL, Makoul GEmpathic communication and gender in the physician–patient encounter. Patient Educ Couns. 2002;48:207–216.
45. Bertakis KD, Azari RPatient-centered care is associated with decreased health care utilization. J Am Board Fam Med. 2011;24:229–239.
46. Kim SS, Kaplowitz S, Johnston MVThe effects of physician empathy on patient satisfaction and compliance. Eval Health Prof. 2004;27:237–251.
47. Griffin SJ, Kinmonth AL, Veltman MW, Gillard S, Grant J, Stewart MEffect on health-related outcomes of interventions to alter the interaction between patients and practitioners: A systematic review of trials. Ann Fam Med. 2004;2:595–608.
48. Stewart MAEffective physician–patient communication and health outcomes: A review. CMAJ. 1995;152:1423–1433.
49. Association of American Medical Colleges. Total graduates by U.S. medical school, sex, and year. https://www.aamc.org/data/facts/enrollmentgraduate/148670/total-grads-by-school-gender.html. Accessed September 3, 2016.
50. Association of Faculties of Medicine of Canada. Canadian medical education statistics 2014. https://www.afmc.ca/pdf/Cmes2014-Enrolment.pdf. Accessed September 3, 2016.
Copyright © 2017 by the Association of American Medical Colleges