Fostering students’ reflection about bias in healthcare: Cognitive dissonance and the role of personal and normative standards

Abstract

Background: To reduce cognitive dissonance about one's beliefs or behavior, individuals may compare their behavior to personal and/or normative standards. The details of this reflection process are unclear.

Aims: We examined how medical students compare their behavior or beliefs to standards in discussions about implicit bias, and explored if and how different reflective pathways (preserving vs. reconciling) are associated with each standard.

Methods: Third-year students engaged in a small-group discussion about bias. Some students and group facilitators also participated in a debriefing about the experience. Using qualitative methods, the transcripts from these 11 sessions were analyzed for evidence of student comparison to a standard and of reflection pathways.

Results: Of 557 text units, 75.8% could be coded with a standard and/or a path of reflection. Students referenced personal and normative standards about equally, and preserved or reconciled existing beliefs about equally. Uses of normative standards were associated with preservation-type reflection, and uses of personal standards with reconciliation-type reflection.

Conclusions: Normative expectations of physicians are sometimes used to provoke students’ consideration of implicit biases about patients. To encourage critical reflection and reconciliation of biased beliefs or behavior, educators should frame reflective activities as a personal exercise rather than as a requirement.

Introduction (Aims)

Implicit bias is the cognition or emotion that occurs in health care when physicians unconsciously classify a patient as a member of a group and apply stereotypes about the group – whether positive or negative – to that patient. When activated, implicit bias is thought to contribute to health disparities and lead to poor quality care (Burgess et al. 2008). Reflection is increasingly used in medical school to raise awareness about and reduce the effects of bias (Mann et al. 2009; Teal et al. 2010, 2012; Thompson et al. 2010). An important trigger for such reflection is the initiation of cognitive dissonance (Aronson 1968), a social psychology concept where a person becomes aware of two conflicting cognitions or ideas. This dissonance can draw attention to discrepancies between beliefs and behavior (Stone & Cooper 2001), such as the chasm between the belief that one is unbiased, and behavioral evidence to the contrary. While it is understood that cognitive dissonance can stimulate reflection (McFalls & Cobb-Roberts 2001; Thompson et al. 2010), shift attitudes (Gringart et al. 2008), and even change behavior (Dickerson et al. 1992), the details of this process are still unclear, particularly among medical students. In previous work, we found two pathways that students take to reduce cognitive dissonance when it occurs. One pathway serves to preserve existing beliefs, while another pathway attempts to reconcile belief and behavior discrepancies (Thompson et al. 2010).

When evaluating their own beliefs and behavior, individuals often use one of two types of standards for comparison (Stone & Cooper 2001), termed personal and normative. Personal standards often relate to internal self guides (e.g., personal morals), whereas normative standards often relate to social context guides, such as the expected behavior of a “good doctor” as articulated informally (e.g., the way things are done at a particular hospital), or formally within the profession (e.g., the Hippocratic Oath). For example, comparison to personal standards could occur when an individual compares their behavior to privately held principles, and comparison to normative standards might occur when an individual compares their behavior to social mores. While there is scant literature on this cognitive process, previous research has found that individuals who have societally derived standards exhibit higher levels of prejudice (Devine et al. 1991). Additionally, there are likely multiple pathways of reflection that students might take to reduce cognitive dissonance. One pathway may preserve existing beliefs, while another pathway may attempt to reconcile belief and behavior discrepancies (Thompson et al. 2010).

In theory, whether students use personal and/or normative standards to evaluate their behavior has implications for understanding how medical students’ process dissonant information and can influence the design of learning activities intended to stimulate self-reflection. However, very little empirical data about these processes of reflection currently exists. We therefore designed this study to examine how medical students reflect on their own biases in healthcare, how personal and normative standards appear in students’ discussions about bias, and whether or how different reflective pathways are associated with such standards.

Table

Practice points

Little is known about cognitive dissonance in medical students’ reflection on implicit bias in clinical care. Prior work on dissonance and reflection informed analysis of students’ discussions of bias. Students’ comparisons of self to normative standards were associated with preservation-type reflection, and personal standards with reconciliation-type. Educators should frame bias-related reflection as a personal exercise.

Methods

Educational activity

We recruited third-year medical students from a longitudinal ambulatory care course to participate in a small discussion group titled “Best Intentions,” which explored implicit bias. Students were invited to participate by the course director and offered course credit comparable to that of other required course workshops for participating.

In preparation for the small group discussion, participants were asked to read a journal article about implicit bias among physicians (Green et al. 2007), primarily to frame the upcoming discussion and establish an understanding about the negative effects of bias. Then, to personalize implicit bias for the students, students completed two Implicit Association Tests (IAT) (Greenwald et al. 1998; Project Implicit 2011): (1) one assessing implicit bias about persons with disabilities and (2) the student's choice of one IAT focused on sexuality, race, or weight. The publicly-available IAT measures the time it takes participants to match pictures or words that represent a social group (race, disability) with value-laden characteristics (happy, angry). This test operationalizes individual bias by assuming that the less time taken to match words, the more likely an individual originally held that association. The IATs were intended to trigger critical reflection in each student regarding his or her potential unknown biases.

After taking the IATs, we asked them to reflect in a one-hour faculty-facilitated group discussion with trained facilitators who utilized a guide developed to lead discussion of students’ experiences with implicit bias. The discussion began with students’ experiences with the IAT itself. Then, facilitators shifted discussion toward students’ clinical experiences (providing services or observing others doing so) in which bias towards patients seemed evident, with focus on how implicit bias might have impacted care and what might have been done to mitigate the bias. The activity strived to encourage students to recognize that implicit bias is inherent in human decisions, and, in particular, medical decision-making.

Data collection

In conjunction with approval by Baylor College of Medicine's Institutional Review Board, we audio-recorded and transcribed the nine small-group discussions for transcription. We also invited randomly selected students to attend a follow-up focus group to discuss their impressions of the reflective activity and faculty to attend a faculty de-briefing after the student discussions. These two debriefing sessions were also audio-recorded and transcribed, resulting in 11 transcripts (designated as groups A-K for this manuscript). With each transcript, text units for coding were identified and consisted of complete statements made by participants.

Analysis

As shown in Table 1, we developed a thematic scheme to code the group transcripts. Two codes for reflective pathways – “reconciliation” and “preservation” – were defined according to Thompson et al.'s (2010) definitions and appear at the top of the table. Statements of preservation acted to maintain an individual's original perspective on bias and avoid critically engaging in discussion of the existence and effect of bias in a clinical setting. In contrast, reconciliatory statements suggested a willingness to adjust behavior or attitudes, or explore the existence of bias on and individual and clinical level. The two codes for the use of referenced standards in assessing one's behavior or belief – “personal” and “normative” – were defined according to literature (Stone & Cooper 2001), and appear on the left in the table. In the context of our discussion groups, for example, use of personal standards could be signaled by a statement such as, “This behavior doesn’t match up with who I think I should be” and normative standards might be signaled by a statement such as, “This behavior doesn’t match up with who society thinks I should be.” The code book permitted analysts to assign either a standard or a reflection pathway or both, based upon students’ statements. The code “No code identified” was used for statements by participants that did not clearly reference a standard and/or a path of reflection.

Table 1  Distribution of coded units within thematic coding scheme

		Reflective pathway			Total
		Preservation: the student experiences no reflection, no development, and attains original position.	Reconciliation: the student experiences progression, development, reflection, or attitude change.	No reflective pathway coded
Standard of reference	Personal standards. A reference to one's own identity or personal standards for conduct, evaluating behavior in terms of its fit to their own self-expectations.	36	106	24	166
	Normative standards. Student evaluates behavior by comparing to external sources of standards.	92	51	40	183
	No standard coded	35	38	135	208
Total		163	195	199	557

Coders could assign a “standard” code and/or a path of reflection code, with up to two codes per text unit. One investigator (RH) coded the units in all the transcripts. We assessed reliability by having each co-author analyze one transcript and compared these to the primary analyst-coded transcript. To further understand the nature of the emergent personal and normative standards, two investigators (RH and CT) used the constant comparative method (Boeije 2002) to identify emergent sub-themes within units coded as using a personal standard and within units coded as using a normative standard. Disagreements were resolved through discussion.

Using the final applied codes for each text unit, we utilized descriptive statistics (frequency, percent) to describe patterns in coding. We used the chi-square test to examine associations between use of comparative standards and reflective pathways.

Results

Participants

On average, the 72 participating students were 26.1 years old. The majority (37, 51.4%) was White; Asians represented 34.7% (25), Latinos 9.7% (7), and Blacks 4.2% (3). Almost two-thirds (44, 61.1%) were male. These students represented 43.9% of their 164-member class, and their demographics were consistent with the medical school's student population and the class from which they were recruited. Ten facilitators (one for each group) led the group sessions of six to eight students. Facilitators were predominantly physicians (seven of 10) and White (six of 10), and half were female.

Coding

Across the eleven transcripts of small-group discussion, follow-up student focus group, and faculty de-briefing, we identified 557 text units for coding. Reliability in analysis was established. Each co-author's analysis of a separate transcript resulted in agreement in coding for 74/83 units (89.2%), 46/65 units (70.8%), and 81/102 units (79.4%) on three separate transcripts.

Frequency of reflection paths and standards

As shown in Table 1, of the 557 units, 422 (75.8%) referenced sufficient information to code a standard for comparison and/or a reflective pathway. Across all group discussions, students referenced both types of standards and both reflective pathways about equally. Of the 557 text units, 64.3% (358) could be coded with a reflective pathway. Of these 358, 45.5% (163) were coded as reflecting a preservation path, while 54.5% (195) were coded as reflecting a reconciliation path. Similarly, 62.6% of the text units (349) were coded as referencing a standard. Of these 349, 47.6% (166) referenced a personal standard, and 52.4% (183) referenced a normative standard.

Of the 183 units coded with the normative standard, 69.9% (128 total) were coded with one of four dominant sub-themes. These included (a) conflation of implicit bias and “evidence-based practice,” (13.1% of 183); (b) students’ stereotypes influenced by experiences with patients and populations (16.3%); (c) assertions of test validity having bearing on perceptions of existence and pervasiveness of bias (24.0%); and (d) student perceptions that society stigmatizes bias (19.1%). Of 166 units coded as referencing personal standards, 73.5% of units fell under one of the three dominant sub-themes. In these cases, a participant (a) drew on perceptions of self for a standard of comparison (28.9% of 166), (b) named personal reflection as a strategy to manage bias (33.1%), and (c) argued that individual bias exists in the healthcare relationships (as opposed to in decision-making about diagnosis or treatment) (11.4%). A complete list of sub-themes for the standards can be found in Table 2.

Table 2  Thematic coding scheme with sub-themes

Normative standards
a.	Clinical models of bias
b.	Doubts of test validity
c.	Evidence confirming the existence of bias (literature/data)
d.	Evidence-based practice is helpful in a clinical setting
e.	Experience with patients and populations as related to bias
f.	External educational requirements
g.	Social stigma of bias, or “good doc ideal”
Personal Standards
a.	Action as a personal strategy to manage bias
b.	Bias exists in the relational points of care
c.	Experience with specific biased events
d.	Personal reflection as a strategy to manage bias
e.	Self-knowledge as a standard of comparison

Note: Bold, Italic = dominant themes.

Association of reflective paths with standards

We were primarily interested in the interplay of the reflective pathways with the two types of standards. As shown in Table 1, comparison to a normative standard was more frequently associated with a preservation path of reflection, while comparison to a personal standard was more frequently associated with a reconciliation path of reflection (χ² (4) = 181.1, p < 0.0001). In the following sections, we present illustrations of preservation and reconciliation as they interact with various sub-themes of normative and personal standards.

Normative standards and paths of reflection

Of the 183 text units in which a student applied a normative standard, 50.3% followed a pathway of reflection that we classified as preservation. An additional 27.9% followed a pathway of reconciliation. The remaining 21.9% could not be coded with a reflection path.

Among the sub-themes of normative standards, students cited “use of evidence-based practice” and “evaluation of patient risk” as a potentially positive use of bias. In these cases, some conflated implicit bias and “evidence-based practice.” One participant in Group A drew on epidemiology in their understanding of bias and maintained that the use of race in diagnosis is acceptable and useful.

I guess epidemiology is pretty important. You want to know what your patient is at risk for, and statistics show that … certain races are at risk for different things … you obviously treat people differently based on that.

A participant in Group L implied that evidence-based medicine and bias are taught institutionally.

You’re just trained to think that way because that's going to make you a better physician … bias almost helps me in that situation … [Medical school] taught us how to do that.”

Like these, several participants conflated evidence-based medicine and epidemiology with bias in clinical settings. For most (though not all), the use of this normative standard was associated with a preservation reflective pathway, in which students maintained existing beliefs about bias.

In the second normative standard sub-theme, some participants discussed the use of stereotypes and how these are confirmed or disconfirmed by the nature and degree of experiences with patients. Stereotypes are socially constructed assumptions about a group. In these cases, the societal stereotype is the normative fulcrum by which individuals evaluate members of the stereotyped group. In this case, the normative standard is the stereotype itself, used as a comparison against which to judge their personal experiences. These experiences can create or confirm original biased perceptions or disconfirm and transform biased perceptions, and in the absence of experience, stereotypes may remain. For example, a student in Group C who believed that those in private care settings were less likely to use drugs had his beliefs reinforced by encountering more drug users at the public hospital.

I mean it is logical. It's not like we’re saying, you know, these people at [a private hospital], … aren’t using drugs. It's just [the patients] more often are at [a public hospital] …. That ultimately is doing a disservice to that patient, but it's just the ‘nature of the beast’.

In contrast, a participant in Group I reflected on recognizing that a stereotype about members of his own racial/ethnic group (i.e., we work too hard to ever be homeless) may be inaccurate.

I can look at situations where I sort of hit myself on the head and been like, “Why did I even think that?” and the one example that comes to mind was being at [the public hospital], … for a lot of us from immigrant background, our parents came here,… and had to work really hard for everything, … I can remember back when I was in Medicine the first time I saw a homeless Indian patient, and honestly, I had to stand there and it really took me by surprise for a little bit, … I sort of maybe subconsciously assumed stereotypes of populations are associated with certain backgrounds.

For some students, such as this student in Group H, stereotypes and bias were created as a result of their experiences.

For me, I think working at [a public hospital] created some new biases for me. I didn’t have a lot of exposure to Hispanics before I came [here]. I kind of noticed myself thinking, “Oh, there's a Hispanic woman in the elevator. She must have three kids and she's like 26 years old and she must have had tubal ligation by now” … I got educated working at [a public hospital], but I also got kind of a new bias working.

This reflection exemplifies the assertion that, while exposure to populations and people can be valuable, some experiences can create harmful bias. This participant, who originally believed they were unbiased, followed a reconciliation pathway to considering and acknowledging the existence and development of bias. Though this participant followed a reconciliation path, most participants who spoke about experiences with patients and bias took a path of preservation (confirming existing stereotypes) more often than reconciliation.

In a third sub-theme, some participants expressed doubt regarding the meaning of their results, citing issues with test validity. The IAT test is a normative standard, an external gauge used to measure bias in participants. Some participants argued that their results were invalid because they perceived the test as “not adhering to proper psychometric standards.” Doubts about the validity of the test were at times connected to their beliefs about the existence and pervasiveness of bias. Some participants shared that they were unmoved by the IAT activity. One participant in Group L expressed doubt about the influence of the test.

That [IAT] test definitely did not convince me to change the way I interact with people. I mean maybe if the test had convinced me better of my biases then I would take a step back and really try to change the way I interact with people, but since the test didn’t convince me, I’m just kind of like ignoring it. If the test had been more effective at getting to me, maybe [I would change].

As a result of the perceived validity of the IAT, this participant exhibited a resistance to change attitudes, beliefs, or behavior, signaling an effort to preserve their existing state through doubts about test results. One participant used concerns about the test validity to preserve a bias, “I don’t believe in this test strongly enough to actually take those results and apply them to my life and change my life in any way.” In contrast, one participant from Group A believed that their results were valid because the IAT fulfilled their expectations of a “good test,”

I also did the first [test] twice [and it] gave me the same exact result, so it kind of made me believe that it was probably accurate. … I did it twice and it told me the exact same thing with almost the exact same percentages, I was like oh, maybe that's true.

In these cases, it is clear that the students were evaluating the test, and that the test's perceived merits may influence how they think about bias, including whether they take a reconciliation or preservation pathway of reflection. Participants who spoke about test validity overwhelmingly took a path of preservation.

In the final dominant sub-theme among normative standards, some participants spoke directly about the social stigma of having bias, especially as physicians. One participant in Group E shared their beliefs about how bias influences patient care.

I don’t think [bias] would affect how I would treat [patients]. [This test] implied that if you [had a bias] that you were not being the ‘ideal doctor’ that wants to help the world and work for no money and be totally magnificent and wonderful, … at 3:00 in the morning when a dirty, nasty homeless man comes in off the streets, maybe you won't be as excited about treating him as you would about somebody else. … I think [society] kind of expect[s] doctors to be like a little bit too wonderful sometimes.

This participant directly referenced societal pressures placed on doctors to fit an ideal, and at once preserved the idea that bias meaningfully enters the clinical realm. One participant in Group C found comfort in a group discussion about personal bias.

It's nice to know that even if you have these [biased] thoughts, that you’re not the only one. Because sometimes in medicine you’re supposed to be this deacon, and [should] not have these terrible thoughts roaming around in your head … and it's nice too know that you are human, and you are experiencing these emotions, and it's OK.

The act of sharing about personal bias helped to manage some of the stigma this participant experiences in regards to bias, taking steps toward reconciliation. Another participant found comfort in surprising results, “It's probably better that I don’t have a preference against one race versus another…and that means I'm maybe less of a bad person than I thought I was.” Participants who felt that society negatively judges individuals who exhibit bias chose preservation and reconciliation reflection pathways about equally.

Personal standards and paths of reflection

Of the text units in which a student applied a personal standard, 21.7% followed a preservation pathway of reflection. In contrast, 63.9% followed a pathway of reconciliation. The remaining 14.5% could not be coded with a reflection path. In the first of three dominant sub-themes among personal standards, participants compared their results to their knowledge of their past behavior and identity or alternately, to perceptions of themselves as trying to attain a personal ‘ideal.” In response to a revelation by another student that they “dread seeing patients with Hispanic surnames because they won’t speak English,” one student from Group G shared a personal experience that shaped how he thinks about and approaches patients with language needs.

Part of my application to medical school was a story about my grandfather … He's the person I probably most respect in this world, and he doesn’t speak English, and I remember going with him as a kid to the doctor …. it's one of my driving factors in medicine, and I’m always willing to interpret [for patients]. I'm always willing to stay behind. I never complain … because I know these people deserve that, and these people probably have grandchildren in medical school or who are doctors, so I pray that people really treat these people the best they can.

This individual acknowledged that bias does have an effect on medical care, and spoke about the motivation behind a personal ideal. A different participant reflected on their results on the IAT, and how it relates to self-knowledge. This participant in Group D believed the IAT test results ring true based on what they expected of themselves, and their memory of personal experiences.

[My results] were expected. I took the disability test and it showed that I had a moderate preference for … non-disabled people … I’ve always preferred being non-disabled. … I have a very deep fear of grave personal injury [during my rotation at a local rehabilitation hospital] … I wasn’t particularly surprised [by my results], but it was sobering just to see it on paper … 

In this case, the student exhibited signs of reconciliation by acknowledging a personal bias, and recognizing the gravity of this revelation. Students who spoke about personal knowledge took a path of reconciliation more often than preservation.

In the second dominant sub-theme among personal standards, some participants planned to hold themselves accountable regarding potential bias and named personal reflection as a strategy to do so. In these cases, students spoke about a personal standard of awareness about bias to avoid its negative effects. One participant in Group G considered how race influences perception of patients, reconciling that it is important to recognize troublesome biases to provide good care.

Race definitely plays a part in how you form the first impression of a patient and how much you feel like you can … build a relationship or a rapport based on what you may have in common … You may not feel as comfortable building a relationship … so I think those are factors that we should always consider in every patient encounter that we have.

These students understood how reflection can be used as a strategy in-the-moment to avoid the negative effects of bias.

One participant in Group G emphasized the importance of self-evaluation in long-term professional development, asserting that the IAT activity is the first step in an on-going process of critical reflection.

It's also just very important the process of self-evaluation, because … we get these lectures now, but when we’re residents and we’re attendings, we’re not gonna have someone once a semester telling us this. So I think [the goal is to] take it upon ourselves at least [to ask] “How have I changed in my ability to relate to people of all different sizes and colors,” and, “Is there some way I'm failing a certain group of patients.” I think we have to be critical of ourselves … when we become residents and doctors.

Another participant, reconciling the complex nature of bias, stated, “I don’t know if you can change the bias but you can overcome it by being aware of it.” A participant in Group K reflected on moments where they experience bias in a clinical setting that confirms the existence and prevalence of bias, while giving a strategy to manage that bias.

Sometimes during the day, you have these few seconds or a minute where you feel kind of disconnected from everybody and you feel like you’re observing yourself. You might make a certain kind of facial expression, and not be aware of it, but if you’re going into the encounter thinking, “I’m gonna be really conscious about how I say things and how I react to patients,” then maybe you will be surprised by your reactions to things and adjust …

This student believed that consistent reflection on individual action could act to manage bias. This participant critically reflected on the gravity of bias in a clinical and interpersonal setting. The participants who spoke about reflection about bias overwhelmingly took a path of reconciliation.

As the final dominant sub-theme among personal standards, some participants believed that bias more often occurs in the relationship-building aspects of a patient encounter than occurs in actual decision-making about diagnosis or treatment. In these cases, participants believed that bias could be avoided in treatment through closely following protocol. One participant from Group K believed that bias only occurs in relational interactions, and as such, can be separated from medical care itself.

I think my experience has been that generally we tend to perform the same services as required for all patients regardless of bias, or at least it feels that way to me … Diagnosis is made and the treatment plan is pretty similar … where I see [bias] come out is how people think of a patient, how much time they’re willing to spend with the patient, and how they interact with a patient – so more in the subjective points of care than objective sort of protocol determined.

While this person critically reflected on the role of bias in care, they did not believe that bias influences medical decision-making. A participant expressed disbelief about the translation of personal bias to a clinical setting.

I mean, fat or skinny, you’re gonna treat [a patient] with the same treatment if they have a cold. I don’t see how [bias] would translate to treating them differently. Maybe your interpersonal skills or bedside manner might change a little if you have a preference, but the medical outcome or the medical treatment will still be the same. … I mean, you might think a fat person is not gonna be compliant because he doesn’t take care of himself and so I’m gonna treat him differently …

This participant acknowledged that bias can influence behavior, but preserves the notion that bias does not affect clinical outcomes. Participants who spoke about bias in the relational points of care took a path of preservation and reconciliation equally.

Discussion

While directionality and causality cannot be established using this data set, these data were sufficient to explore how students naturally associate a particular standard with a path of reflection when considering and discussing implicit bias. When working to preserve original attitudes and believes about bias, students tended to draw on normative standards, and when working to reconcile discrepancies and critically reflect on bias, students more often drew on personal standards.

Many medical school activities are normative in nature. For example, many professionalism curricula are based on externally articulated position statements, such as the American College of Physicians Charter or the Hippocratic Oath (Goldberg 2008). While these statements help to define and guide the practice of medicine, our data suggests that these kinds of statement may not be effective to encourage meaningful reflection about bias. In contrast, these data suggest that educators addressing implicit bias should actively frame reflective activities as a personal exercise, outside of the normative sphere of medical school. Activities that are grounded in normative expectations such as “good doctors are aware of and manage their bias” may require considerable reframing to activate comparison to personal standards instead. For example, facilitators might consider framing an activity as an exercise in personal development, in contrast to meeting industry standards. Further, activities that rely on the IAT alone to trigger reflection may be less successful than the use of the IAT as a trigger for group-based discussion. The IAT itself is inherently normative; results are provided in comparison to others. Consistent with published literature (Kember et al. 1996; Lockyer et al. 2004), group discussion often served as a secondary trigger for continued or additional reflection, through exposure to a diversity of beliefs and observations. When those beliefs or observations were offered in conjunction with comparison to personal standards, other students were sometimes motivated to consider their own personal standards. In this way, group discussion offered a second chance for reconciliation for those who would have maintained perceptions if unchallenged.

Because personal or normative standards are associated with different paths of reflection, educators can adjust reflective activities to encourage students to compare themselves to a type of standard that produces a particular path of reflection. These data raise questions as to the normative or personal nature of critical reflection about other realms in medicine in addition to bias. In learning areas, where students’ perceptions of themselves are often challenged by educational activities, such as professionalism (Arnold 2002; Varga-Atkins et al. 2010), cultural competence (Lie et al. 2010; Teal et al. 2010), and communication with patients and colleagues (Blanch-Hartigan 2011), reconciliation of attitudes, beliefs, and actions becomes a goal for educators. In these cases, it may be desirable to frame activities as related to personal standards.

Our study is not without limitations. To protect participant's privacy, the audio tapes of the focus groups used for analysis were destroyed immediately after transcription. Analysis may have benefitted from audio or video recordings of the discussions to better glean non-verbal cues. The a priori code book based in existing literature may have limited interpretation of the data. It is possible that our own framing of the activities, which attempted to “normalize” the occurrence of bias may have prompted normative standards. However, the standards and cognitive pathways appeared strongly in the data. Educational research is needed to explore the results of our study and to test interventions with alternate framing to determine student outcomes. Despite these limitations, our data offers medical educators much-needed suggestions for how they might create meaningful reflection about implicit bias, and perhaps other equally challenging educational arenas.

Declarations of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the article.