A distressing feature in the life of which you are about to enter … is the uncertainty which pertains not alone to our science and art, but also to the very hopes and fears which make us men. In seeking out the absolute Truth we aim at the unattainable, and must be content with finding broken portions.
—Sir William Osler1
A substantial body of literature over the past several decades suggests that tolerance for ambiguity is an important competency for physicians.2–4 As Geller4 recently pointed out,
in medical practice, an individual’s low tolerance for ambiguity has been associated with … increased test-ordering tendencies and failure to comply with evidence-based guidelines,5 greater likelihood of recalling screening mammograms,6 increases in patient charges,7 withholding negative genetic test results,3 fear of malpractice litigation and defensive practice,8 and discomfort in the context of death and grief.9,10
Tolerance for ambiguity influences practice patterns and resource utilization11; it is estimated that 17% of excessive costs in medical care result from physicians’ anxiety related to how they manage uncertainty.7 Intolerance of uncertainty has also been linked to higher rates of burnout.12
In addition to negative factors related to low tolerance for ambiguity, there also may be specific positive factors related to having a high tolerance for ambiguity, in addition to minimizing the negative factors just mentioned. For example, our increasingly complex health care system requires physician leadership, and higher tolerance for ambiguity is correlated with leadership capacity among medical students.13 Also, Geller4 suggests that there can be many benefits if schools more intentionally admit students who possess high tolerance for ambiguity. These benefits include improving the “quality of care in ambiguous conditions,” reducing “imbalances in physician supply and practice patterns,” and enhancing “the humility necessary for moral character formation.”
As academic medicine moves toward a competency-based model of medical education, the relevance of tolerance for ambiguity has been increasingly recognized in both undergraduate and graduate medical education. Educators and scholars have called for ambiguity tolerance to be assessed and evaluated both when students are admitted to medical school4,14 and also as medical students transition through the educational continuum.4,15,16 The Association of American Medical Colleges (AAMC) includes tolerance of and adaptation to stressful or changing environments as part of the Resiliency and Adaptability competency, which is one of the core competencies for entering medical students.17 It also includes comfort with ambiguity in its published Core Entrustable Professional Activities,18 believed essential for graduated students entering residency programs. The Accreditation Council for Graduate Medical Education (ACGME) considers tolerance for ambiguity as an essential “reporting milestone” under the Professionalism competency for pediatrics residents19; programs assess each pediatrics resident’s progress on this milestone every six months and report it to the ACGME.
Although the conceptual literature on ambiguity tolerance is substantial and rich, empirical studies that focus on medical students are sparse and have been limited to small samples of students, often at just one medical school. The more consistent findings among these studies indicate that tolerance for ambiguity is associated with sociodemographic characteristics, specifically sex and age, as well as a student’s attitude regarding underserved populations.16,20,21 Notably, no studies that we know of followed students longitudinally to examine stability of tolerance for ambiguity over time; a cross-sectional study that assessed students enrolled in medical school in Germany20 found below-average tolerance for ambiguity, with no differences across the students enrolled in the different years of school during that academic year.
Empirical studies have produced mixed findings regarding whether tolerance for ambiguity influences specialty preference.4 Geller et al21 found tolerance for ambiguity to be lower among students who choose highly structured specialties, such as surgery, and higher among students who choose specialties that are inherently ambiguous, such as pediatrics and psychiatry. DeForge and Sobal,22,23 on the other hand, found little evidence of an association between tolerance for ambiguity and specialty choice.
In their 1993 study, Geller et al3 developed a modified scale for tolerance for ambiguity. The authors quote Budner’s24 definition of intolerance for ambiguity: “the tendency to perceive situations that are novel, complex or insoluble, as sources of threat.” They combined 18 survey items from various measures developed in prior research and included them in a national survey of physicians’ knowledge and attitudes about genetic testing. Although tested on a population of physicians, the scale was an attempt “to measure ambiguity tolerance as a more general personality trait.”3 Psychometric analyses found that 7 of the 18 items were a “good fit” for the data.
Purpose and hypotheses
In 2013, the modified tolerance for ambiguity (TFA) scale described above was, for the first time, included in the AAMC Matriculating Student Questionnaire (MSQ).25 Our study, reported below, used data from this large, nationally representative sample of over 13,000 respondents to systematically examine how entering students’ characteristics and preferences are associated with tolerance for ambiguity.
We first examined the reliability of the TFA scale. We then examined the extent to which tolerance for ambiguity differs by respondents’ sociodemographic characteristics (sex and age), specialty choice, career intentions, and levels of perceived stress. We began our study with the following three hypotheses:
- Women will have higher tolerance for ambiguity than men.
- Older students will have higher tolerance for ambiguity than younger students.
- Students who express an intention to work in an underserved area will have higher tolerance for ambiguity than students who express no such intention.
In addition, previous research21 suggests that tolerance for ambiguity may differ across students’ specialty preferences. However, because no previous studies have examined as wide a range of specialty preferences as we did in the present study, we did not hypothesize specific results. Instead, we performed exploratory analyses in order to develop hypotheses for future work.
Participants and procedures
The MSQ contains a wide range of survey items, including premedical experiences, the medical school selection process, personal characteristics and attitudes, and specialty preferences and career plans. In 2013, individuals accepted for admission to any of the 140 U.S. medical schools that were accredited at that time by the Liaison Committee on Medical Education and that enrolled students in 2013 were invited by the AAMC to participate in the MSQ between June and mid-September. A total of 20,792 individuals were invited to participate via e-mail in the MSQ online survey. Of the 20,055 individuals who actually matriculated to medical school in 2013, 14,888 consented to participate in the MSQ, representing a 74.2% survey participation rate. As explained in the next paragraph, the number of participants whose responses we studied was slightly lower.
Our study was deemed exempt from institutional review board review by the human subjects office at the AAMC.
The seven-item TFA3 scale is a measure of one’s ability to cope with situations of uncertainty. Participants were provided the following instruction: “Please indicate the extent to which you agree with the following statements”; seven statements were then given, such as “Before any important task, I must know how long it will take.” The six response options were strongly agree, moderately agree, slightly agree, slightly disagree, moderately disagree, and strongly disagree; these options are scored from 1 to 6, with strongly agree = 1 and strongly disagree = 6. TFA scores are calculated by summing across the seven statements. The possible range of scores is 7 to 42 (7, when a student chooses “strongly agree” for all seven statements, and 42, when a student chooses “strongly disagree” for all seven statements—see Table 1). Thus, higher scores are correlated with higher tolerance for ambiguity. Only survey participants who provided responses to every item on the TFA scale are included in the present analyses. In total, 13,867 medical students responded to all items on the TFA scale, representing 69.1% of all 2013 matriculating students.
The Perceived Stress Scale (PSS)26 is widely used for measuring the perception of stress. This 10-item scale measures the degree to which situations in one’s life are considered stressful. The PSS was included in the 2013 MSQ. Participants were provided the following instruction: “The following questions ask you about your feelings and thoughts during the last month. In each case, indicate how often you felt or thought a certain way.” Response options were never, almost never, sometimes, fairly often, and very often. PSS scores are calculated by summing across the 10 items, which are measured on a 0- to 4-point scale (never = 0, very often = 4). The possible range of scores is 0 to 40, and higher scores are correlated with higher perceived levels of stress. Only the survey participants who provided responses to every item on the PSS are included in our analyses of PSS responses: 13,314 medical students responded to all items on the PSS.
The MSQ assessed specialty preference with the following question: “What general specialty are you considering?” Respondents could select 1 of 26 specialties or a response option of “undecided.” In total, 13,635 medical students selected one of the response options, representing 68.0% of all 2013 matriculating students. The survey also included the following item: “Do you plan to work primarily in an underserved area?” Response options were yes, no, and undecided. In total, 13,612 medical students responded to one of these options.
The sex and age of each respondent were obtained from the AAMC Student Records System, which houses centralized information on the U.S. medical school student population. Respondents’ ages were grouped into four categories: 18–22; 23–25; 26–30; and 31–54.
We first examined the internal consistency of the TFA and PSS scales using the Cronbach alpha test of reliability.27 Analyses of variance were then conducted to examine the relationship of the TFA scores with sex, age, plans to work in an underserved area, stress, and specialty preference. We also report effect size results (Cohen d) in order to discern the extent to which the magnitude of the difference between two means is practically significant.28 Generally, an effect size is small if d is equal to or greater than 0.2 but less than 0.5, medium if d is equal to or greater than 0.5 but less than 0.8, and strong if d is equal to or greater than 0.8.
Analyses of reliability
The Cronbach alpha test of reliability for the TFA and PSS scales showed values of 0.749 and 0.857, respectively. These tests suggest that, for each scale, the items have high internal consistency; that is, the items consistently measure the underlying concept. Results from a confirmatory factor analysis to further examine the dimensionality and fit for the two scales showed that a one-factor model was a reasonable fit for the data for both scales (results available from us upon request).
Descriptive statistics and analysis
Table 2 presents the number of MSQ respondents for each of the categories described earlier—that is, sex, age, plans to work in an underserved area, and perceived levels of stress. Additionally, the standard deviation and mean TFA score associated with each item are displayed. Finally, the F statistic for each analysis of variance and effect size (Cohen d) values are presented.
As shown in Table 2, the mean tolerance for ambiguity score differed significantly between women (24.8) and men (25.2; F = 12.5, P < .001), although not in the hypothesized direction. However, the effect size (d = 0.06) was negligible. As projected, the results show significant differences in mean TFA scores for younger and older MSQ participants (F = 102.5, P < .001). MSQ respondents aged 18 to 22 years had an average TFA score of 24.1, and MSQ respondents aged 23 to 25 years had an average TFA score of 24.9. Those averages increased to 26.5 and 27.1 for age groups 26 to 30 years and 31 to 54 years, respectively. The difference in the average TFA score between the youngest (18–22 years) and oldest (31–54 years) age groups showed a medium effect size (d = 0.52).
MSQ respondents who indicated an intention to work in an underserved area had a higher mean TFA score (25.8) than did those who indicated no such intention (24.1) and those who were undecided regarding future plans to work in an underserved area (25.0; F = 44.5, P < .001). The difference in the average TFA score between those who indicated no intention to work in an underserved area and those who indicated an intention to work in an underserved area showed a small effect size (d = 0.29).
On the scale ranging from 0 to 40, the mean PSS score was 12.5 and the standard deviation was 5.5. Analyses showed a relationship between TFA scores and perceived levels of stress among MSQ participants (F = 208.8, P < .001), with those expressing higher levels of stress reporting lower tolerance for ambiguity. Here, the difference in the average TFA score between the first quartile group (scores between 0 and 8) and fourth quartile group (scores between 15 and 40) showed a medium effect size (d = 0.66).
Table 3 displays descriptive statistics for TFA scores for all specialty preferences, listed in ascending order of TFA scores. There was an overall significant relation between specialty preference and TFA score (F = 2.2, P < .001). However, because we had no specific hypotheses about this relationship, follow-up tests were not conducted. Specialties with the lowest mean TFA scores were dermatology (24.0), physical medicine and rehabilitation (24.2), otolaryngology (24.4), and anesthesiology (24.5). Furthermore, interest in specialties such as psychiatry (25.3), radiation oncology (25.3), emergency medicine (25.5), neurological surgery (26.1), and medical genetics (26.4) were associated with the highest mean TFA scores. It should be noted that even the greatest difference in the average TFA score between specialties showed only a small effect size—for instance, Cohen d for the difference between the average TFA score for dermatology and the average TFA score for neurological surgery was 0.34.
In the national sample of matriculating medical students whose TFA scores we studied, tolerance for ambiguity was higher in men and in older students. Additionally, high tolerance for ambiguity was associated with an intention to work in an underserved area and with lower levels of perceived stress.
In contrast to previous research, the present study did not show higher tolerance for ambiguity in women but, rather, a small advantage for men. The narrowing of any difference between men and women may be due to the generational influences of the “Millennial generation.” Ninety-eight percent of the 13,867 survey participants who responded to the TFA items are in the Millennial age group of those born between 1980 and 1999. Some research suggests that gender distinctions and gender roles may be diminishing among those in the Millennial cohort.29
Similar to findings of previous studies, age was related to tolerance for ambiguity. In earlier studies, however, tolerance for ambiguity was related not to the individual’s age at the time he or she participated in the studies but, instead, to the age at which the individual started medical school.20 In the present study, students’ ages at matriculation and when participating in the study were the same. Whether it is age alone or particular experiences in which older students participated during their “gap” between college and medical school that led to this difference in tolerance for ambiguity is not known.
Respondents who expressed an interest in working in an underserved area had higher tolerance for ambiguity than those who did not express such an interest. With disparities of health care access and outcomes recognized by socioeconomic status, race and ethnicity, and geography, there will continue to be a need for clinicians dedicated to patients living in underserved areas.30,31 Therefore, medical schools committed to addressing these problems may consider prioritizing personal characteristics such as tolerance for ambiguity in the admission process to enhance the likelihood that the future health care workforce will better address disparities in health care access. Given the small effect size found for this association, it may be prudent not to consider this as prescriptive for medical schools but, rather, that admission committees may wish to consider tolerance for ambiguity along with other personal factors in the context of a holistic review. We plan future research to examine how the relationships between tolerance for ambiguity and interest in working in an underserved area change over the course of medical school.
Although the present study did not make predictions about which specialty preferences would be associated with different levels of tolerance for ambiguity, the exploratory analysis was consistent with some prior work in this area. For example, we found that the specialty preferences associated with the highest levels of tolerance for ambiguity included medical genetics, emergency medicine, and psychiatry; these areas have all been proposed by other researchers9,21,32 as disciplines requiring a high tolerance for ambiguity. It should be noted, however, that the range of scores across the specialties was very small, which is also consistent with a recent study using a different tolerance for ambiguity scale.33 Therefore, these relationships will require further study.
Perceived stress was also associated with tolerance for ambiguity, as students with lower tolerance for ambiguity reported higher perceived stress levels. High levels of stress and psychological distress commonly have been found among medical students, with medical students showing higher rates of depression,34 anxiety,35 and suicidal ideation36 than their peers. Students with low tolerance for ambiguity may be at particular risk.
Since at least 1962, those with intolerance for ambiguity have been recognized as perceiving situations that are new and complex as “sources of threat” and experiencing significant stress from those situations.24 Because students completed this survey either just before entering medical school or during the first weeks of medical school, those with lower tolerance for ambiguity may have experienced stress from the novelty of the transition to medical school or uncertainty of the immediate future. It is unknown how this type of stress correlates with the risk of burnout. With over half of medical students meeting criteria for burnout and the known associations between burnout, lower empathy, lower interest in caring for the underserved, and increased unprofessional behavior,37 it will be critical to determine whether students with lower tolerance for ambiguity are at an increased risk for burnout. Most interventions to address medical student distress have focused on access to mental health services and wellness programs38; cognitive, behavioral, and mindfulness-based strategies39; and, more recently, curricular changes.40 It is unknown whether these interventions are equally efficacious for students with different levels of tolerance for ambiguity or, relatedly, whether such interventions could increase an individual’s tolerance for ambiguity over time.
The present study is limited by its reliance on self-report to assess important personal characteristics. Furthermore, because of this study’s very large sample size, there were many areas of statistical significance but low to moderate effect sizes. Further research is necessary to determine what score differences represent meaningful qualitative distinctions. In addition, the tolerance for ambiguity measure used in this study was designed by its authors to examine tolerance for ambiguity as a personality trait. Other researchers have suggested that tolerance for ambiguity should also be viewed as an attitudinal style that may be more changeable with experience.41 Whether the measure used here can detect changes over time will need to be tested.
No study has been published demonstrating whether ambiguity tolerance among medical students changes over time. Given its implications for levels of stress, specialty preference, and intention to care for underserved populations, such a study should be an important item on educational and health services research agendas. Because the AAMC will administer surveys to the cohort of medical students included in this study during their second year and again just prior to graduation, a longitudinal examination will be possible, and will be important. Although medical students enter medical school with a certain predisposition to respond to ambiguity, it has been proposed that the “medical socialization process is likely to exert a mediating influence on how graduating students deal with ambiguity.”5 Alternatively, it is possible that tolerance for ambiguity is more like a personality trait (i.e., relatively stable over time and unchanging in the face of new experiences). Either finding will have a critically important, if different, implication for the formation of our nation’s future physicians.
If tolerance for ambiguity can be taught or strengthened through the learning environment, medical schools will need to develop evidence-based curricular and pedagogical approaches to nurture this quality in their students. If we find that TFA is relatively immutable and a “trait,” schools should become more strategic in calibrating the appropriate level of tolerance for ambiguity in their entering students. Schools may incorporate this trait’s assessment more intentionally into their admission processes by including a validated tool into the application materials or by incorporating interview methods, such as multiple mini-interviews. Alternative interview methods may be better than traditional interviews in identifying important noncognitive traits such as tolerance for ambiguity.42
1. Osler WReynolds R, Stone J. Aequantimitas. In: On Doctoring: Stories, Poems Essays. 1995 New York, NY Simon and Schuster
2. Atkinson P. Training for certainty. Soc Sci Med. 1984;19:949–956
3. Geller G, Tambor ES, Chase GA, Holtzman NA. Measuring physicians’ tolerance for ambiguity and its relationship to their reported practices regarding genetic testing. Med Care. 1993;31:989–1001
4. Geller G. Tolerance for ambiguity: An ethics-based criterion for medical student selection. Acad Med. 2013;88:581–584
5. Evans L, Trotter DR. Epistemology and uncertainty in primary care: An exploratory study. Fam Med. 2009;41:319–326
6. Ghosh AK. On the challenges of using evidence-based information: The role of clinical uncertainty. J Lab Clin Med. 2004;144:60–64
7. Allison JJ, Kiefe CI, Cook EF, Gerrity MS, Orav EJ, Centor R. The association of physician attitudes about uncertainty and risk taking with resource use in a Medicare HMO. Med Decis Making. 1998;18:320–329
8. Benbassat J, Pilpel D, Schor R. Physicians’ attitudes toward litigation and defensive practice: Development of a scale. Behav Med. 2001;27:52–60
9. Geller G, Rushton CH, Francomano C, Kolodner K, Bernhardt BA. Genetics professionals’ experiences with grief and loss: Implications for support and training. Clin Genet. 2010;77:421–429
10. Kvale J, Berg L, Groff JY, Lange G. Factors associated with residents’ attitudes toward dying patients. Fam Med. 1999;31:691–696
11. Gerrity MS, DeVellis RF, Earp JA. Physicians’ reactions to uncertainty in patient care. A new measure and new insights. Med Care. 1990;28:724–736
12. Cooke GP, Doust JA, Steele MC. A survey of resilience, burnout, and tolerance of uncertainty in Australian general practice registrars. BMC Med Educ. 2013;13:2
13. Sherrill WW. Tolerance of ambiguity among MD/MBA students: Implications for management potential. J Contin Educ Health Prof. 2001;21:1117–1122
14. Albanese MA, Snow MH, Skochelak SE, Huggett KN, Farrell PM. Assessing personal qualities in medical school admissions. Acad Med. 2003;78:313–321
15. Luther VP, Crandell SJ. Ambiguity and uncertainty: Neglecting elements of medical education curricula? Acad Med. 2011;86:799–800
16. Wayne S, Dellmore D, Serna L, Jerabek R, Timm C, Kalishman S. The association between intolerance of ambiguity and decline in medical students’ attitudes toward the underserved. Acad Med. 2011;86:877–882
20. Weissenstein A, Ligges S, Brouwer B, Marschall B, Friederichs H. Measuring the ambiguity tolerance of medical students: A cross-sectional study from the first to sixth academic years. BMC Fam Pract. 2014;15:6
21. Geller G, Faden RR, Levine DM. Tolerance for ambiguity among medical students: Implications for their selection, training and practice. Soc Sci Med. 1990;31:619–624
22. DeForge BR, Sobal J. Intolerance of ambiguity in students entering medical school. Soc Sci Med. 1989;28:869–874
23. DeForge BR, Sobal J. Investigating whether medical students’ intolerance of ambiguity is associated with their specialty selections. Acad Med. 1991;66:49–51
24. Budner S. Intolerance of ambiguity as a personality variable. J Pers. 1962;30:29–50
26. Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav. 1983;24:385–396
27. Cronbach LJ, Shavelson RJ. My current thoughts on coefficient alpha and successor procedures. Educ Psychol Meas. 2004;64:391–418
28. Cohen J Statistical Power Analysis for the Behavioral Sciences. 19882nd ed Hillsdale, NJ Lawrence Erlbaum Associates
30. Smedley BD, Stith AY, Nelson AR Unequal Care: Understanding and Eliminating Racial and Ethnic Disparities in Health Care. 2003 Washington, DC National Academies Press
31. Betancourt JR, Maina AW. The Institute of Medicine report “Unequal Treatment”: Implications for academic health centers. Mt Sinai J Med. 2004;71:314–321
32. Kuhn G, Goldberg R, Compton S. Tolerance for uncertainty, burnout, and satisfaction with the career of emergency medicine. Ann Emerg Med. 2009;54:106–113.e6
33. Hancock J, Roberts M, Monrouxe L, Mattick K. Medical student and junior doctors’ tolerance for ambiguity: Development of a new scale. Adv Health Sci Educ Theory Pract. doi: 10.1007/s10459-014-9510-z
34. Rosenthal JM, Okie S. White coat, mood indigo—Depression in medical school. N Engl J Med. 2005;353:1085–1088
35. Dyrbye LN, Thomas MR, Shanafelt TD. Systematic review of depression, anxiety, and other indicators of psychological distress among U.S. and Canadian medical students. Acad Med. 2006;81:354–373
36. Dyrbye LN, Thomas MR, Massie FS, et al. Burnout and suicidal ideation among US medical students. Ann Intern Med. 2008;149:334–341
37. Dyrbye LN, Massie FS Jr, Eacker A, et al. Relationship between burnout and professional conduct and attitudes among US medical students. JAMA. 2010;304:1173–1180
38. Drolet BC, Rodgers S. A comprehensive medical student wellness program—Design and implementation at Vanderbilt School of Medicine. Acad Med. 2010;85:103–110
39. Regehr C, Glancy D, Pitts A, LeBlanc VR. Interventions to reduce the consequences of stress in physicians: A review and meta-analysis. J Nerv Ment Dis. 2014;202:353–359
40. Slavin SJ, Schindler DL, Chibnall JT. Medical student mental health 3.0: Improving student wellness through curricular changes. Acad Med. 2014;89:573–577
41. Bovier PA, Perneger TV. Stress from uncertainty from graduation to retirement—A population-based study of Swiss physicians. J Gen Intern Med. 2007;22:632–638
42. Lemay JF, Lockyer JM, Collin VT, Brownell AK. Assessment of non-cognitive traits through the admissions multiple mini-interview. Med Educ. 2007;41:573–579