The environment in which a student learns plays an important role in his or her well-being,1 academic performance,2 and development of professional attributes central to medical practice.3,4 There is evidence suggesting that the learning environment is a critical element in the establishment of professional identity, which in turn fosters “competency” in many areas.5 The learning environment includes the social, psychological, and physical contexts that affect or are affected by academic activities. It involves perceived support structures available to students, level of autonomy for learning, students’ emotional response, and the inherent meaning that students find in the educational process.6 Learning in medical school does not happen solely in the lecture hall, nor does all learning come from what is intentionally taught.7 Students learn as part of an informal curriculum, often referred to as the “hidden curriculum,”8 that pervades both the classroom and the clinical environments. The professional development of students during medical school is a result of many facets that make up the learning environment.
The learning environment also has become a major focus of curriculum reform in medical education.9 Research on the learning environment in the context of curriculum reform is scarce, but leaders in medical education are promulgating the necessity of collaborative learning environments that expose the “hidden” curriculum.10 In 2008, the Liaison Committee on Medical Education (LCME)—the accrediting body for U.S. and Canadian medical degree programs—added a new standard requiring all medical schools to assess the school’s learning environment and make improvements where indicated.11 To this end, all medical students and residents are now asked about their perceptions of the learning environments through national surveys.12,13
Students enter medical school with their own individual sets of expectations. The first year of medical school may validate their expectations or establish new expectations as they begin their physician training. It is important to measure students’ perceptions of the learning environment in such a pure classroom-based environment prior to their exposure to high-stakes examinations, clinical training, and other experiences. Measuring student attitudes after student exposure to each academic year of medical school may allow institutions to compare positively and negatively perceived aspects of their environments and develop appropriate interventions to help students succeed on their path to becoming physicians.
The American Medical Association (AMA)-sponsored Learning Environment Study (LES) was developed in 2010 to evaluate factors that contribute to a medical school learning environment. This study reports the initial LES data for student perceptions of the learning environment at the end of the first year of medical school. Specifically, we examined two research hypotheses:
- At the end of the first year of medical school, medical students’ demographic characteristics and self-reported attributes may be associated with their perceptions of various aspects of the learning environment.
- In turn, distinct culture at the campus or school level may be associated with variations in students’ perceptions of the learning environment.
The LES collaborative consisted of 28 North American, LCME-accredited medical schools at 38 campuses, each enrolling one or two cohorts for a four-year longitudinal study either from 2010 to 2014 or from 2011 to 2015. Cohort 1 (matriculating in 2010) consisted of students from 12 schools that had been a subset of the AMA’s Innovative Strategies for Transforming the Education of Physicians (ISTEP), a group that focused on developing ethics curricula in medical school.14 When the LCME established requirements on the learning environment, some of the ISTEP schools expressed an interest in exploring factors of the learning environment and comparing their results with a national cohort. With the support of the AMA, these 12 schools decided on appropriate instrumentation and protocol for the LES in early 2010. Cohort 2 (matriculating in 2011) included students from 11 of 12 schools in the first cohort and those from an additional 16 schools that requested to participate after presentations at national meetings about the LES. Each school selected between online and paper survey administration for LES surveys, and data collection dates were dependent on each school’s schedule, spanning approximately three months for the majority of the schools. Otherwise, participating schools followed the same data collection protocol and used the same instrumentation.
We invited a total of 6,551 students to participate from the two cohorts. At matriculation (fall of 2010 and 2011), students provided consent for their data to be included in the multi-institutional collaborative study. In addition to a demographic questionnaire, students completed four professional attribute tools at matriculation. We selected these tools because of potential relationships with perceptions of the Medical School Learning Environment Survey (MSLES): Ways of Coping Questionnaire,15,16 Tolerance of Ambiguity Scale,17 Jefferson Scale of Empathy,18,19 and Patient-Practitioner Orientation Scale.20,21 During the eighth month of the first year of medical school, we administered the 17-item MSLES survey to students. All data were self-reported, and participation was voluntary.
Human subject research approval for the multi-institution collaborative study was obtained from the University of Michigan Medical School institutional review board (IRB); each participating medical school also received approval from its own IRB to collect data on its students. We identified the learning environment for each student as the campus attended in the first year of medical school; three schools had students enrolled in multiple separate campus locations at matriculation.
The MSLES, developed originally by Marshall,6 was designed to measure learners’ perceptions of various aspects of the learning environment including social, academic, and emotional constructs. For this study, we administered Rosenbaum and colleagues’22 17-item version of the questionnaire to medical students at the end of the first year of medical school. This iteration of the MSLES included 16 items from the original instrument, as well as an additional item about vertical integration. Items are rated on a five-point frequency scale (1 = “never,” 2 = “rarely,” 3 = “sometimes,” 4 = “often,” 5 = “very often”); the outcome variable used is the average of all responses on the tool.
We included the following self-reported demographic characteristics: gender, race/ethnicity, previous health-related work experience, physician in the family, the type of location in which the student grew up, and marital status. Gender was reported as female or male. Ethnicity was reported as white; Asian; or underrepresented in medicine (UIM), which included Native American, Alaska Native/Eskimo/Inuit, African American/African descent/black, and Hispanic/Latino/Chicano; and other. We categorized participants who selected both a UIM category and other ethnicities as UIM. Previous health-related work experience and the presence of a physician in the family were both reported as yes or no. Location in which a student grew up was reported as being city, suburban, rural, or town. Marital status was reported as being single, married, divorced, or living with a partner.
We performed one-way ANOVA and chi-square tests on continuous and categorical variables for each student demographic and professional attribute variables, respectively, to determine whether students differed statistically significantly across campuses. We specified the campus as the primary location of the student during the first year of medical school. Of three schools that had multiple campuses, two schools had students identify their campus when completing the surveys. The third school identified the primary campus in a separate dataset containing only the LES ID code and campus location. Students who completed at least one demographic item or at least one professional attribute scale were included. We conducted a second set of one-way ANOVA models for all demographic and professional attribute variables to determine whether there was a statistically significant association with MSLES responses and to calculate the percent variance explained by each variable. Seven items on the MSLES were reverse coded to help interpretation before calculating the MSLES score; higher scores denote a more positively perceived learning environment. We calculated the MSLES score as a simple average of responses on each of all 17 items; data from students who answered all 17 items were included in the MSLES analyses.
We calculated descriptive statistics (mean, standard deviation, and 95% confidence interval) by campus for each of the professional attributes where there were more than 5 responses (34 campuses). We performed a mixed-effects generalized linear model controlling for all student demographic and professional attributes, cohort, and random campus effects. Mixed models are hierarchical models that incorporate random and fixed effects. We calculated the type III sum of squares to determine the percent variance explained by each variable included in the model. All analyses were performed using SAS Enterprise Guide statistical software version 6.4 (SAS Inc., Cary, North Carolina).
At matriculation in either 2010 or 2011, all 6,551 students at 28 medical schools and 38 campuses were invited to participate in the LES study. Between the two cohorts, 4,664 students enrolled in the study (71% response rate) (see Table 1). There were 1,612 (35%) participants from 12 medical schools (14 campuses) in Cohort 1 and 3,052 (65%) participants from 27 schools (37 campuses) in Cohort 2. One school was unable to collect data at the second time point and therefore was excluded from the MSLES analyses, but was included in all other analyses. Some student responses on the MSLES survey taken in the spring could not be matched to their surveys taken at matriculation because of incorrectly entered subject IDs. This accounted for the higher attrition rates in some schools. The participation rates by school, ranging from 29% to 92%, with the median of 79% at matriculation, were calculated by the ratio of participants to class enrollment from the LCME Part II survey and Association of Faculties of Medicine Canada enrollment data. Of the 3,422 students who answered at least one question on the MSLES questionnaire at the end of the first year of medical school, 3,375 students (98.6%) completed the entire MSLES survey.
There were statistically significant differences (P < .0001) for all demographic characteristics, except gender, as presented in Table 1. Students’ professional attributes were also statistically significantly different across the 28 schools at matriculation as shown in Table 1. Figure 1 displays the mean and 95% confidence interval of MSLES scores by campus at the end of the first year of medical school; schools differed significantly (P < .0001) with respect to students’ perceptions of the learning environment, as reported at the end of their first year of enrollment. Table 2 displays the sample size, observed mean, and standard deviation for each item on the MSLES tool. Results show high scores on items about the relationship with their medical school peers and low scores on items about their ability to find time for family and friends.
Several demographic and professional attribute variables at matriculation were significantly associated (P < .05) with total MSLES score in the mixed-effects general linear model: cohort, race/ethnicity, previous health experience, marital status, active ways of coping, and total empathy score (Table 3). Cohort and race/ethnicity were each associated with a 0.2% variance in total MSLES scores. Previous health experience and marital status were each associated with 0.3% and 0.4% variance, respectively, in total MSLES scores. Active ways of coping were associated with 0.1% of variance in total MSLES scores, and total empathy accounted for 0.7% of variance in total MSLES scores. Despite these statistically significant relationships, in aggregate these variables of student demographics, student professional attributes, and cohort together accounted for 2.2% of the variance as depicted in Figure 2. Medical school campus location was the variable that explained the largest amount of variance in the perception of the learning environment at the end of the first year of medical school. When controlling for all the variables, within-campus effect accounted for 15.6% of the variance in MSLES scores. The covariates measured in the generalized linear model accounted for 17.8% of the variance in MSLES scores.
Creating a positive medical school learning environment continues to be a major challenge for medical schools.10 Student perceptions of the learning environment have been linked to academic performance, and positive student reports are correlated with better performance on United States Medical Licensing Exam Step 1.2 The reasons for these differences in perceptions need to be investigated and strategies for improvement sought.
Through each school’s unique admissions process, students arrive at medical school with varying characteristics, including demographics, background, experiences, and personal and professional attributes. It is a reasonable hypothesis that these characteristics are a major influence on how students perceive the medical school learning environment. Despite the statistically significant differences across participating schools with respect to these student characteristics, our findings did not demonstrate specific salient factors as to why students perceived the learning environment differently at the end of the first year of medical school. Student scores on the Jefferson Empathy Scale were highly significant (P < .0001) predictors of MSLES scores, yet the total empathy scores accounted for less than 1% of the variance in MSLES scores. The attribute of active ways of coping was not a significant explanation of variance in MSLES in the bivariate analysis, but when controlling for all the other variables in the model it did reach levels of statistical significance. In aggregate, all of the variables related to student demographics, background, and attributes accounted for only 2.2% of the variation in their perceptions of the medical school learning environment. This suggests that, holding all other demographics and professional attributes constant, the ways that students interact with the stressors of medical school may also be related to how they perceive the learning environment.
The significant differences in MSLES scores across schools and campuses suggest that the medical school culture at each location plays a significant role in student experiences and perceptions in the first year of medical school. Although it is not known what other factors may have been related to how students perceived their learning environment, the campus itself accounted for almost all, 88.9% (16% of 18% measured), of the total variance accounted for in the study. Elements in the milieu of each medical school and campus create a school’s culture, and thus affect student perceptions in the first year of medical school. We observed less variation within a school and more variation among schools with regard to perceptions of the learning environment. This is expected if learning environments actually do differ on institutional, curricular, and “hidden” curricular characteristics, independent of subjective student perceptions. These factors may include measurable attributes, such as grading policy; existence of learning communities; advisors and mentors; class size and demographics; and intrinsic factors that are less easy to measure, such as perceived support and competition. Identification of specific elements contributing to a medical school’s learning environment culture will be important to accentuate the positive influences and mitigate negative ones.
Although we found that students’ demographic characteristics accounted for very little of the total variance in MSLES scores, it may be prudent for schools to examine the perceptions of their own population of students in order to create, develop, and foster positive learning environments lest personal well-being and academic performance suffer. Student characteristics and attributes like gender, race/ethnicity, marital status, or ways of coping may not have had an overall impact on this 28-school dataset, but some students in a particular school environment may have experienced differences which affected their perceptions. Our findings may suggest that the culture of a school may significantly impact student perceptions very early on, well before students’ clinical experiences.
Our study provided an opportunity to go beyond the single-institution focus of most studies of the learning environment and to hypothesize about relationships between student demographics, self-reported professional attributes, and the institutional culture. In the past, most studies evaluated the impact of these variables through pre–post intervention comparisons within a single school.
Analyses of such a large dataset allow for a greater statistical power and potential generalizability. Additional research on measurable aspects of the learning environment—such as pass/fail grading system; presence of learning communities or other support structures; class size; financial resources; and sociocultural issues—may provide more insight into how to create, develop, and foster positive perceptions of the learning environment.
Limitations of this study include those for self-reported survey research related to perceptions. One is that the data were obtained on a voluntary basis, and there was a 27% drop-off between administration of LES surveys at matriculation and the end of the first year, which may have led to sampling bias and some differences from the demographics reported at matriculation. The individuals who did not respond may be systematically different from those who did respond, thus distorting the results. The sample of participants in this study may not be a representation of each school’s full student body. This study does not include a measure of student academic performance in medical school, a variable that could influence perceptions of the medical school learning environment. Some schools had higher attrition rates because some student responses on the MSLES survey taken in the spring could not be matched to their surveys taken at matriculation. Finally, the large sample size of this cohort study results in statistically significant differences in measurements that must be carefully considered for their real significance and interpreted thoughtfully and with care.
This study of the medical school learning environment at 28 U.S. and Canadian schools revealed that students’ perceptions of the learning environment differ significantly across school and campuses after completing only one year of medical school. The student’s school or campus location, with its inherent local institutional culture, explains almost 90% of the measured variance in student perception of the learning environment. The relationships between the MSLES scores, student demographic, and personal attribute measures, although statistically significant, only explained about 2% of the variance. This finding suggests that medical schools can examine elements of their institutional learning environment, such as grading policies, learning communities, and curricular change efforts, to enhance student experiences in undergraduate medical education. Further study is needed to identify specific factors in the learning environment that may contribute to student perceptions, as all variables measured in this study account for 18% of the variance in students’ perception of the learning environment.
1. Dyrbye LN, Thomas MR, Massie FS, et al. Burnout and suicidal ideation among U.S. medical students. Ann Intern Med. 2008;149:334341.
2. Wayne SJ, Fortner SA, Kitzes JA, Timm C, Kalishman S. Cause or effect? The relationship between student perception of the medical school learning environment and academic performance on USMLE Step 1. Med Teach. 2013;35:376380.
3. Christianson CE, McBride RB, Vari RC, Olson L, Wilson HD. From traditional to patient-centered learning: Curriculum change as an intervention for changing institutional culture and promoting professionalism in undergraduate medical education. Acad Med. 2007;82:10791088.
4. Cruess RL, Cruess SR, Boudreau JD, Snell L, Steinert Y. Reframing medical education to support professional identity formation. Acad Med. 2014;89:14461451.
5. Jarvis-Selinger S, Pratt DD, Regehr G. Competency is not enough: Integrating identity formation into the medical education discourse. Acad Med. 2012;87:11851190.
6. Marshall RE. Measuring the medical school learning environment. J Med Educ. 1978;53:98104.
7. Hundert EM, Hafferty F, Christakis D. Characteristics of the informal curriculum and trainees’ ethical choices. Acad Med. 1996;71:624642.
8. Hafferty FW, Franks R. The hidden curriculum, ethics teaching, and the structure of medical education. Acad Med. 1994;69:861871.
9. Association of American Medical Colleges. AAMC statement on the learning environment. 2014. https://www.aamc.org/download/408212/data/learningenvironmentstatementdownload.pdf
. Accessed January 14, 2016.
10. Irby DM, Cooke M, O’Brien BC. Calls for reform of medical education by the Carnegie Foundation for the Advancement of Teaching: 1910 and 2010. Acad Med. 2010;85:220227.
11. Liaison Committee on Medical Education. http://www.lcme.org
. Accessed January 14, 2016.
12. Association of American Medical Colleges. AAMC year two questionnaire (Y2Q). https://www.aamc.org/download/426076/data/y2q2014report.pdf
. Accessed January 14, 2016.
13. Association of American Medical Colleges. Graduation questionnaire. 2013. https://www.aamc.org/data/gq/
. Accessed January 14, 2016.
14. American Medical Association. Innovative strategies for transforming the education of physicians, ISTEP. http://www.ama-assn.org/ama/pub/education-careers/istepinnovative-strategies-transforming-education-physicians.page
. Accessed January 14, 2016.
15. Brown S. Factor structure of a brief version of the ways of coping questionnaire: A study with veterinary medicine students. Meas Eval Couns Dev. 1994;27:308315.
16. Folkman S, Lazarus RS. Manual for the Ways of Coping Questionnaire. 1988.Palo Alto, Calif: Consulting Psychological Press.
17. Budner S. Intolerance of ambiguity as a personality variable. J Pers. 1962;30:2950.
18. Hojat M, Gonnella JS, Nasca TJ, Mangione S, Veloksi JJ, Magee M. The Jefferson Scale of Physician Empathy: Further psychometric data and differences by gender and specialty at item level. Acad Med. 2002;77(10 suppl):S58S60.
19. Hojat M, Gonnella JS, Nasca TJ, Mangione S, Vergare M, Magee M. Physician empathy: Definition, components, measurement, and relationship to gender and specialty. Am J Psychiatry. 2002;159:15631569.
20. Krupat E, Putnam SM, Yeager C. The fit between doctors and patients: Can it be measured? J Gen Intern Med. 1996;11:134.
21. Krupat E, Rosenkranz SL, Yeager CM, Barnard K, Putnam SM, Inui TS. The practice orientations of physicians and patients: The effect of doctor–patient congruence on satisfaction. Patient Educ Couns. 2000;39:4959.
22. Rosenbaum ME, Schwabbauer M, Kreiter C, Ferguson KJ. Medical students’ perceptions of emerging learning communities at one medical school. Acad Med. 2007;82:508515.