The hypothesis which we tested was whether personality traits as measured by the NEO-PI-R Test have predictive value in early medical school performance and whether this predictive value was stronger than traditional premed metrics (MCAT, GPA, etc.). Obviously, if support for this hypothesis could be obtained from this study, it would argue possibly for an expanded role of the NEO-PI-R Test in the medical school admissions process or at least for additional confirmatory retrospective and validatory prospective studies.
At MUSM, the Admissions Committee did not formally use the NEO-PI-R test to evaluate prospective applicants and were completely blinded to the NEO-PI-R Test results. Therefore, any correlations between personality scores and academic performance were made on an unselected and therefore unbiased population. In the present study we re-analyzed the MUSM raw data. We also made comparisons between the MUSM and CUSM data.
The present study also examined 2 years of CUSM applicant and matriculant data for NEO-PI-R, premedical parameters, demographic data and medical school performance data for potential predictive value of the NEO-PI-R v traditional premed parameters.
Even though the MUSM data and the CUSM data were derived from different populations of medical school applicants, approximately 15 years apart, with different demographic features, (eg.,the male / female ratio was much higher at MUSM), from different schools with different admission criteria, and from different geographic areas of the United States, the NEO-PI-R was remarkably consistent in the personality mean scores and ranges between the two groups of students. 29 of 30 facets of personality showed no differences in score distribution between the populations (p = .87; p = .78). The single facet showing a difference between the two populations was (A6) Tender-Mindedness (p = .007). This facet accounted for a difference in its member domain (A) Agreeableness (p = .034). The fact that 29/30 personality facets showed no differences between the MUSM and CUSM student populations demonstrated the remarkable consistency of the NEO-PI-R. This consistency spanned decades, schools, demographics and geographies.
Re-analysis of the MUSM data revealed a number of interesting findings. For one there were significant differences in one major personality domain as well as many of its facets between males v females. The one major domain which showed differences was (C) Conscientiousness with females scoring higher (p = .012). Females also scored higher in two of its facets: (C2) Order (p = .026) and (C6) Deliberation (p = .02). Within the domain of (A) Agreeableness, the facet (A4) Compliance showed higher scores in males (p = .032).
A number of personality domains and facets correlated with either academic success or failure in both males and females. Academic success was defined by separate and cumulative course performance and academic failure was defined as having to repeat a single course or multiple courses or dropping out of school. The predictive values of these personality domains and facets were compared to the predictive values of traditional premed metrics like MCAT VR (verbal reasoning), MCAT PS (physical sciences: chemistry, physics) and MCAT BS (biological sciences: biology, biochemistry, genetics, physiology, molecular biology, microbiology, evolution, organic chemistry). At the time of the MUSM study, the MCAT was divided into MCAT VR, MCAT PS and MCAT BS. The MCAT BS scores positively correlated with 7 different courses performances (p = .05; Pearson 0.6) and the MCAT PS positively correlated with 2 course performances (p = .05; Pearson 0.6) whereas MCAT VR negatively correlated with 4 course performances (p = .01; Pearson − 0.7). However, none of the MCAT scores correlated with academic failure.
A number of personality domains and facets also correlated positively (significant and positive Pearson coefficients) with course performances. Most of these fell within the (C) Conscientiousness domain which included (C3) Dutifulness, 5 courses (p = .03; Pearson − 0.8); (C4) Achievement Striving, 4 courses (p = .04; Pearson 0.7); and (C5) Self-Discipline, 7 courses (p = .02; Pearson 0.9). Collectively, the facets within the (C) Conscientiousness domain correlated with academic success in more courses than the MCAT BS and MCAT PS scores.
However, the most striking finding in the MUSM data was the negative correlations (significant and negative Pearson coefficients) with academic failure. The personality domains and facets which provided strong negative correlations with academic failure (repeating a single course, multiple courses or dropping out of school) fell mainly within the (N) Neuroticism domain including facets (N2) Angry Hostility (p = .05; Pearson − 0.7) and (N3) Depression (p = .05; Pearson − 0.7) and the (O) Openness to Experience domain which included facets (O2) Aesthetics (p = .05; Pearson − 0.7) and (O3) Feelings (p = .05; Pearson − 0.7). The facets within the (O) Openness to Experience domain negatively correlated with repeating not just one but multiple courses (p = .028; Pearson − 0.9). Select personality domains and facets therefore potentially add value to the admissions process as a negative predictor of academic failure.
Similarly to the MUSM students whose admissions to medical school were not at all based on the NEO-PI-R test, CUSM did not use the NEO-PI-R test to formally influence admissions. In the first class which was admitted (the class of 2022), 29 of 30 facets of personality predictably showed no differences in score distribution between the accepted vs rejected applicants (p = .250). In the second class which was admitted (the class of 2023), there were differences in only 1 domain: (N) Neuroticism. In fact, all of the facets within this domain showed differences between accepted vs. rejected applicants (p = .02). Although the NEO-PI-R test was not formally used as an Admissions Criteria and whose results were not made available to the Admissions Committee, it was entirely possible that the interviewers were sensitive to neurotic personality traits of certain applicants that negatively impacted their decisions on acceptance. It would seem then from this observation that this domain may have factored into the admission decision.
Analysis of the CUSM data revealed both similarities and differences compared to the MUSM data. The personality profiles of males vs females were again different but mainly fell in facets within the (E) Extraversion, (O) Openness to Experience and (A) Agreeableness domains (p = .02). CUSM accepted approximately equal number of males and female students whereas MUSM accepted only a limited number of female students at that time. The difference in male / female ratio between the two classes could explain the discrepancy in the differing personality facets.
Since there is currently more of an emphasis on evaluating medical school student performance to comply with the rigors of the LCME accreditation process than there was 15 years ago, CUSM used a number of performance metrics that were not available at MUSM which included Multiple Choice Questions (MCQs), NBME (both raw and scaled), Laboratory, Case Presentation, iRAT, OSCE, Course Final Grade (derived from a composite of measurements depicted below) and Overall Averages.
The Course Final Grade (Raw Score) was derived from a composite of the detailed measurements as depicted (Table 3). In addition, other premedical metrics that were available included overall MCAT, overall GPA (grade point average) and BCPM (Biology, Chemistry, Physics, Mathematics) grade point average. Presently, only an overall- MCAT score was available because the MCAT was no longer broken into MCAT VR, MCAT PS and MCAT BS as it was for the MUSM data.
At CUSM, presently, academic failure was defined as the need to repeat a course but since no CUSM students to date, however, have been required to repeat a course due to students’ 100% successful attempts at remediation, academic failure per se could not be correlated with NEO-PI-R measurements, Academic performance (success or lack thereof) based on various assessments including class rank (Table 3) could be measured and was used in this study.
With traditional premed metrics, MCAT scores surprisingly did not significantly correlate with any of the above-mentioned assessments (p = .5). However, BCPM significantly correlated with 3 of the assessments (p = .01; Pearson 0.7) and was therefore the best of the objective metrics.
However, the most striking finding discovered was the very strong negative correlations (significant and negative Pearson coefficients) with academic performance by certain personality domains and facets. The personality domains and facets which provided strong negative correlations with academic performance fell mainly within the (N) Neuroticism domain including facets (N2) Angry Hostility, (N3) Depression, (N5) Impulsiveness and (N6) Vulnerability (all, p = .02; Pearson − 0.8). These facets negatively correlated with as many as 4 of the assessments, which were more assessments than those that correlated with the BCPM. Interestingly, the (N) Neuroticism domain including facets (N2) Anger Hostility and (N3) Depression were also the same personality domain and facets that predicted academic failure at MUSM.