- Top of page
- Discussion and Conclusions
Background. UK-trained medical students and doctors from minority ethnic groups underperform academically. It is unclear why this problem exists, which makes it difficult to know how to address it.
Aim. To investigate whether demographic and psychological factors mediate the relationship between ethnicity and final examination scores.
Sample. Two consecutive cohorts of Year 5 (final year) UCL Medical School students (n= 703; 51% minority ethnic). A total of 587 (83%) had previously completed a questionnaire in Year 3.
Methods. Participants were administered a questionnaire in 2005 and 2006 that included a short version of the NEO-PI-R, the Study Process Questionnaire, and the General Health Questionnaire (GHQ) as well as socio-demographic measures. Participants were then followed up to final year (2007–2010). White and minority ethnic students’ questionnaire responses and final examination grades were compared using univariate tests. The effect of ethnicity on final year grades after taking into account the questionnaire variables was calculated using hierarchical multiple linear regression.
Results. Univariate ethnic differences were found on age, personality, learning styles, living at home, first language, parental factors, and prior education. Minority ethnic students had lower final exam scores, were more likely to fail, and less likely to achieve a merit or distinction in finals. Multivariate analyses showed ethnicity predicted final exam scores even after taking into account questionnaire factors.
Conclusions. Ethnic differences in the final year performance of two cohorts of UCL medical students were not due to differences in psychological or demographic factors, which suggests alternative explanations are responsible for the ethnic attainment gap in medicine.
- Top of page
- Discussion and Conclusions
The underachievement of minority ethnic groups is a problem across UK higher education (Broecke & Nicholls, 2007; Connor, Tyres, Modood, & Hillage, 2004; Richardson, 2008). Within higher education, medical students are selected to be high academic achievers, but despite this, minority ethnic students still achieve lower grades at medical school than their white colleagues. This ethnic attainment gap has been found at various medical schools in the United Kingdom (Woolf, Potts, & McManus, 2011), the United States (Colliver, Swartz, & Robbs, 2001; Dawson et al., 1994; Koenig, Sireci, & Wiley, 1998; Veloski, Callahan, Xu, Hojat, & Nash, 2000; Xu, Veloski, Hojat, Gonnella, & Bacharach, 1993), and in other ‘Western’ English-speaking countries (Kay-Lambkin, Pearson, & Rolfe, 2002; Liddell & Koritas, 2004).
The reasons for the ethnic gap in attainment are unclear. Often all that can be said is that a difference exists, with no apparent mechanism, and with confounding always a possibility. In rare cases, racial discrimination can be invoked, as in studies of selection for medical school (McManus, Richards, Winder, Sproston, & Styles, 1995) and medical jobs (Esmail & Everington, 1993). The force of the claim in each of those cases comes from the study design. The Esmail and Everington study sent matched application forms, differing only in name and ethnicity, and hence differences in interview rates were directly dependent upon ethnicity and ethnicity alone. The McManus et al. (1995) study relied, in effect, on Mendelian randomization, applicants of mixed ethnicity (with one parent white and the other non-White, receiving more offers if the male parent were white (and hence the surname was European), than if the female parent were white (and the surname was non-European).
As far as we know, those two studies are unique in the UK medical education literature in being able directly to ascribe observed ethnic differences to discrimination. Other studies invoke the much more complex phenomenon of underperformance, which may arise from a mix of social and cultural factors, possibly stretching back generations. Cohort studies, in which participants are not randomized, have to use statistical methods to disentangle any effects of overt discrimination from those of underperformance. That being said, we can be fairly sure that overt discrimination in examinations per se is unlikely to be the cause of the ethnic gap, not least because the effects are also found in machine-marked knowledge examinations (Dewhurst, McManus, Mollon, Dacre, & Vale, 2007; Woolf et al., 2011). More subtle effects, such as stereotyping (Lempp & Seale, 2006; Woolf, Cave, Greenhalgh, & Dacre, 2008) or stereotype threat (Steele & Aronson, 1995) may affect minority students’ learning and performance, but difficulty in measuring such subtle effects means much of the quantitative research in this area has concentrated on seeing whether data routinely collected by universities can help explain the ethnic gap in attainment. We briefly review the evidence for the three most common suggestions – socio-economic group, prior attainment and language – before explaining our approach to investigating this problem.
Minority ethnic groups tend to have higher levels of socio-economic deprivation (Platt, 2009; The Cabinet Office, 2003), and both ethnicity and socio-economic group are linked to attainment in compulsory education (Department for Education and Skills, 2006; Strand, 2008; Zwick & Green, 2007). For example, in the United Kingdom, much of the difference in the examination performance of white British and Pakistani and Bangladeshi pupils at age 16 is due to the latter's lower socio-economic group (Strand, 2008). Ethnicity and socio-economic group also have indirect effects on General Certificate of Secondary Education (GCSE) and A-level achievement via parental education and schooling (McManus, Woolf, & Dacre, 2008). In higher education, Fielding, Charlton, Kounali, and Leckie (2008) analysed the degree results of 66,432 students taking finals at UK universities in 2004–2005 [note: although medicine and dentistry was included, medicine does not have a classified degree, and therefore the medicine and dentistry results in the study refer only to intercalated degrees (cfRichardson, 2008)]. Fielding et al. (2008) used multilevel modelling techniques to examine which factors mediated or moderated the relationship between ethnicity and degree attainment. The results showed that socio-economic group did account for a small amount of the ethnic gap, but did not fully explain it.
Medical students traditionally come from the highest socio-economic groups (Seyan, Greenhalgh, & Dorling, 2004) and therefore ethnic differences in medicine may be related to socio-economic status in a way not found in the general university population. Some reports have claimed that socio-economic group relates to medical school performance (The Royal Commission on Medical Education, 1968) but this relationship has not been found in other studies (McManus & Richards, 1986). Few studies have analysed differences in medical school attainment by both socio-economic group and ethnicity. A systematic review of UK studies (Woolf et al., 2011) found only two, and both showed effects of ethnicity on attainment, which were not accounted for by differences in socio-economic status or schooling (Lumb & Vail, 2004; Yates & James, 2010).
School-leaving examination performance is probably the strongest predictor of medical school exam performance (Ferguson, James, & Madeley, 2002; James & Chilvers, 2001; McManus, Smithers, Partridge, Keeling, & Fleming, 2003), but few UK studies have looked at ethnic differences in medical school attainment controlling for this factor. The two studies we found showed ethnic differences persisted when controlling for prior attainment (Lumb & Vail, 2004; Yates & James, 2007).
Language or communication difficulties
Studies from around the globe have found that medical students who are native speakers of the language in which they are being assessed achieve higher scores (Chur-Hansen, 1997; Frischenschlager, Haidinger, & Mitterauer, 2005; Liddell & Koritsas, 2004). Although language and ethnicity are linked, they are of course different. UK studies looking at whether language ability explains ethnic differences in medical school attainment have mixed results. Wass, Roberts, Hoogenboom, Jones, and Van der Vleuten (2003) found that minority ethnic students achieved lower final year communications skills grades, and qualitative analysis suggested that a sub-group of male minority ethnic students performed particularly poorly and that white examiners and minority candidates may have had subtly different interpretations of ‘good’ communication. Haq, Higham, Morris, and Dacre (2005) however found ethnic differences in the Year 3 medical school attainment of native English speakers, suggesting language ability cannot fully explain minority underperformance.
The current study
The current study sought to explore whether the above-mentioned factors, or other factors previously found to influence medical school attainment (e.g., Ferguson et al., 2002) could account for ethnic differences in attainment at one UK medical school. In particular, we sought to measure the effects of psychological factors including personality, stress (psychological morbidity), study habits, and negative experiences in addition to the demographic and academic factors usually examined in studies of this type.
Discussion and Conclusions
- Top of page
- Discussion and Conclusions
This study of two cohorts of UK students from one medical school showed that minority ethnic students performed more poorly than their white colleagues in final year practical (OSCE) and written examinations. This ethnic gap in attainment could not be explained by students’ self-reported motivation for becoming a doctor, whether they completed their secondary education in the United Kingdom, how well they did in their examinations prior to medical school, how conscientious they were, how old they were, or what their father's socio-economic status was, or whether they transferred to UCL from Oxbridge – despite those factors themselves all significantly predicting final year exam scores. Other factors such as having parents who are medical doctors, speaking English as a first language, having a parent who speaks English as a first language, living at home during term time, study habits, desire to drop out of medical school, desire to practice medicine upon qualifying, experiences of negative events, stress (all measured in Year 3), did not explain the ethnic attainment gap, nor did they predict final year performance in the final multivariate model.
This study is the first to explore in depth a large number of possible psychological and demographic reasons for the ethnic difference in attainment frequently found in medical education at both undergraduate and postgraduate levels. The longitudinal study design allowed causal inferences to be made between questionnaire variables (measured in Year 3) and examination performance (measured in Year 5 – final year). The relatively large sample size provided sufficient statistical power to minimize type II errors. The multivariate analysis allowed us to show that, for example, although speaking English as a first language predicted performance, this was not the reason that minority ethnic students underperformed.
Although the sample size was fairly large, it was not large enough to distinguish the performance of students from the different minority groups in the regression analyses. It is known that differences in the educational attainment of various minority ethnic groups exist. For example, at GCSE Indian students achieve higher grades than white and than other minority ethnic groups (Department for Education and Skills, 2006; Strand, 2008). By the time students reach higher education, the picture has changed slightly with all minority groups having lower attainment than the white British group, despite minority ethnic groups varying in the proportions attending higher education (HE), in the types of university they attend, and in the courses they study (Fielding et al., 2008). The white/non-white distinction in our study therefore appears to be of some importance and have some validity. We did also however conduct some sub-analyses that showed the white British group achieved higher scores than the Indian, ‘white other’, and ‘all other minority ethnic’ groups on the practical OSCE exam, whereas on the written exam, the difference between the white British and Indian groups was non-significant. This suggests research to explore differences in medical school attainment between various minority ethnic groups is worthwhile.
Students who did not respond to the questionnaire in Year 3 had significantly worse final year examination performance and were significantly more likely to be from minority ethnic groups, and those two factors were probably confounded. It is probable that the non-respondents were also different in ways we could not measure. For example, the administration of the questionnaires in lectures meant no distinction could be made between non-respondents who chose not to participate and those who did not attend the lectures. Those who did not attend the lectures are likely to have been disorganized and/or lower on conscientiousness, which are themselves predictors of lower examination performance (Ferguson et al., 2002; Stanley, Khan, Hussein, & Tweed, 2006; Wright & Tanner, 2002). All that being said, the pattern of ethnic differences in OSCE and written examinations was similar in the respondent sample as in the total sample indicating that non-responder bias was not too much of a problem.
All of the measured variables in this study explained only 18–19% of the variance in final year exam results. While we included A-level and GCSE scores, we did not include medical school examination results from before final year, which almost certainly reduced the amount of variance explained. The reason for excluding previous examination data was their complexity (due to the structure of the course, retakes, and interruptions, a student in a particular Year 3 cohort could be in a different Year 1, Year 2, Year 4, and Year 5 cohorts from his or her colleagues and in addition, Year 1 and Year 2 exam data were not available for Oxbridge transfer students). The questionnaire was also unable to measure more subtle psychological processes around identity and stereotyping, or perceptions of institutional climate, all of which may contribute to ethnic differences in attainment (e.g., Cohen, Garcia, Apfel, & Master, 2006; Steele, 2010; Steele & Aronson, 1995). In addition, the timing of the questionnaire administration meant that many of the psychological factors were measured at the start of Year 3, which was several years before students sat their final year examinations. For the stable ‘facts’ about students such as their first language, or their GCSE results, this was not a problem; but for the more changeable factors such as stress or experience of negative events, the time lag may have meant that any effects relating to these factors were not observed.
This study was conducted with two cohorts at a single London medical school, which limits its generalizability, although multivariate analyses from two other UK medical schools (Lumb & Vail, 2004; Yates & James, 2007, 2010) have also shown ethnic in medical student attainment persist after controlling for socio-demographic factors and previous attainment. Results from a third UK medical school however found that adjusting for sex, disability, year, and interaction effects removed a previously significant effect of ethnicity on performance in a progress test taken by medical students every year (Ricketts, Brice, & Coombes, 2010). Studies from the United States have found that minority ethnic students underperform in national licensing examinations compared to their performance in the medical college admissions test and in medical school examinations (Kleshinski, Khuder, Shapiro, & Gold, 2007; Koenig et al., 1998; Veloski et al., 2000; White, Dey, & Fantone, 2007; Xu et al., 1993).
This study found that speaking English as a first language, being schooled in the United Kingdom, and having at least one parent who speaks English as a first language were all predictors of good performance. It may be that students born and brought up outside the United Kingdom find medical school more difficult; however, being schooled outside the United Kingdom was a significant predictor only for the practical OSCE examination and not for the written exam. The OSCE requires communication skills whereas the written exam does not, so country of schooling may be a proxy for communication or cultural differences. Student support programmes in many medical schools help students manage issues of fitting in, while also not assuming that everyone from outside the United Kingdom struggles with these problems (Hawthorne, Minas, & Singh, 2004; Winston, van der Vleuten, & Scherpbier, 2010; Yates & James, 2006). More research is required to establish the long-term effectiveness of such programmes in reducing ethnic differences in attainment.
Having a father from the top socio-economic group had a small but significant positive effect on final OSCE, but not written scores. As mentioned previously, few studies have analysed the effect of socio-economic group on medical student academic attainment, and those that have generally find it does not predict performance. However, this may reflect that medical students from lower socio-economic groups are under-represented in medicine. There is a strong drive to widen access to students from non-traditional backgrounds (Mathers & Parry, 2009; Powis, Hamilton, & McManus, 2007), which makes it important to monitor attainment by socio-economic background to discover whether those from lower socio-economic groups are disadvantaged in the medical educational process.
Having studied medicine at Oxbridge universities was one of the strongest predictors of good performance in our sample. Oxford and Cambridge students tend to transfer to another medical school for their later medical school training, but we are not aware of any other studies that have included ‘Oxbridge’ as a predictor of undergraduate exam results, although one study found Oxford and Cambridge graduates achieved the highest scores of all UK medical schools on the postgraduate Membership of the Royal College of Physicians (UK) examination (McManus, Elder, et al., 2008). In many ways this is unsurprising. The fact that the ‘Oxbridge’ variable remained statistically significant after controlling for GCSE and A-level results however shows that prior school attainment was not sufficient to explain this difference. An analysis of earlier medical school performance would help disentangle the effects of pre-medical school factors and differences in teaching and learning during the early years of medical school.
Interestingly, being motivated to study medicine for financial and status reasons was a positive predictor of good performance on the OSCE. This was an unexpected finding because being a ‘good doctor’ is generally perceived to be related to altruistic motivations. The finding requires replication, but it perhaps underlines that being good at exams is not necessarily the same thing as being a good doctor (Taylor, 2006).
In summary, the results of this study clearly show that ethnic group had an independent and negative effect on final year examination performance, and although ethnic differences existed on a number of demographic variables, the relationships between ethnic group and examination performance was found to be virtually unmediated by age, socio-economic group, sex, schooling, parents’ education, language, personality, study habits, or motivation. The study focussed on mainly stable, student-related variables, and many of them were found to predict performance. However, together they explained less than 20% of the variance in scores, meaning that many unmeasured variables played a large and significant part in the examination performance of our sample.
Having ruled out many possible explanations for the ethnic attainment gap, we are left still with the task of explaining this phenomenon. The list of contenders is long, possibly endless; but in studying both what predicts performance and what explains the ethnic gap in performance, medical education researchers could take their lead from the wealth of research on school examination performance (e.g., Burgess & Greaves, 2009). If similar large-scale datasets existed in medicine in the United Kingdom, it would make it easier to disentangle the student-related, teacher-related, and medical school-related factors influencing students’ performance and in particular, the factors influencing the ethnic gap in medical students’ performance. Another approach would be to examine the issue in more detail at different medical schools using qualitative research techniques. For example, previous qualitative research on the same population (Woolf et al., 2008) has highlighted the importance of the student–teacher relationship to learning and the possible contributory effects of social psychological phenomena such as stereotyping on the ethnic attainment gap.
Finally, although understanding how demographic and other stable student-related factors correlate with performance is interesting and useful, for the benefit of those teaching our future doctors it is arguably more important that researchers strive to understand the micro-structure, as well as the macro-structure of medical education. By understanding how day-to-day occurrences in seminars, lectures and on the wards influence the learning, academic attainment, and clinical performance of students, we may also discover what is responsible for the ethnic attainment gap.