This small study suggests the UKCAT has very limited predictive value for the performance over the first two years of preclinical study at Nottingham. The total score had only very modest correlation with Themes A and C. This effect appeared to be exerted via Quantitative Reasoning in Theme A, and Verbal Reasoning in Theme C. Socio-demographic variables also had little influence, apart from male sex and white ethnicity in Theme C.
Strengths and weaknesses of the study
This paper adds to the currently sparse evidence relating to the UKCAT. Although the study group includes only 78% of the intake cohort, these students did not differ in socio-demographic terms from their peers, who had either not taken the UKCAT or not given permission for their data to be used. Thus there is no a priori reason to suppose that our findings are unrepresentative.
We had to exclude nine students (4% of the study group) from the statistical analyses because they did not have full datasets. The reasons were varied, and included academic difficulties, health problems, and personal issues. With such small numbers we decided that it was not appropriate to examine whether overwhelming academic failure was related to UKCAT score, although this is a potentially important issue.
We did not attempt to compare students' performance in the UKCAT with their school leaving examinations. As noted, only one publication to date has looked at this issue. That study investigated the sub-group of medical school applicants who went on to achieve at least three passes at A level, and demonstrated a modest correlation between UKCAT scores and A levels. In the current study, so many students had average A-level tariff scores at the maximum (120) that we could not use them in a comparable analysis.
The variance contributed by the explanatory variables was small. This is in line with other research at Nottingham.
Socio-demographic predictors of pre-clinical performance
Socio-demographic influences on performance were generally slight, and corresponded with previous research both at Nottingham and elsewhere[19–21]. We have no ready explanation for the poorer performance of males in Behavioural Sciences, and of non-White students in Epidemiology. These findings merit further internal investigation.
Differential effects of the UKCAT section scores
Our study suggests that the total UKCAT score has little predictive relationship with preclinical performance. It must be remembered that students on the course are already a highly selected group, and their UKCAT scores probably lie within a relatively small range, compared to the wider pool of applicants. Although school examination results have been shown to relate to academic performance on the course in the past [1, 2, 4], we might speculate that it is less likely that this correlation would be shown now, when an increasing majority of medical students have top grades. Perhaps we should not expect UKCAT scores to correlate highly either, in the select group of students who have achieved admission. However, an overall relationship between UKCAT scores and course progress might still be observed if the UKCAT is able to identify students with good potential abilities. The section scores may prove more sensitive in this respect.
The modest relationship of UKCAT Quantitative Reasoning with Theme A is not unexpected, since that part of the course is the most 'scientific' in terms of course content and assessment. The ability of Verbal Reasoning to predict performance in Theme C is interesting. The types of assessment are mixed and include oral presentation and an essay, as well as short-answer and multiple choice questions. The subject matter perhaps requires a deeper level of thought and understanding, and better articulation, than Themes A and B, which are based more on acquired knowledge. Theme D is assesses through coursework, practical communication and clinical skills, so it is unsurprising that clear cognitive predictors did not emerge, as other factors such as personality are likely to be more important. It is of note that the correlation coefficients between Themes A and B in years one and two are stronger than those between Themes C and D and the OSCEs. We speculate that this could reflect that the format and content of the respective assessments (see Supplement 1). Themes A and B comprise science knowledge in both years and use similar assessments, whereas the content and assessments of Themes C and D are more varied. Certainly the standard of clinical performance required for year two OSCEs is intended to be higher than year 1.