![Sample our Education journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5932/TOC-Subject-Sample-2021-Education.jpg"})
Abstract
In the criterion‐based clinical assessment context, and more broadly, examination resits continue to present an ongoing dilemma to educators. The number of students involved is generally a small percentage of the cohort, but it is precisely with these students that institutions bear a heavy responsibility in ensuring they reach a sufficiently high standard of performance, and are accurately assessed. The students must be given a “fair” opportunity to reach the appropriate standard(s), but at the same time, especially when clinical skills are being assessed, the public must be protected from incompetent practitioners. However, this creates a delicate tension for institutions since resits present an additional financial burden, and student failures adversely effect external funding.
This paper investigates and compares the main clinical assessment and resit profiles of students in two medical schools, each of which employs a different standard setting methodology. The analysis demonstrates that resit students improve their performance by a substantial amount; this finding applies across all undergraduate clinical levels. The key issue under investigation is whether resits, as currently constituted, offer an unfair advantage to the resit student. In essence, are resits are too easy?
Notes
1. For those wishing to acquaint themselves with a good working knowledge of assessment methodologies, the authors would recommend the book by Cizek (Citation2001).
2. Issues surrounding this form of assessment may be found in Goldstein (Citation2004).
3. In other words, each grade category which forms part of the comparison must contain a reasonable number of students. Kramer et al. (Citation2003) details the borderline regression method; Pell and Roberts (Citation2006) discuss practical issues in applying the borderline regression method.
4. The standard errors for the point estimates of the assessment means and assessment pass marks have been assumed to be the same. Strictly speaking this is not so, but the pooling of variance for about 20 stations is complex and would usually yield a smaller standard error, so the methodology used is conservative.
5. It should be noted that for such small groups the computed values for Cronbach’s alpha are unlikely to be representative of larger groups, however, they do demonstrate the uniformity of performance between the stations of the students.
6. Two students had not attempted all stations.
7. For large cohorts it is usual to have a number of circuits (either concurrent or consecutive) with one assessor assessing at most 20 students, resulting in within station between assessor variance which adversely affects the alpha score.
8. Although not given here, the coefficient of determination (R 2) and the curve fit for each resit station has been computed and found to be comparable to the main assessment values.
9. In the main assessment 250–300 assessors are used per cohort; in order to minimise bias, these assessors are randomly assigned to assessments within parallel circuits and rotations with respect to their areas of expertise. However, because of the small student numbers resits usually consist of a single circuit, or two at the most, with the assessors hand picked from within the institution.
10. For school A, 2006 Y3 stripping out the first time assesses, giving a failure of 1 out of 6.