%0 Journal Article
%T Leniency and halo effects in marking undergraduate short research projects
%A Brian H McKinstry
%A Helen S Cameron
%A Robert A Elton
%A Simon C Riley
%J BMC Medical Education
%D 2004
%I BioMed Central
%R 10.1186/1472-6920-4-28
%X Review of grades awarded by supervisors, second markers and control markers to the written reports of 4th year medical students who had participated in an SSM during two full academic years (n = 399). Paired t-tests were used to compare mean marks, Pearson correlation to look at agreement between marks and multiple linear regression to test the prediction of one mark from several others adjusted for one another.There was a highly significant difference of approximately half a grade between supervisors and second markers with supervisors marking higher. (t = 3.12, p < 0.01, difference in grade score = 0.42, 95% CI for mean difference 0.18¨C0.80). There was a high correlation between the two marks awarded for performance of the project and the written report by the supervisor (r = 0.75), but a low-modest correlation between supervisor and second marker (r = 0.28). Linear regression analysis of the influence of the supervisors' mark for performance on their mark for the report gave a non-significant result. This suggests a leniency effect but no halo effect.This study shows that with the use of structured marking sheet for assessment of undergraduate medical students, supervisors marks are not associated with a halo effect, but leniency does occur. As supervisor assessment is becoming more common in both under graduate and postgraduate teaching new ways to improve objectivity in marking and to address the leniency of supervisors should be sought.There is compelling evidence from the literature that supervisors may be unreliable when asked to assess the performance of their own students. Effects such as the so-called 'halo' effect [1] in which a good or bad performance in one area affects the assessor's judgement in other areas and 'leniency'[2] where assessors are reluctant for a variety of reasons including fear of impairing the student-teacher relationship, fear of a negative emotional reaction from the student, or of poor reflection on the teacher's own expertise may
%U http://www.biomedcentral.com/1472-6920/4/28