Figure 3

Mean score of candidates on the assessments made at each of the five stations by candidate ethnicity (solid circles, white candidates; open circles, non-white candidates), in relation to the ethnicity of the two examiners (white-white, both examiners white; white-non-white, one examiner white and the other non-white; non-white-non-white, both examiners non-white). Error bars indicate ± one standard error of the mean, and are wider for the white-non-white and particularly for the non-white-non-white groups, as there were fewer candidates in those groups).