Investigating possible ethnicity and sex bias in clinical examiners: an analysis of data from the MRCP(UK) PACES and nPACES examinations

Table 1 The table shows separately for analyses of hawk-dove differences, male–female differences, White-nonWhite differences, and differences between odd and even numbered candidates (see columns), the numbers of examiners who reached statistical significance (rows) on various criteria

	Hawk-Dove		Male–female		White-NonWhite		Odd-Even numbering
	Positive: Examiner hawkish		Positive: Males score higher than females		Positive: Whites score higher than non-Whites
	PACES	nPACES	PACES	nPACES	PACES	nPACES	PACES	nPACES
Negative effect: P < .05 corrected	34 (1.9%)	35 (2.3%)	0	0	2 (0.1%)	1 (0.1%)	0	0
Negative effect: P < .05 uncorrected (chance expectation = 2.5%)	198 (11.1%)	235 (15.7%)	73 (4.1%)	63 (4.2%)	73 (4.4%)	48 (3.6%)	60 (3.2%)	51 (3.2%)
Not significant (uncorrected, p > .05)	1339 (74.8%)	989 (66.0%)	1638 (91.5%)	1379 (92.1%)	1491 (90.4%)	1229 (92.2%)	1680 (93.9%)	1396 (93.2%)
Positive effect: P < .05 uncorrected (chance expectation = 2.5%)	192 (10.7%)	200 (13.4%)	79 (4.4%)	55 (3.7%)	82 (5.0%)	55 (4.1%)	50 (3.0%)	51 (3.6%)
Positive effect: P < .05 corrected	27 (1.5%)	39 (2.6%)	0	0	1 (0.1%)	0	0	0
N examiners	1790	1498	1790	1497	1649	1333	1790	1498

Levels of statistical significance are divided into five groups, those who are significant at a Bonferroni corrected level of p < .05 (first and fifth rows), those who are significant at a non-Bonferroni-corrected level of p < .05 (second and fourth rows), and those who are not significant at a non-Bonferroni-corrected level of p < .05 (middle row). ‘Positive’ refers., arbitrarily, to examiners being more hawkish (i.e. giving lower overall scores), giving higher scores to male candidates, giving higher scores to White candidates, or giving higher scores to odd-numbered candidates. By chance alone one would expect 95% of candidates to be in the ‘non-significant’ group, with the remaining 5% of candidates distributed evenly between negative and positive effects.

ISSN: 1472-6920