The Academic Backbone: longitudinal continuities in educational achievement from secondary school and medical school to MRCP(UK) and the specialist register in UK medical students and doctors

Table 6 Correlations in the Westminster Cohort Study

	AH5 score	Best three A-levels	Clinical performance	PRHO ratings	MRCP(UK) Part 1	On Specialist Register
	Continuous	Continuous censored	Ordinal	Continuous censored	Ordinal	Binary
AH5 score	1	0.281	0.155	0.074	-0.162	0.185
		P <0.001	P = 0.001	NS	NS	P <0.001
		n = 501	n = 486	n = 408	n = 106	n = 458
Best three A-levels	0.299 ± 0.041 (0.223 to 0.377)	1	0.146	0.078	0.098	0.190
			P = 0.001	NS	NS	P <0.001
			n = 483	n = 405	n = 105	n = 454
Clinical performance	0.209 ± 0.065 (0.079 to 0.336)	0.202 ± 0.057 (0.091 to 0.312)	1	0.175	0.174	0.126
				P <0.001	NS	P = 0.008
				n = 395	n = 104	n = 447
PRHO ratings	0.069 ± 0.048 (-0.037 to 0.163)	0.086 ± 0.054 (-0.020 to 0.190)	0.260 ± 0.063 (0.141 to 0.400)	1	0.093	0.076
					NS	NS
					n = 83	n = 372
MRCP(UK) Part 1	-0.154 ± 0.101 (-0.351 to 0.036)	0.138 ± 0.107 (-0.079 to 0.346)	0.302 ± 0.143 (-0.004 to 0.562)	0.124 ± 0.110 (-0.105 to 0.323)	1	0.261
						P = 0.009
						n = 98
On Specialist Register	0.229 ± 0.061 (0.109 to 0.348)	0.245 ± 0.062 (0.108 to 0.353)	0.215 ± 0.077 (0.065 to 0.363)	0.096 ± 0.069 (-0.042 to 0.231)	0.325 ± 0.129 (0.053 to 0.572cpa)	1

Abbreviations: A-level, Advanced level; BMS, Basic medical sciences; GCSE, General Certificate of Secondary Education; MCMC, Markov Chain Monte Carlo; NS, not significant; O-level, Ordinary level; PRHO, Pre-Registration House Officer.
^aCorrelations in entrants to medical school between measures of academic and professional attainment in the Westminster Cohort study.
^bCorrelations above the diagonal are simple Pearson correlations (for example, point-biserials, φ correlations, where a measure is binary or ordinal), and have different N values for various reasons.
Correlations shown in bold are significant at P<0.05.
Correlations in the lower triangle were calculated taking account of censoring, and for ordinal values are equivalent to tetrachoric/polychoric correlations.
^eValues are mean ± SD (95% CI) unless otherwise stated, and the 95% confidence intervals are the result of the final 2000 MCMC steps in a chain of 5000.

ISSN: 1741-7015