Comparison of the sensitivity of the UKCAT and A Levels to sociodemographic characteristics: a national study

BMC Medical Education

Table 2 Results of univariate regression analyses of standardised A level summed tariff and UKCAT total score

Predictor variable	Outcome variable	Intercept (score)	Coefficient	p^§	95% confidence intervals for regression coefficient
State school^†	UKCAT	.16	-.31	<0.001	-.35 to - .27
	A level	.21	-.40	<0.001	-.45 to - .36
Non-white ethnicity	UKCAT	.16	-.43	<0.001	-.47 to - .38
	A level	.11	-.29	<0.001	-.34 to - .25
>20 years at application	UKCAT	.01	-.68	<0.001	-.93 to - .43
	A level	.01	-.88	<0.001	-1.21 to - .55
Non-professional background	UKCAT	.02	-.41	<0.001	-.51 to - .32
	A level	.02	-.36	<0.001	-.45 to - .26
Male sex	UKCAT	-.08	.17	<0.001	.13 to .22
	A level	-.01	.01	0.62	-.04 to .06
EASL	UKCAT	.07	-.47	<0.001	-.53 to - .41
	A level	.03	-.20	<0.001	-.26 to - .14

The dependent variables (both standardised as z scores) are regressed on to the WP sociodemographic categories and sex, which serve as the predictor variables.
ATobit regression was used where A level attainment was the dependent variable whilst linear regression was used where UKCAT score was the outcome.
^§p values must be interpreted cautiously where confidence intervals are derived by bootstrapping due to the possibility of asymmetric probability distributions.
^†Excludes state grammar schools.

ISSN: 1472-6920