Skip to main content

Table 1 Descriptive statistics for predictor-outcome correlations and criterion-related construct validities

From: Construct-level predictive validity of educational attainment and intellectual aptitude tests in medical student selection: meta-regression of six UK longitudinal studies

   

Predictor-outcome correlations

Criterion-related construct validities

  

N participants

Pearson

Corrected*

r

Equivalent N

 

N studies

Mean (Range)

Mean (Range)

Mean (Range)

Mean (Range)

Mean (Range)

Predictor

      

A-levels

22

920 (R 55; 3,096)

.180 (R .041; .306 )

.231 (R .053; .415)

.634 (R .147; .943)

61 (R 8; 245)

AS-levels

1

1911

.182

.228

.458

266

GCSEs/O-levels

20

849 (R 52; 2,657)

.162 (R -.026 ; .269)

.176 (R −.045; .273)

.336 (R −.079; .626)

231 (R 16; 1,271)

Highers

1

777

-.001

.076

.107

336

‘HighersPlus’

1

771

.143

.180

.292

265

Advanced Highers

1

735

.345

.358

.506

247

Ed. Attainment GCE

1

2768

.312

.347

.923

745

Ed. Attainment SQA

1

722

.419

.425

.623

317

Aptitude tests (AH5, UKCAT)

9

934 (R 156; 4,841)

.132 (R .037; .276)

.152 (R .045; .260)

.228 (R.062; .449)

445 (R 62; 2,720)

Outcome

      

BMS first year

15

1521 (R 542; 4,841)

.207 (R −.001 ; .419)

.248 (R .076; .425)

.498 (R .107; .943)

517 (R 24; 2,720)

BMS overall

9

1152 (R 502; 3,096)

.174 (R .037; .282)

.215 (R .053; .394)

.491 (R .065; .903)

203 (R 10; 513)

Finals

11

786 (R 314; 2,413)

.187 (R .051 ; .306)

.222 (R .080; .328)

.488 (R.097; .871)

136 (R 10; 389)

MRCP(UK) Pt1

5

492 (R 202; 957)

.192 (R .126; .245)

.221 (R .143; .308)

.456 (R .168; .692)

107 (R 32; 209)

MRCP(UK) Pt2

5

363 (R 98; 753)

.205 (R .085 ; .299)

.227 (R .071; .217)

.442 (R .326; .743)

44 (R 12; 80)

MRCP(UK) Clinical

5

277 (R 52; 597)

.141 (R .058; .236)

.144 (R .071; .217)

.317 (R .147; .627)

40 (R 10; 81)

On Specialist Register

9

984 (R 393; 2,664)

.084 (R -.026; .419)

.113 (R −.045; .249)

.367 (R −.079; .803)

126 (R 8; 2,720)

Cohort

      

Westminster

4

470 (R 454; 486)

.169 (R .146 ; .190)

.226 (R .188; .249)

.565 (R .386; .803)

49 (R 17; 96)

1980 cohort

8

449 (R 314; 562)

.164 (R −.026; .306)

.192 (R −.046; .315)

.457 (R −.079; .864)

73 (R 14; 178)

1985 cohort

6

643 (R 347; 851)

.164 (R .066 ; .240)

.187 (R .102; .253)

.566 (R .214; .903)

98 (R 8; 289)

1990 cohort

18

1234 (R 156; 3,096)

.133 (R .037; .276)

.157 (R .045; .280)

.357 (R .062; .692)

190 (R 42; 513)

UCLMS cohorts

12

362 (R 52; 668)

.218 (R .058; .299)

.261 (R .071; .415)

.475 (R .147; .743)

96 (R 10; 346)

UKCAT-12

9

1,938 (R 722; 4841)

.198 (R −.001; .419)

.232 (R .076; .425)

.467 (R .107; .943)

688 (R 25; 2720)

All construct validities

57

935 (SD = 956) (R 52; 4841)

.171 (SD = .092) (R .−026; .419)

.203 (SD = .101) (R −.045; .425)

. 450 (SD = .248) (R −.079; .943)

213 (SD = 396) (R 8; 2720)

  1. *The corrected correlation takes into account both right-censoring and the use of ordinal values (see statistical appendix for details).
  2. Note that figures in brackets are ranges (indicated by R:) and are not confidence intervals.