Skip to main content

Table 1 Comparison of student and school-level characteristics between samples

From: Sources of potential bias when combining routine data linkage and a national survey of secondary school-aged children: a record linkage study

  

School consented to data linkage question

 
 

Full SHW survey sample (n = 103,971; schools = 193)

Data linkage sub-sample (n = 18,956; schools = 39)

Data linkage control sample (n = 8374; schools = 22)a

Regression-based P-value (data linkage vs. non-data linkage sample)b

 

n (%)

n (%)

n (%)

 

Student-level

 Gender

  Male

50,452 (48.5)

9545 (50.4)

3788 (45.2)

0.078

  Female

51,458 (49.5)

9064 (47.8)

4421 (52.8)

0.110

  Did not want to answer

2061 (2.0)

347 (1.8)

165 (2.0)

0.283

 School grade

  Year 7

22,634 (21.8)

3787 (20.0)

2064 (24.7)

0.005

  Year 8

22,421 (21.6)

4122 (21.8)

1998 (23.9)

0.755

  Year 9

22,208 (21.4)

4347 (22.9)

1628 (19.4)

0.023

  Year 10

19,704 (19.0)

3500 (18.5)

1417 (16.9)

0.580

  Year 11

17,004 (16.4)

3200 (16.9)

1267 (15.1)

0.552

 Weekly cigarette smoker

  No

96,170 (96.6)

17,656 (96.9)

7745 (96.4)

–

  Yes

3422 (3.4)

571 (3.1)

291 (3.6)

0.168

 cannabis use

  No

94,156 (95.6)

17,799 (96.2)

7604 (95.5)

–

  Yes

4305 (4.4)

699 (3.8)

357 (4.5)

0.061

 Ever truant

  No

68,071 (73.9)

13,666 (75.4)

5457 (73.2)

–

  Yes

23,989 (26.1)

4455 (24.6)

1995 (26.8)

0.046

 Ever excluded

  No

86,614 (92.2)

17,216 (93.0)

6947 (91.1)

–

  Yes

7370 (7.8)

1304 (7.0)

675 (8.9)

0.091

SWEMWBS (sd.)

21.955 (4.51)

22.094 (4.33)

21.867 (4.40)

0.048

FAS (sd.)

9.28 (2.34)

9.333 (2.32)

9.166 (2.34)

0.557

School-level

 Mean FAS (sd.)

9.309 (0.63)

9.353 (0.50)

9.192 (0.653)

0.595

  1. Notes: SWEMWBS, Short Warwick-Edinburgh Mental Wellbeing Scale; FAS, Family Affluence Scale
  2. aThe data linkage control sample refers to the 22/64 schools which consented to data linkage, but which were not randomly assigned to the data linkage sample following the 2:1 randomization process (see Fig. 1)
  3. bVariation in student socio-demographics and risk behaviours between samples were explored using binary and linear regression models (depending on the outcome variable) with adjustment for clustering by school. As the data linkage sample was nested within the full SHW survey sample, reported p-values were drawn from models comparing students attending data linkage (n = 39) versus non-data linkage (n = 154) schools