Skip to main content

Table 3 Results from "Baby Walker" trial.

From: Spatial effects should be allowed for in primary care and other community-based cluster RCTS

    

AIC

Model

b(Treat)

se

AIC

Difference

Ignore clustering N/S gradient only

-0.660

0.141

1245.57

0.00

Ignore clustering + N/S + E/W

-0.651

0.141

1247.09

1.53

Rand Pract + N/S

-0.655

0.148

1247.48

1.91

Rand Pract + N/S + E/W

-0.646

0.148

1249.03

3.46

Ignore clusteringa

-0.618

0.138

1254.50

8.93

Robust se (pract as cluster)

-0.618

0.164

1254.50

8.93

Rand Practb

-0.629

0.167

1255.29

9.73

Rand Pract + MM

-0.609

0.158

1256.92

11.35

  1. LR test for random effect, model (a) vs model (b): chisquared 2(01) = 0.06 P = 0.4017