Skip to main content

Table 2 Summary of the factorial design of the simulation study for the primary objective

From: Mind the gap: covariate constrained randomisation can protect against substantial power loss in parallel cluster randomised trials

Parameter Values
Number of clusters in each treatment arm (K) 5, 9, 13
Number of observations per cluster (M) 300
Intra-cluster correlation (ρ) 0.001, 0.01, 0.05, 0.1
Standardised treatment effect (θ) 0 or 0.5 (0.2 (13 clusters per arm),
0.25 (9 clusters per arm, ICC = 0.01 or 0.001))
Number of covariates in the data generation model (L) 4
Magnitude of prognostic effect of covariates 2
Number of covariates balanced in the randomisation (C) 1, 2, 3, 4
Candidate set 10%, 20%, …, 90% worst schemes, 100% (simple randomisation), 10%, …, 90% best schemes
Number of covariates adjusted in the analysis (A) 1, 2, 3, 4