Table 5:

Simulation results for the effect of treatment with 12 clusters, based on an analysis model that does not control for cluster-level covariates

Covariate SD	Degree Confounding	%Bias	Var	MSE	Cov	Power	CI Width	Type 1 Error	Balance
Covariate Constrained Randomization
	β = 0.0	−0.44	0.76	0.76	0.95	1.00	3.81	0.05	1.27
σx = 0.5	β = 0.5	−0.51	0.78	0.79	0.95	1.00	3.98	0.04	1.27
	β = 1.0	−0.57	0.86	0.86	0.96	0.99	4.46	0.03	1.27
	β = 0.0	−0.45	0.76	0.76	0.95	1.00	3.81	0.05	1.27
σx = 1	β = 0.5	−0.57	0.86	0.86	0.96	0.99	4.46	0.03	1.27
	β = 1.0	−0.70	1.18	1.18	0.98	0.93	6.03	0.02	1.27
	β = 0.0	−0.44	0.76	0.76	0.95	1.00	3.81	0.05	1.27
σx = 2	β = 0.5	−0.71	1.18	1.18	0.98	0.93	6.03	0.02	1.27
	β = 1.0	−0.97	2.44	2.44	0.99	0.49	10.11	0.01	1.27
Simple Randomization
	β = 0.0	−0.02	0.75	0.75	0.95	1.00	3.80	0.05	2.98
σx = 0.5	β = 0.5	−0.06	0.81	0.81	0.95	1.00	3.96	0.05	2.98
	β = 1.0	−0.09	0.99	0.99	0.95	0.99	4.40	0.05	2.98
	β = 0.0	−0.02	0.75	0.75	0.95	1.00	3.80	0.05	2.98
σx = 1	β = 0.5	−0.09	0.99	0.99	0.95	0.99	4.40	0.05	2.98
	β = 1.0	−0.16	1.72	1.72	0.95	0.92	5.83	0.05	2.98
	β = 0.0	−0.02	0.75	0.75	0.95	1.00	3.80	0.05	2.98
σx = 2	β = 0.5	−0.16	1.72	1.72	0.95	0.92	5.83	0.05	2.98
	β = 1.0	−0.30	4.70	4.70	0.95	0.53	9.60	0.05	2.98