Skip to main content
. Author manuscript; available in PMC: 2023 Sep 28.
Published in final edited form as: Biometrics. 2022 Dec 8;79(3):2551–2564. doi: 10.1111/biom.13803

TABLE 3.

Estimation of the average treatment effects with varying types of treatment effect heterogeneity across exposure time in 1000 simulated data sets with continuous outcomes. Each simulated data set represents a stepped-wedge CRT with E = 7 exposure-time periods, T = 8 steps, 14 clusters, and n = 100 individuals per cluster per time periods. The Monte Carlo error associated with 95% coverage for 1000 simulation iterations is 0.7%. We use Δ to denote the average treatment effect, Δ^ to denote its estimate, E^() and SD^() to denote sample average and sample standard deviation across simulated experiments, and SE^() to denote the estimated standard error.

Bootstrap
Model-based
Within cluster
Within cluster period
Model E^(Δ^) E^(σ^α) E^(σ^δ) SD^(Δ^) E^(SE^(Δ^)) E^(SE^(Δ^)) Coverage (%) E^(SE^(Δ^)) Coverage (%)
g(e) = 2, Δ = 2, σα = 0.141, σδ = 0
1 2.000 0.138 - 0.032 0.032 0.032 94.2 0.032 94.4
2 2.000 0.138 - 0.052 0.050 0.048 92.8 0.048 92.5
3 2.000 0.138 - 0.037 0.037 0.036 94.1 0.036 93.8
4 2.001 0.138 - 0.053 0.051 0.049 93.0 0.049 93.7
5 2.000 0.138 0.010 0.033 0.034 0.036 96.4 0.036 96.7
g(e) = 2 + δe with δe ~ 𝓝(0, 22), Δ = 2, σα = 0.141, σδ = 2
1 2.193 0.213 - 0.033 0.055 0.047 0.3 0.032 0.0
2 2.255 0.220 - 0.053 0.084 0.064 0.8 0.048 0.2
3 3.552 0.536 - 0.039 0.058 0.050 0.0 0.038 0.0
4 2.002 0.139 - 0.052 0.051 0.049 92.5 0.049 94.3
5 2.003 0.139 2.001 0.052 0.758 0.049 92.5 0.049 94.2
g(e) = (e − 1.840)/1.080, Δ = 2, σα = 0.141, σδ = 2
1 −1.236 0.992 - 0.034 0.041 0.043 0.0 0.033 0.0
2 2.000 0.139 - 0.051 0.050 0.048 93.5 0.048 92.7
3 −0.444 0.742 - 0.039 0.044 0.046 0.0 0.038 0.0
4 2.000 0.139 - 0.052 0.051 0.049 93.3 0.049 92.6
5 1.997 0.139 1.999 0.052 0.757 0.049 93.2 0.049 92.3
g(e) = −2.536I(e = 1) + 2.756I(e >1), Δ = 2, σα = 0.141, σδ = 2
1 −0.760 0.919 - 0.034 0.058 0.070 0.0 0.033 0.0
2 2.788 0.302 - 0.055 0.080 0.063 0.0 0.052 0.0
3 2.000 0.139 - 0.037 0.037 0.036 93.6 0.036 93.4
4 2.000 0.139 - 0.054 0.051 0.049 91.6 0.049 89.9
5 1.998 0.140 2.000 0.054 0.758 0.049 91.7 0.049 90.6