Skip to main content

Advertisement

Table 2 Parameter estimates between standard analysis and pooled analysis with different pool sizes for a binary outcome with a matched design. See Results section for detailed simulation setting. In addition, model-based SE (ModelSE), the Monte Carlo Standard Error (EmpSE) and coverage (nominal: 0:95) are also shown

From: Addressing data privacy in matched studies via virtual pooling

Parameters Unpooled Pooled
g = 4 g = 6 g = 10
β = 0.3
Estimate 0.301 0.303 0.307 0.336
EmpSE 0.014 0.022 0.028 0.077
ModelSE 0.014 0.022 0.029 0.060
Coverage 0.958 0.956 0.964 0.964
γ = 0.2
Estimate 0.202 0.204 0.207 0.234
EmpSE 0.077 0.101 0.128 0.219
ModelSE 0.076 0.100 0.122 0.199
Coverage 0.954 0.956 0.944 0.952
δ= 0.15
Estimate 0.149 0.150 0.150 0.165
EmpSE 0.037 0.049 0.062 0.104
ModelSE 0.037 0.049 0.060 0.098
Coverage 0.952 0.958 0.952 0.960
ω = 0.09
Estimate 0.088 0.088 0.091 0.104
EmpSE 0.050 0.067 0.080 0.134
ModelSE 0.049 0.063 0.076 0.123
Coverage 0.964 0.936 0.942 0.954
ϑ = 0.05
Estimate 0.050 0.051 0.051 0.054
EmpSE 0.013 0.018 0.023 0.039
ModelSE 0.013 0.018 0.022 0.037
Coverage 0.954 0.944 0.952 0.944