Skip to main content
. 2013 Sep 20;37(7):658–665. doi: 10.1002/gepi.21758

Table 2.

Results from simulation study with correlated variants

r2 Mean F Method Mean Median SD Mean SE Coverage
0.00 42.6 2SLS 0.195 0.191 0.090 0.090 94.8
IVW 0.195 0.191 0.090 0.082 92.8
Likelihood 0.200 0.196 0.092 0.087 94.1
0.06 47.8 2SLS 0.196 0.193 0.086 0.085 94.5
IVW 0.197 0.194 0.086 0.073 90.3
Likelihood 0.201 0.198 0.087 0.077 92.0
0.13 52.6 2SLS 0.197 0.194 0.080 0.081 95.0
IVW 0.199 0.196 0.080 0.066 89.5
Likelihood 0.202 0.199 0.081 0.070 91.2
0.26 63.3 2SLS 0.197 0.193 0.074 0.073 94.6
IVW 0.199 0.196 0.074 0.055 85.1
Likelihood 0.201 0.198 0.074 0.058 87.1
0.41 74.8 2SLS 0.198 0.196 0.067 0.067 95.0
IVW 0.201 0.199 0.068 0.046 82.1
Likelihood 0.202 0.200 0.068 0.048 84.2

Instrumental variable estimates of causal effect +0.2 from simulated data with correlated variants (correlation measured by r2, the average squared correlation between variants) using individual-level data (two-stage least squares method, 2SLS) and summarized data (inverse-variance weighted, IVW, and likelihood-based methods) with mean F statistic, mean and median estimates across simulations, SD of estimates, mean SE of estimates, coverage (%) of 95% confidence interval