Skip to main content
. Author manuscript; available in PMC: 2015 Jul 1.
Published in final edited form as: Comput Stat Data Anal. 2014 Jul;75:53–65. doi: 10.1016/j.csda.2014.02.009

Table 5.

Simulated lung cancer data with weak and strong random effects

Effect Parameter True Value B1 B2 B3 Estimate 95% CI Pr(Inclusion)
β 1 −0.5 −0.070 −0.010 (−0.082, 0.000) 0.168
β 2 0.3 0.239 0.332 0.165 (0.000, 0.494) 0.609
β 3 0.5 0.312 0.408 0.334 0.323 (0.000, 0.535) 0.907
β 4 −0.8 −0.575 −0.546 −0.510 −0.554 (−0.788, −0.373) 1.000
β 5 0.3 0.320 0.348 0.301 0.335 (0.201, 0.467) 1.000
Σ 00 0 0.000 (0.000, 0.000) 0.022
Weak Σ 11 0.1 0.000 (0.000, 0.000) 0.001
Σ 22 0.2 0.320 0.367 0.322 0.343 (0.130, 0.658) 1.000
Σ 33 0 0.000 (0.000, 0.000) 0.005
Σ 44 0 0.000 (0.000, 0.000) 0.009
Σ 55 0 0.001 (0.000, 0.000) 0.023
Pr(Selection) 0.380 0.337 0.086
β 1 1 1.155 1.330 1.253 1.200 (0.921, 1.504) 1.000
β 2 −2 −2.062 −2.100 −2.099 −2.074 (−2.297, −1.815) 1.000
β 3 0 0.021 0.000 (0.000, 0.000) 0.038
β 4 −1 −0.914 −1.028 −0.928 −0.943 (−1.117, −0.829) 1.000
β 5 0 0.169 0.039 (0.000, 0.249) 0.235
Σ 00 0 0.000 (0.000, 0.000) 0.001
Strong Σ 11 16/15 0.619 0.900 0.677 0.688 (0.328, 1.346) 1.000
Σ 22 32/15 1.548 1.341 1.550 1.492 (0.707, 2.566) 1.000
Σ 33 0 0.003 (0.000, 0.000) 0.027
Σ 44 0 0.000 (0.000, 0.000) 0.001
Σ 55 0 0.000 (0.000, 0.000) 0.001
Pr(Selection) 0.712 0.223 0.025