Skip to main content
. 2017 Dec 4;17:160. doi: 10.1186/s12874-017-0434-1

Table 1.

Parameters of marginal true effect for the 15 simulated sub-cases

Situations Case 1
α1 = 0.004, ω1 = 0.004
Strong confounding for A2
Case 2
α1 = 0.004, ω1 = 0.001
Weak confounding for A2
Case 3
α1 = 0.004, ω1 = 0.000
No confounding for A2
A β1 = 0.5, β2 = 0.5, β3 = 0 β1 = 0.5, β2 = 0.5, β3 = 0 β1 = 0.5, β2 = 0.5, β3 = 0
B β1 = 0.5, β2 = 0.5, β3 = 0.5 β1 = 0.5, β2 = 0.5, β3 = 0.5 β1 = 0.5, β2 = 0.5, β3 = 0.5
C β1 = 0, β2 = 0.5, β3 = 0 β1 = 0, β2 = 0.5, β3 = 0 β1 = 0, β2 = 0.5, β3 = 0
D β1 = 0, β2 = 0, β3 = 0.5 β1 = 0, β2 = 0, β3 = 0.5 β1 = 0, β2 = 0, β3 = 0.5
E β1 = 0, β2 = 0, β3 = 0 β1 = 0, β2 = 0, β3 = 0 β1 = 0, β2 = 0, β3 = 0

α1 and ω1 are the coefficients of the time-dependent confounder in the functions of prediction for treatment A1 and A2, respectively