Skip to main content
. Author manuscript; available in PMC: 2018 Jan 22.
Published in final edited form as: Int J Biostat. 2016 Nov 1;12(2):/j/ijb.2016.12.issue-2/ijb-2015-0034/ijb-2015-0034.xml. doi: 10.1515/ijb-2015-0034

Table 4b.

Results for various IPW and TMLE approaches in the plasmode simulation study (with truncation in the propensity score at 0.025 and 0.975)

Approach Outcome Model Treatment Model Estimate SE Bias RMSE Finite fitted values*
IPW1 A NULL −1.19 0.04 0.86 0.86 100%
IPW2 A C1 −0.84 0.04 0.51 0.51 100%
IPW3 A C2 −0.59 0.07 0.26 0.27 100%
IPW4 A C3 −0.45 0.08 0.11 0.13 100%
TMLE1 A NULL −1.19 0.04 0.86 0.86 100%
TMLE2 A C1 −0.84 0.04 0.51 0.51 100%
TMLE3 A C2 −0.54 0.07 0.21 0.22 100%
TMLE4 A C3 −0.31 0.09 0.03 0.09 100%
TMLE5 A,C1 NULL −0.84 0.04 0.51 0.51 100%
TMLE6 A,C1 C1 −0.84 0.04 0.51 0.51 100%
TMLE7 A, C1 C2 −0.55 0.07 0.22 0.23 100%
TMLE8 A, C1 C3 −0.31 0.09 0.03 0.09 100%
TMLE9 A, C2 NULL −0.58 0.05 0.25 0.26 100%
TMLE10 A, C2 C1 −0.58 0.05 0.25 0.25 100%
TMLE11 A,C2 C2 −0.56 0.07 0.23 0.24 100%
TMLE12 A, C2 C3 −0.33 0.08 0.01 0.08 100%
TMLE13 A, C3 NULL −0.34 0.05 0.005 0.06 100%
TMLE14 A, C3 C1 −0.34 0.05 0.002 0.06 100%
TMLE15 A, C3 C2 −0.33 0.08 0.000 0.06 100%
TMLE16 A, C3 C3 −0.34 0.08 0.002 0.08 100%
*

Refers to settings in which the first step of the TMLE returned non-infinite predictions.