Double Robustness in Estimation of Causal Treatment Effects
Double Robustness in Estimation of Causal Treatment Effects
Double Robustness in Estimation of Causal Treatment Effects
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
6. Some Interest<strong>in</strong>g Issues<br />
Issue 1: How do we get standard errors for � ∆DR?<br />
• One way: use standard large sample theory , which leads to the<br />
so-called sandwich estimator<br />
�<br />
�<br />
� n�<br />
�n−2 �Ii = ZiYi<br />
e(Xi, � β) − {Zi − e(Xi, � β)}<br />
e(Xi, � β)<br />
m1(Xi, �α1)<br />
�<br />
(1 − Zi)Yi<br />
−<br />
1 − e(Xi, � β) + {Zi − e(Xi, � β)}<br />
1 − e(Xi, � β) m0(Xi,<br />
�<br />
�α0) − � ∆DR<br />
i=1<br />
• Use the bootstrap (i.e., resample B data sets <strong>of</strong> size n)<br />
<strong>Double</strong> <strong>Robustness</strong>, EPID 369, Spr<strong>in</strong>g 2007 37<br />
�I 2 i