Nonparametric inverse-probability-weighted estimators based on the highly adaptive lasso.
Nonparametric inverse-probability-weighted estimators based on the highly adaptive lasso.
复制标题
DOI：
10.1111/biom.13719
复制
发表时间：
2023-06
期刊：
影响因子：
1.9
通讯作者：
van der Laan, Mark J.
中科院分区：
文献类型：
--
作者：
Ertefaie, Ashkan;Hejazi, Nima S.;van der Laan, Mark J.
逆概率加权估计量是估计因果效应的最古老且可能最常用的一类程序。通过加权机制调整选择偏差，这些程序通过构建一个伪群体来估计感兴趣的影响，其中选择偏差被消除。尽管它们易于使用，这些估计需要正确的规范的加权机制的模型，是已知的效率低下，并遭受灾难的维度。我们提出了一类非参数逆概率加权估计，其中加权机制是通过高度自适应套索的欠平滑来估计的，这是一种非参数回归函数，被证明以接近n-1/3的速率收敛到真正的加权机制。我们证明了我们的估计是渐近线性的，方差收敛到非参数有效界。与双稳健估计不同，我们的方法既不需要推导有效的影响函数，也不需要指定条件结果模型。我们的理论发展具有广泛的影响，在大型统计模型和各种问题设置的有效的逆概率加权估计的建设。我们评估我们的估计在模拟研究中的实际表现，并证明使用我们提出的方法与数据从一个大规模的流行病学研究。
Inverse probability weighted estimators are the oldest and potentially most commonly used class of procedures for the estimation of causal effects. By adjusting for selection biases via a weighting mechanism, these procedures estimate an effect of interest by constructing a pseudo-population in which selection biases are eliminated. Despite their ease of use, these estimators require the correct specification of a model for the weighting mechanism, are known to be inefficient, and suffer from the curse of dimensionality. We propose a class of nonparametric inverse probability weighted estimators in which the weighting mechanism is estimated via undersmoothing of the highly adaptive lasso, a nonparametric regression function proven to converge at nearly n−1/3-rate to the true weighting mechanism. We demonstrate that our estimators are asymptotically linear with variance converging to the nonparametric efficiency bound. Unlike doubly robust estimators, our procedures require neither derivation of the efficient influence function nor specification of the conditional outcome model. Our theoretical developments have broad implications for the construction of efficient inverse probability weighted estimators in large statistical models and a variety of problem settings. We assess the practical performance of our estimators in simulation studies and demonstrate use of our proposed methodology with data from a large-scale epidemiologic study.
登录
查看更多内容
影响因子：
4.5
作者：
GEMAN, S;HWANG, CR
通讯作者：
HWANG, CR
影响因子：
1.9
作者：
Seaman SR;White IR;Copas AJ;Li L
通讯作者：
Li L
DOI：
10.1111/j.1467-985x.2006.00407.x
发表时间：
2006-01-01
影响因子：
2
作者：
Carpenter, James R.;Kenward, Michael G.;Vansteelandt, Stijn
通讯作者：
Vansteelandt, Stijn
影响因子：
2.7
作者：
Benkeser D;Carone M;Laan MJV;Gilbert PB
通讯作者：
Gilbert PB
影响因子：
6.1
作者：
Hirano, K;Imbens, GW;Ridder, G
通讯作者：
Ridder, G