Strati cation and weighting via the propensity score in estimation of causal treatment e ects : a comparative study

By J. Lunceford, Marie Davidian. 2004

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Balanced Policy Evaluation and Learning (2017, background): “…T ( X )] , a normalized IPW (NIPW) estimator is given by normalizing the weights so they sum to n , a common practice in causal eﬀect estimation [2, 24]: Both IPW approaches are subject to considerable variance because the plugged-in propensities are in the denominator so that small errors can…”
Doubly Robust Policy Evaluation and Optimization (2014, methodology): “Dou-bly robust (or doubly protected) estimation (Cas-sel, S¨arndal and Wretman, 1976; Robins, Rot-nitzky and Zhao, 1994; Robins and Rotnitzky, 1995; Lunceford and Davidian, 2004; Kang and Schafer, 2007) is a statistical approach for estimation from incomplete data with an important property: if…”
The Finite Sample Performance of Inference Methods for Propensity Score Matching and Weighting Estimators (2018, unknown): “…number of simulation studies has investigated and compared the finite sample behavior of various point estimators, see Frölich (2004), Zhao (2004), Lunceford and Davidian (2004), Busso, DiNardo, and McCrary (2014), Huber, Lechner, and Wunsch (2013), and Frölich, Huber, and Wiesenfarth (2017).”
Stable Prediction across Unknown Environments (2018, unknown): “They either employ propensity scores (Rosenbaum and Rubin (1983); Lunceford and Davidian (2004); Austin (2011); Kuang et al. (2017b, 2016)), or optimize balancing weights directly (Hainmueller (2012); Zubizarreta (2015); Athey et al. (2016); Kuang et al. (2017a)).”
Evaluating online ad campaigns in a pipeline: causal models at scale (2010, methodology): “P The theory in [16], [13], and [8] shows that the minimum asymptotic variance is achieved by modifying ˆ∆ ∗ by the errors Z i − ˆ p i as follows: where Usually, ˆ m 1 ( X ) is obtained by regressing the observed Y on X for the exposed, and ˆ m 0 ( X ) is obtained by regressing Y on X for the…”