When: 29 April, 1-2 pm
Where: This seminar is given online. E-mail Dan Hedlin if you want to attend.

Confounder adjustment is a key objective when doing causal analyses with observational data. Properties of semi-parametric estimators using fitted propensity scores, conditional outcomes and a combination thereof with different degrees of flexibility of parametric models have been in focus in the causal literature. Early guidance to model selection suggested that model specification, fitting and balance checking could be performed in an iterative procedure. This was followed by proposals of, now standard, doubly robust AIPW estimators that fit parametric models for the propensity score and conditional outcomes given covariates.

In more recent years, a novel class of weighting estimators have been proposed that directly aim at incorporating covariate balance in the estimation process through calibration/entropy maximization.

Since covariate balance is not a sufficient condition for identification of the true propensity score the general calibration estimator, using finite constraints, has an asymptotic error which will depend on the covariance of the error of an implicit propensity score fit and the conditional outcomes. Here, alike properties of multiple robustness as for the AIPW estimators are inherited in the estimation procedure.

In this talk we study weighting estimators within the more recent calibration/entropy balancing proposals (Chan et al. 2016) and other alternatives to propensity score estimation such as RKHS (Wong and Chan, 2018) and CBPS (Fan et al. 2018). The study includes simulations comparing parametric and nonparametric approaches for estimation of the propensity score. In addition, we evaluate proposed variance estimators from an earlier study of entropy balancing estimators using Kullback-Leibler and quadratic Rényi divergence (Källberg and Waernbaum, 2019).

Joint with David Källberg and Emma Persson, Umeå University.