Semiparametric Inference
Overview
Semiparametric inference studies how to estimate and quantify uncertainty for targets such as treatment effects or policy values when parts of the data-generating process are modeled flexibly. Common approaches use influence functions and tangent-space geometry to derive efficiency bounds, then apply orthogonal scores, debiased one-step estimators, or targeted maximum likelihood with cross-fitting. These methods are used with machine-learned nuisance functions in causal studies, survival analysis, missing-data problems, and adaptive experiments. A central challenge is maintaining valid confidence intervals under nuisance-estimation error, weak overlap, dependence, or adaptive sampling without imposing assumptions so strong that useful efficiency gains disappear.
Latest papers
34 papers3 papers in the last four weeks, level with the four weeks before. 0.0% of all new papers.