Double robustness

Daniel, Rhian M. ORCID: https://orcid.org/0000-0001-5649-9320 2014. Double robustness. Wiley StatsRef: Statistics Reference Online, John Wiley & Sons, Ltd, (10.1002/9781118445112.stat08068)

Full text not available from this repository.

Abstract

Double robustness is a prevalent topic in current statistical thinking, especially in causal inference and missing data methods. The most popular class of doubly robust (DR) estimators – known as locally efficient augmented inverse probability weighted (AIPW) estimators – possess additional properties, linked to but distinct from double robustness, which lead to tractable statistical inferences and other attractive features. In this article, we start by briefly introducing the concepts of nuisance models and multiple (double) robustness, before turning to a particular setting for a more formal introduction, namely that of estimating a population mean from incomplete data. For this setting, we describe the DR AIPW estimator and discuss the other properties alluded to above as well as double robustness itself. We then show how this setting is more general than it might seem at first sight, before going on to discuss many extensions, both to more complex settings and to recently developed improved DR estimation strategies.

Item Type:	Book Section
Date Type:	Publication
Status:	Published
Schools:	Medicine
Publisher:	John Wiley & Sons, Ltd
ISBN:	9781118445112
Last Modified:	26 Apr 2024 10:00
URI:	https://orca.cardiff.ac.uk/id/eprint/108140

Actions (repository staff only)

Edit Item