Statistics > Methodology
[Submitted on 25 Aug 2016 (this version), latest version 12 May 2019 (v3)]
Title:Bayesian Inference for Sequential Treatments under Latent Sequential Ignorability
View PDFAbstract:We focus on causal inference for time-varying treatments, i.e., situations in which units are assigned to treatment at multiple time points, aiming to assess the effect of different treatment sequences on some final outcome observed at the end of the study. An assumption which is usually invoked in studies with longitudinal treatment is Sequential Ignorability (SI), which consists in assuming that treatment assignment at each time point is independent of missing past potential outcomes and future potential outcomes given past observed outcomes and covariates. SI may be questionable when participation in the treatment depends on individual choices, and treatment assignment may depend on unobservable quantities associated with future potential outcomes. In order to relax SI, we propose to rely on Principal Stratification, formulating a mild version of SI that we call Latent Sequential Ignorability (LSI), where latent indicates that treatment assignment is conditionally independent on future potential outcomes given past treatments, covariates and principal stratum membership, which is a latent variable since it is defined by the joint value of observed and missing intermediate outcomes. Adopting a Bayesian approach for inference, we discuss and compare SI and LSI, using both theoretical arguments as well as simulation studies in which we investigate the relative performance of the two alternative assumptions when one holds and inference is conducted under both assumptions. Simulation results show that when SI does not hold, inference performed under SI may lead to misleading conclusions. Conversely, LSI generally leads to well-shaped posterior distributions centred around the true values, irrespective of which assumption holds. We apply our framework to an illustrative example based on real data, in which we investigate the effects of interest free loans on firm employment policies.
Submission history
From: Federico Ricciardi [view email][v1] Thu, 25 Aug 2016 14:44:07 UTC (5,873 KB)
[v2] Fri, 16 Jun 2017 11:36:56 UTC (5,877 KB)
[v3] Sun, 12 May 2019 11:39:41 UTC (5,274 KB)
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.