Bayesian Causal Inference in Sequentially Randomized Experiments with Noncompliance

TL;DR

This paper introduces a Bayesian framework for causal inference in sequential experiments with non-compliance, addressing a gap in existing methods that typically assume fixed treatments.

Contribution

It proposes a novel latent mixture Bayesian approach to estimate treatment effects in sequential experiments with non-compliance, extending beyond traditional methods.

Findings

Effective estimation of treatment effects in complex sequential settings.

Addresses non-compliance issues often overlooked in sequential experiments.

Provides a new methodological framework for causal inference.

Abstract

Scientific researchers utilize randomized experiments to draw casual statements. Most early studies as well as current work on experiments with sequential intervention decisions has been focusing on estimating the causal effects among sequential treatments, ignoring the non-compliance issues that experimental units might not be compliant with the treatment assignments that they were originally allocated. A series of methodologies have been developed to address the non-compliance issues in randomized experiments with time-fixed treatment. However, to our best knowledge, there is little literature studies on the non-compliance issues in sequential experiments settings. In this paper, we go beyond the traditional methods using per-protocol, as-treated, or intention-to-treat analysis and propose a latent mixture Bayesian framework to estimate the sample-average treatment effect in sequential experiment having non-compliance concerns.