Towards causally interpretable meta-analysis: transporting inferences from multiple studies to a target population
share
We take steps towards causally interpretable meta-analysis by describing conditions under which we can transport causal inferences from a collection of randomized trials to a new target population. We discuss the conditions that allow the identification of causal quantities in the target population and provide identification results for potential (counterfactual) outcome means and average treatment effects. Our results highlight the importance of accounting for variation in the treatment assignment mechanisms across the randomized trials when transporting inferences. Last, we propose estimators of the potential outcome means that rely on different working models and provide code for their implementation in statistical software.
READ FULL TEXT