A Note on "Generalizability of Study Results"
University of California Los Angeles Los Angeles United States
Pagination or Media Count:
We compare two approaches to the analysis of generalizability 1potential outcomes and 2 model-based data fusion. We show that, while the inferential power of the former is curtailed by a priori assumptions of conditional exchangeability, the latter unveils and leverages the full range of opportunities that are licensed by both the available data and ones model of reality, regardless of whether the model supports the assumption of conditional exchangeability.