Bayesian Causal Inference with Bipartite Record Linkage

In some scenarios, the observational data needed for causal inferences are spread over two files. particular, we consider scenarios where one file includes covariates and treatment measured on a set of individuals, second responses another, partially overlapping individuals. absence error-free direct identifiers like social security numbers, straightforward merging separate files is not feasible, so that records must be linked using error-prone variables such as names, birth dates, demographic characteristics. Typical practice in situations generally follows two-stage procedure: first link probabilistic linkage technique, then make with dataset. This does propagate uncertainty due to imperfect linkages inference, nor it leverage relationships among study improve quality linkages. We propose joint model simultaneous Bayesian inference effects addresses these deficiencies. Using simulation studies theoretical arguments, show can accuracy estimated effects, well record linkages, compared modeling option. illustrate constructed debit card possession household spending.