Using Interviewer Random Effects to Calculate Unbiased HIV Prevalence Estimates in the Presence of Non-Response: A Bayesian Approach

Mark E. McGovern, Harvard University
David Canning, Harvard University

Selection bias in HIV prevalence estimates occurs if refusal to test is correlated with HIV status. Interviewer identity is plausibly correlated with consenting to test, but not with HIV status, allowing a Heckman-type correction that produces consistent prevalence estimates. We innovate by adopting an interviewer random effects Bayesian estimator which improves on existing approaches by producing prevalence estimates that are unbiased as well as consistent and which allows the construction of bootstrapped standard errors. The model is used to produce new estimates and confidence intervals for HIV prevalence among men in Zambia and Ghana.

  See paper

Presented in Session 44: Innovative Methods in HIV-related Research