TY - JOUR

T1 - Quanti cation of empirical determinacy: the impact of likelihood weighting on posterior location and spread in Bayesian meta-analysis estimated with JAGS and INLA

AU - Hunanyan, Sona

AU - Rue, Haavard

AU - Plummer, Martyn

AU - Roos, Malgorzata

N1 - KAUST Repository Item: Exported on 2022-09-14
Acknowledgements: We thank both reviewers and the associate editor for their thoughtful comments and constructive suggestions. Support by the Swiss National Science Foundation (no. 175933) granted to Malgorzata Roos is gratefully acknowledged.

PY - 2021/9/24

Y1 - 2021/9/24

N2 - The popular Bayesian meta-analysis expressed by the normal-normal hierarchical model synthesizes knowledge from several studies and is highly rele-vant in practice. The normal-normal hierarchical model is the simplest Bayesian hierarchical model, but illustrates problems typical in more complex Bayesian hierarchical models. Until now, it has been unclear to what extent the data deter- mines the marginal posterior distributions of the parameters in the normal-normal hierarchical model. To address this issue we computed the second derivative of the Bhattacharyya coe cient with respect to the weighted likelihood. This quan-tity, which we de ne as the total empirical determinacy (TED), can be written as the sum of two terms: the empirical determinacy of location (EDL), and the empirical determinacy of spread (EDS).
We implemented this method in the R package ed4bhm and considered two case studies and one simulation study. We quanti ed TED, EDL and EDS under di erent modeling conditions such as model parametrization, the primary outcome, and the prior. This clari es to what extent
the location and spread of the marginal posterior distributions of the parameters are determined by the data. Although these investigations focused on Bayesian normal-normal hierarchical model, the method proposed is applicable more gen-erally to complex Bayesian hierarchical models.

AB - The popular Bayesian meta-analysis expressed by the normal-normal hierarchical model synthesizes knowledge from several studies and is highly rele-vant in practice. The normal-normal hierarchical model is the simplest Bayesian hierarchical model, but illustrates problems typical in more complex Bayesian hierarchical models. Until now, it has been unclear to what extent the data deter- mines the marginal posterior distributions of the parameters in the normal-normal hierarchical model. To address this issue we computed the second derivative of the Bhattacharyya coe cient with respect to the weighted likelihood. This quan-tity, which we de ne as the total empirical determinacy (TED), can be written as the sum of two terms: the empirical determinacy of location (EDL), and the empirical determinacy of spread (EDS).
We implemented this method in the R package ed4bhm and considered two case studies and one simulation study. We quanti ed TED, EDL and EDS under di erent modeling conditions such as model parametrization, the primary outcome, and the prior. This clari es to what extent
the location and spread of the marginal posterior distributions of the parameters are determined by the data. Although these investigations focused on Bayesian normal-normal hierarchical model, the method proposed is applicable more gen-erally to complex Bayesian hierarchical models.

UR - http://hdl.handle.net/10754/672029

UR - https://arxiv.org/pdf/2109.11870.pdf

M3 - Article

JO - Bayesian Analysis

JF - Bayesian Analysis

ER -