|
Abstract:
|
Understanding the relationship between prenatal mercury exposure and neurodevelopment in children is of great interest to many practitioners. Typically, analyses rely on separate models fit to each outcome. If the effect of exposure is very similar across outcomes, separate models lack power to detect a common exposure effect. Furthermore, the outcomes cluster into broad domains and domain-specific effects are also of interest. We fit a Bayesian model which allows the mercury effect to vary across outcomes, while allowing for shrinkage of these effects within domains, and to a lesser extent between domains. We will discuss the benefits and challenges of fitting this model within a Bayesian framework, and present results of the model applied to multiple outcomes measured in children at 9 years of age in the Seychelles.
|
