Bivariate model for a meta analysis of diagnostic test accuracy

[arthur-albuquerque]

Of note, we discuss the model you referenced in this post:

Posterior Predictive Distribution in Bivariate Meta-Analysis

I want to generate the posterior predictive distribution in a bivariate meta-analysis using brms. The model is described as (ref): [image] And here is the posterior predictive distribution: [image] As shown below, one needs to use brms::fcor() to include the within-study covariance-variance matrix in the brms model. However, I have never generated the posterior predictive distribution in a model with fcor(). I tried to apply the brms::posterior_predict() syntax I usually use for univariate…

This model was applied to a context where there were multiple studies with 2 outcomes, both of which were estimated in the same population (in each study). In this case, both within- and between-study correlations have a role. I decided to assume within-study correlation = 0.5, and incorporated this info into the model through a covariance-variance matrix generated by metafor::vcor().

Hope this is helpful.