Bias in main effects for logistic model with random intercept and small cluster sizes

[maj684]

@martinmodrak

ly on fixed values but instead draw the true parameters from the priors in your model, as in Simulation-based calibration.

Many thanks Martin. I am proceeding down the path of examining the prior predictive distribution. For the case of a logistic regression with varying intercept and binary response, I haven’t seen an example anywhere, so I will post the model when I get back to it. I can see that it could be a little bit challenging constructing and visualising the results and that it would probably be necessary to do this on the basic of a summary statistic.