
Aspects of Bayesian Inference in Skewed Mixed Logistic Regression Models 573
27.4 Monte Carlo Study
One of the conclusions drawn by [19] is that in the logistic regression model
with skew-normal random effects, the skewness parameter is usually not well
estimated. Since the prior distribution of random effects influences the fixed
effect interpretation, our goal would be to consider different prior specifications
for the skewness parameter and evaluate gain, if any, in fixing the skewness
parameter (see also [16]).
We performed a Monte Carlo study where we considered 100 replications
of the mixed logis tic model given in (27.1) assuming the fixed effects β =