Use a logit proposal function into the MCMC model

[bakaburg1]

At the moment the MCMC proposal distribution is built on the raw prevalence and values outside the [0,1] interval are discarded by manually setting the score to -Inf.

Instead the parameter should be evaluaed after transforming it into the logit space in order to have a Normal distributed esploration but evaluation in the costrained transformed space.

[bakaburg1]

Unfortunately this is less simple than expected, since it breaks proposal symmetry apparently. So there would be more steps to take, to make it work, unless the exploration is directly made in the logit space, but I don't think I can make this in MCMCpack