That works - thanks! I’m really sorry though the problem is still there… I had a (4 x 4) matrix of success rates and my hacky solution was to convert this into a (4 x 4) boolean “stacked up” 1000 times so that the mean of 1000 for each cell equals the success rate I had to begin with… for each cell in (4 x 4) I have different numbers of trials (not 1000 for all of them) but I’m not sure if this is a good way. It’s still not converging to reproduce the observed data although the order of magnitude is better than before. I get "The chain reached maximum tree depth", "The number of effective samples is smaller than 200/25%" warning every time. My target_accept is 0.95. If there is a way to troubleshoot this problem I’d love to know… thank you so much.