ah, thank you, both of those suggestions are helpful!!
I have been trying to create an ordinal probit model, but the approach I first took had a direted cycle (used the values of the latent variables within each ordinal group as the boundaries for the new estimated partitions/cut points). I am not estimating the cut points from a gamma distribution instead of from the latent variable "w" but it still says I have a "possible directed cycle involving wlo" and I have no idea where it is. model reproduced below: model { for (i in 1:N){ #get your latent variable...