Diagonal reference models are especially suited for the estimation of effects of movements across levels of categorical variables like education or social class. In social stratification, it enables us to estimate the weight of origen and destination. Their use is straightforward with DRM in Stata and the function Dref of gnm in R. However, I am working with a dataset with 30 countries and I would like to model those weights as random effects. I haven't find a multilevel extension of DRM or a workaround. Any idea?