I want to set up a scenario where errors are correlated. In the first setting, a betaregression should be fit, the second scenario includes a conditional logit model.