We make a methodological contribution to the latent class literature by re-examining censored variable analysis within a panel data context. Specifically, we extend the standard latent class tobit panel approach to include random effects, to allow for heteroskedasticity and to incorporate the...