We describe a regression model for multivariate responses of mixed type, where association between outcomes is modeled through dependent, outcome- specific, latent effects. Parametric specifications of this model already exist in the literature; in this paper, we show how model parameters can be estimated in a Finite Mixture (FM) context. While ML estimation, once identifiability has been estab- lished, is quite straightforward, a relevant question arises when dependence should be tested vs independence. We review the performance of LRT and penalized like- lihood criteria to assess the presence of dependence between outcome-specific ran- dom effects. The model behavior is investigated through the analysis of simulated datasets.
Regression models for multivariate mixed responses
Nieddu L;
2009-01-01
Abstract
We describe a regression model for multivariate responses of mixed type, where association between outcomes is modeled through dependent, outcome- specific, latent effects. Parametric specifications of this model already exist in the literature; in this paper, we show how model parameters can be estimated in a Finite Mixture (FM) context. While ML estimation, once identifiability has been estab- lished, is quite straightforward, a relevant question arises when dependence should be tested vs independence. We review the performance of LRT and penalized like- lihood criteria to assess the presence of dependence between outcome-specific ran- dom effects. The model behavior is investigated through the analysis of simulated datasets.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.