Graphical Models for Clustered Binary and Continuous Responses

Graphical models for clustered data mixed with discrete and continuous responses are developed. Discrete responses are assumed to be regulated by some latent continuous variables and particular link functions are used to describe the regulatory mechanisms. Inferential procedures are constructed using the full-information maximum likelihood estimation and observed/empirical Fisher information matrices. Implementation is carried out by stochastic versions of the generalized EM algorithm. As an illustrative application, clustered data from a developmental toxicity study is re-investigated using the directed graphical model and the proposed algorithms. A new interesting directed association between two mixed outcomes reveals. The proposed methods also apply to cross-sectional data with discrete and continuous responses.

[1]  Louise Ryan,et al.  Bivariate Latent Variable Models for Clustered Discrete and Continuous Outcomes , 1992 .

[2]  J Rochon,et al.  Analyzing bivariate repeated measures for discrete and continuous outcome variables. , 1996, Biometrics.

[3]  Jason Roy,et al.  Scaled marginal models for multiple continuous outcomes. , 2003, Biostatistics.

[4]  James O. Berger,et al.  Bayesian Analysis: A Look at Today and Thoughts of Tomorrow , 2000 .

[5]  Hans-Hermann Bock,et al.  Classification and Related Methods of Data Analysis , 1988 .

[6]  Diana L Miglioretti,et al.  Latent transition regression for mixed outcomes. , 2003, Biometrics.

[7]  James J. Heckman,et al.  A Life Cycle Model of Female Labour Supply , 1980 .

[8]  T. Haavelmo.  The Statistical Implications of a System of Simultaneous Equations , 1943 .

[9]  G. C. Wei,et al.  A Monte Carlo Implementation of the EM Algorithm and the Poor Man's Data Augmentation Algorithms , 1990 .

[10]  É. Moulines,et al.  Convergence of a stochastic approximation version of the EM algorithm , 1999 .

[11]  N. Wermuth,et al.  Graphical Models for Associations between Variables, some of which are Qualitative and some Quantitative , 1989 .

[12]  C J Price,et al.  The developmental toxicity of ethylene glycol in rats and mice. , 1985, Toxicology and applied pharmacology.

[13]  O. Cappé,et al.  Markov Chain Monte Carlo: 10 Years and Still Running! , 2000 .

[14]  A. Kuk,et al.  MAXIMUM LIKELIHOOD ESTIMATION FOR PROBIT-LINEAR MIXED MODELS WITH CORRELATED RANDOM EFFECTS , 1997 .

[15]  A. Zellner.  An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests for Aggregation Bias , 1962 .

[16]  T. Louis.  Finding the Observed Information Matrix When Using the EM Algorithm , 1982 .

[17]  D. Rubin,et al.  Maximum likelihood from incomplete data via the EM - algorithm plus discussions on the paper , 1977 .

[18]  G. McLachlan,et al.  The EM algorithm and extensions , 1996 .

[19]  L Ryan,et al.  A Scaled Linear Mixed Model for Multiple Outcomes , 2000, Biometrics.

[20]  R V Gueorguieva,et al.  Joint analysis of repeatedly observed continuous and ordinal measures of disease severity , 2006, Statistics in medicine.

[21]  D. Dunson,et al.  Bayesian latent variable models for clustered mixed outcomes , 2000 .

[22]  S. Lauritzen,et al.  The TM algorithm for maximising a conditional likelihood function , 2001 .

[23]  M. Pourahmadi,et al.  Bayesian analysis of covariance matrices and dynamic models for longitudinal data , 2002 .

[24]  J. Koster,et al.  Markov properties of nonrecursive causal models , 1996 .

[25]  A. Agresti,et al.  A Correlated Probit Model for Joint Modeling of Clustered Binary and Continuous Responses , 2001 .

[26]  L Ryan,et al.  Multivariate linear mixed models for multiple outcomes. , 1999, Statistics in medicine.

[27]  David B Dunson,et al.  A Bayesian Approach for Joint Modeling of Cluster Size and Subunit‐Specific Outcomes , 2003, Biometrics.

[28]  R. P. McDonald,et al.  Structural Equations with Latent Variables , 1989 .