Semiparametric Maximum Likelihood Variance Component Estimation Using Mixture Moment Structure Models
- 1 June 2006
- journal article
- conference paper
- Published by Cambridge University Press (CUP) in Twin Research and Human Genetics
- Vol. 9 (3) , 360-366
- https://doi.org/10.1375/183242706777591245
Abstract
Nonnormal phenotypic distributions introduce significant problems in the estimation and selection of genetic models. Here, a semiparametric maximum likelihood approach to analyzing non-normal phenotypes is described. In this approach, distributions are explicitly modeled together with genetic and environmental effects. Distributional parameters are introduced through mixture constraints, where the distribution of effects are discretized and freely estimated rather than assumed to be normal. Semiparametric maximum likelihood estimation can be used with a variety of genetic models, can be extended to a variety of pedigree structures, and has various advantages over other approaches to modeling nonnormal data.This publication has 4 references indexed in Scilit:
- Kolmogorov's Structure Functions and Model SelectionIEEE Transactions on Information Theory, 2004
- Information Theory and an Extension of the Maximum Likelihood PrinciplePublished by Springer Nature ,1998
- Practical Bayesian density estimation using mixtures of normalsJournal of the American Statistical Association, 1997
- Latent Variables Analysis: Applications for Developmental Research.Journal of the American Statistical Association, 1995