Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A general maximum likelihood analysis of variance components in generalized linear models.

M Aitkin1

  • 1Department of Statistics, University of Newcastle, UK. Murray.Aitkin@newcastle.ac.uk

Biometrics
|April 25, 2001
PubMed
Summary

This study introduces an EM algorithm for nonparametric maximum likelihood estimation in generalized linear models. This method offers a robust alternative for analyzing complex variance component structures and estimating distributions without parametric assumptions.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Stillbirths among offspring of male radiation workers at Sellafield nuclear reprocessing plant.

Lancet (London, England)·1999
Same author

Meta-analysis by random effect modelling in generalized linear models.

Statistics in medicine·1999
Same author

Detailed modelling of caffeine metabolism and examination of the CYP1A2 gene: lack of a polymorphism in CYP1A2 in Caucasians.

Pharmacogenetics·1999
Same author

Ambulatory blood pressure in schoolchildren.

Archives of disease in childhood·1999
Same author

Nurse education--is it limiting us?

The Queensland nurse·1991
Same author

A method of fitting the gravity model based on the Poisson distribution.

Journal of regional science·1982

Area of Science:

  • Statistics
  • Statistical Modeling
  • Computational Statistics

Background:

  • Generalized linear models (GLMs) with variance components present analytical challenges.
  • Existing methods like MQL, PQL, and GEE offer approximations but may lack full nonparametric estimation capabilities.
  • Parametric assumptions for mixing distributions can impact the sensitivity of maximum likelihood (ML) estimates.

Purpose of the Study:

  • To present an Expectation-Maximization (EM) algorithm for nonparametric maximum likelihood (NPML) estimation in GLMs with variance component structures.
  • To provide a flexible alternative to approximate methods and GEE analyses.
  • To enable robust estimation of the mixing distribution without specifying a parametric form.

Main Methods:

  • The study details an EM algorithm, generalizing existing methods for overdispersion.

Related Experiment Videos

  • The algorithm is initially based on Gaussian quadrature with a normal mixing distribution.
  • It is adapted for completely unknown mixing distributions, facilitating fully nonparametric ML estimation.

    • Main Results:

    • The EM algorithm provides a straightforward method for fully nonparametric ML estimation of the mixing distribution.
    • This approach enhances the reliability of GLM parameter estimates by avoiding sensitivity to parametric form specification.
    • The nonparametric analysis extends to general random parameter models, allowing NPML estimation of joint distributions.

    Conclusions:

    • The developed EM algorithm offers a powerful tool for nonparametric ML estimation in complex statistical models.
    • It provides computational savings and improved accuracy compared to methods relying on parametric assumptions or numerical integration.
    • The method is applicable to various models, including variance component, longitudinal, and small-area estimation problems.