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

Likelihood ratio testing of variance components in the linear mixed-effects model using restricted maximum likelihood

C H Morrell1

  • 1Mathematical Sciences Department, Loyola College, Maryland 21210-2699, USA. chm@Loyola.edu

Biometrics
|January 12, 1999
PubMed
Summary

This study compares likelihood ratio tests for linear mixed-effects models. Restricted maximum likelihood (REML) test statistics align well with maximum likelihood (ML) and show rejection proportions closer to the nominal 5% level.

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

Low levels of prostate-specific antigen predict long-term risk of prostate cancer: results from the Baltimore Longitudinal Study of Aging.

Urology·2001
Same author

Impact of donor/recipient traits independent of rejection on long-term renal function.

American journal of kidney diseases : the official journal of the National Kidney Foundation·2001
Same author

Today men with prostate cancer have larger prostates.

Urology·2000
Same author

The effect of gestational parity on FEV1 in a group of healthy volunteer women.

Respiratory medicine·1999
Same author

The relationship of obesity and the development of coronary heart disease to longitudinal changes in systolic blood pressure.

Collegium antropologicum·1999
Same author

Construction of hearing percentiles in women with non-constant variance from the linear mixed-effects model.

Statistics in medicine·1997

Area of Science:

  • Statistics
  • Statistical modeling
  • Mixed-effects models

Background:

  • Linear mixed-effects models are widely used in various scientific fields.
  • Testing random components is crucial for model selection and interpretation.
  • The behavior of likelihood ratio test statistics under different estimation methods requires thorough investigation.

Purpose of the Study:

  • To investigate the distribution of the likelihood ratio test statistic for random components in linear mixed-effects models.
  • To compare the performance of restricted maximum likelihood (REML) with maximum likelihood (ML) for this test statistic.
  • To evaluate rejection proportions under the null hypothesis using a mixture of chi-square distributions.

Main Methods:

  • Extensive Monte Carlo simulations were conducted.

Related Experiment Videos

  • The likelihood ratio test statistic was computed using both REML and ML.
  • The distribution of the test statistic was analyzed when adding one random component.
  • Rejection proportions were calculated under the null hypothesis.

    • Main Results:

    • The restricted likelihood ratio test statistic demonstrated reasonable agreement with the maximum likelihood test statistic.
    • For most parameter combinations, rejection proportions for both REML and ML were below the nominal 5% level.
    • On average, REML exhibited rejection proportions closer to the nominal level compared to ML.

    Conclusions:

    • The likelihood ratio test using REML is a viable approach for testing random components in linear mixed-effects models.
    • REML provides a more accurate control of Type I error rates compared to ML in this context.
    • The findings support the use of REML for hypothesis testing in linear mixed-effects models.