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

Bias in misspecified mixtures

G Gray1

  • 1Department of Statistics, North Carolina State University, Raleigh 27695-8203.

Biometrics
|June 1, 1994
PubMed
Summary
This summary is machine-generated.

Finite mixture models can suffer from severe bias in maximum likelihood estimators (MLEs) due to model misspecification. Applying the Box-Cox transformation can significantly reduce this asymptotic bias, enabling more reliable statistical inference.

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

Analysis of adverse events following COVID-19 vaccination and infection: A retrospective, comparative cohort study using a claims database from Discovery Health, a managed care organisation in South Africa.

South African medical journal = Suid-Afrikaanse tydskrif vir geneeskunde·2026
Same author

Clot twist - D-dimer analysis of healthy adults receiving heterologous or homologous booster COVID-19 vaccine after a single prime dose of Ad26.COV2.S in a phase II randomised open-label trial, BaSiS.

South African medical journal = Suid-Afrikaanse tydskrif vir geneeskunde·2025
Same author

Vaccine pharmacovigilance in South Africa: successes and limitations of current approaches.

Expert opinion on drug safety·2024
Same author

Death trends for 2010 - 2022 for members of a large private medical scheme in South Africa.

South African medical journal = Suid-Afrikaanse tydskrif vir geneeskunde·2024
Same author

Frequency, kinetics and determinants of viable SARS-CoV-2 in bioaerosols from ambulatory COVID-19 patients infected with the Beta, Delta or Omicron variants.

Nature communications·2024
Same author

The Functional Assessment of Cancer Therapy Scale: Development and Validation of the General Measure.

Journal of clinical oncology : official journal of the American Society of Clinical Oncology·2023

Area of Science:

  • Statistics
  • Statistical Modeling

Background:

  • Finite mixture models are widely used in biology, often assuming normal component densities.
  • Inference issues in mixture models are known, but extreme biases from misspecification are less documented.

Purpose of the Study:

  • To investigate and quantify asymptotic biases in maximum likelihood estimators (MLEs) of finite mixture models under misspecification.
  • To evaluate the effectiveness of the Box-Cox transformation in mitigating these biases.

Main Methods:

  • Calculation of asymptotic biases for MLEs when assuming normality for skewed or unequal variance component distributions.
  • Application and estimation of the Box-Cox transformation parameter (lambda) within the mixture model likelihood.

Main Results:

Related Experiment Videos

  • Even assuming equal variances can lead to substantial asymptotic bias in MLEs.
  • Assuming normality for skewed distributions also causes significant bias.
  • The Box-Cox transformation effectively reduces asymptotic bias, especially for mixtures of skewed components.

Conclusions:

  • Model misspecification in finite mixture analysis, particularly normality assumptions, can lead to severe estimator bias.
  • Incorporating the Box-Cox transformation with an estimated lambda parameter offers a robust method to reduce bias and improve inference quality.