Jove
Visualize
Contact Us

Related Experiment Videos

Misspecifying latent class models by mixture binomials.

A K Formann1

  • 1Institut für Psychologie, Universität Wien, A-1010 Wien, Liebiggasse 5, Austria.

The British Journal of Mathematical and Statistical Psychology
|January 31, 2002
PubMed
Summary

Latent class models can exhibit overdispersion, making them appear similar to mixture binomial models. This can mislead researchers into accepting mixture binomials without verifying latent class model fit, potentially misinterpreting results.

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

Neurodevelopmental schizophrenia: obstetric complications, birth weight, premorbid social withdrawal and learning disabilities.

Neuropsychobiology·2001
Same author

Rater agreement and the generalized Rudas-Clogg-Lindsay index of fit.

Statistics in medicine·2000
Same author

Quality of life and psychosocial adjustment of young patients after treatment of bone cancer.

Cancer·1998
Same author

Latent class analysis in medical research.

Statistical methods in medical research·1996
Same author

Measurement errors in caries diagnosis: some further latent class models.

Biometrics·1994
Same author

Fixed-distance latent class models for the analysis of sets of two-way contingency tables.

Biometrics·1993
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

Area of Science:

  • Psychometrics
  • Statistical modeling
  • Cognitive development

Background:

  • Statistical models are crucial for analyzing performance data, especially in cognitive research.
  • Understanding heterogeneity in subject performance and task difficulty is key to accurate modeling.
  • Previous research often relies on simpler models like the mixture binomial, potentially overlooking complex data structures.

Purpose of the Study:

  • To investigate the relationship between latent class models and simpler statistical models (mixture binomial, independence, binomial).
  • To determine how heterogeneity in subject performance and task difficulty affects score distributions.
  • To highlight potential pitfalls in model selection when analyzing empirical data, using Piaget's water-level tasks as a case study.

Main Methods:

Related Experiment Videos

  • Considered four statistical models: unconstrained latent class, mixture binomial, independence, and binomial.
  • Analyzed the conditions under which overdispersion and underdispersion arise in these models.
  • Compared score distributions generated by latent class models and mixture binomials, particularly in the presence of overdispersion.

Main Results:

  • Latent class models can produce over- and underdispersion, encompassing the other three models as special cases.
  • Overdispersion can cause latent class and mixture binomial models to generate nearly indistinguishable score distributions.
  • The score distribution alone may not be sufficient to detect a lack of fit for mixture binomial models when a latent class model is more appropriate.

Conclusions:

  • It can be misleading to accept mixture binomial models as well-fitting without assessing latent class models, a common issue in empirical research.
  • The study underscores the importance of considering more complex models like latent class models to accurately capture heterogeneity in performance and task difficulty.
  • Findings have implications for the interpretation of results in fields utilizing similar statistical approaches, such as developmental psychology.