Modeling Heterogeneous Variance–Covariance Components in Two-Level Models

Leckie, George; French, Robert; Charlton, Chris; Browne, William J.

doi:10.3102/1076998614546494

Cited by 82 publications

(101 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the item scale parameters are poorly estimated with low coverage, relatively high rmse, and diminished confidence interval width. This finding agrees with results reported in Leckie, French, Charlton, and Browne (2014) for a location-scale model of continuous outcomes, who also noted poor coverage and spurious precision for scale (regression parameter) estimates if the random scale effect was not included in the model. Thus, if one is interested in assessing item scale effects, it is important to allow for random subject scale effects in the model.…”

Section: Simulation Studysupporting

confidence: 91%

A mixed-effects location-scale model for ordinal questionnaire data

Hedeker

Mermelstein

Demirtaş

et al. 2016

Health Serv Outcomes Res Method

View full text Add to dashboard Cite

In health studies, questionnaire items are often scored on an ordinal scale, for example on a Likert scale. For such questionnaires, item response theory (IRT) models provide a useful approach for obtaining summary scores for subjects (i.e., the model's random subject effect) and characteristics of the items (e.g., item difficulty and discrimination). In this article, we describe a model that allows the items to additionally exhibit different within-subject variance, and also includes a subject-level random effect to the within-subject variance specification. This permits subjects to be characterized in terms of their mean level, or location, and their variability, or scale, and the model allows item difficulty and discrimination in terms of both random subject effects (location and scale). We illustrate application of this location-scale mixed model using data from the Social Subscale of the Drinking Motives Questionnaire (SS-DMQ) assessed in an adolescent study. We show that the proposed model fits the data significantly better than simpler IRT models, and is able to identify items and subjects that are not well-fit by the simpler models. The proposed model has useful applications in many areas where questionnaires are often rated on an ordinal scale, and there is interest in characterizing subjects in terms of both their mean and variability.

show abstract

Section: Simulation Studysupporting

confidence: 91%

A mixed-effects location-scale model for ordinal questionnaire data

Hedeker

Mermelstein

Demirtaş

et al. 2016

Health Serv Outcomes Res Method

View full text Add to dashboard Cite

show abstract

“…Alternatively those familiar with the Bayesian estimation framework may choose to fit this extended mixed-effects location scale model using the Stat-JR software 27 as this has been shown to be possible in a recent application of this model to cross-sectional clustered data. 28 Lastly, our findings are based on pain assessment after orthodontic separators placement. We are not aware of any literature applying similar analyses to studies involving comprehensive fixed orthodontic appliance.…”

Section: Limitations and Future Directionsmentioning

confidence: 99%

Orthodontic pain trajectories in adolescents: Between-subject and within-subject variability in pain perception

Sandhu¹,

Leckie

2016

American Journal of Orthodontics and Dentofacial Orthopedics

Self Cite

View full text Add to dashboard Cite

show abstract

“…A more appealing approach is to treat the covariance matrices as exchangeable by specifying them as being drawn from a distribution, with unknown hyperparameters to be estimated. We are actively exploring this new class of model (Leckie et al ., ) and are implementing this and related extensions in the new Stat‐JR software (Charlton et al ., ) being developed at the Centre for Multilevel Modelling.…”

Section: Discussionmentioning

confidence: 99%

A Multilevel Modelling Approach to Measuring Changing Patterns of Ethnic Composition and Segregation Among London Secondary Schools, 2001–2010

Leckie

2014

Journal of the Royal Statistical Society Series A: Statistics in Society

Self Cite

View full text Add to dashboard Cite

Summary Multilevel binomial logistic regression has recently been proposed for the special case of statistically modelling changing composition and segregation of two groups of individuals over two occasions among organizational units, enabling inferences to be made about the underlying social processes which generate these patterns. A simulation method can then be used to re‐express the model parameters in the metric of any desired two‐group segregation index. We generalize this combined modelling and simulation approach by proposing multilevel random‐coefficient multinomial logistic regression for the general case of statistically modelling multiple groups of individuals over multiple occasions and multiple organizational scales. We illustrate this combined approach with an application to modelling changing three‐group white–black–Asian ethnic composition and segregation among London secondary schools and local authorities during the first decade of the 21st century.

show abstract

Modeling Heterogeneous Variance–Covariance Components in Two-Level Models

Abstract: His research interests are in the use of statistical modeling techniques, in particular MCMC methods, for analyzing complex datasets in many fields including veterinary epidemiology, ecology and education.

Cited by 82 publications

References 42 publications

A mixed-effects location-scale model for ordinal questionnaire data

A mixed-effects location-scale model for ordinal questionnaire data

Orthodontic pain trajectories in adolescents: Between-subject and within-subject variability in pain perception

A Multilevel Modelling Approach to Measuring Changing Patterns of Ethnic Composition and Segregation Among London Secondary Schools, 2001–2010

Contact Info

Product

Resources

About