PubMed Health⌕ Search
Home/Archive/PubMed

PubMed · 11764269

Parameterization of multivariate random effects models for categorical data.

S Rabe-Hesketh·A Skrondal
View original source ↗Export metadata ↓

Abstract

Alternative parameterizations and problems of identification and estimation of multivariate random effects models for categorical responses are investigated. The issues are illustrated in the context of the multivariate binomial logit-normal (BLN) model introduced by Coull and Agresti (2000, Biometrics 56, 73-80). We demonstrate that the BLN model is poorly identified unless proper restrictions are imposed on the parameters. Moreover, estimation of BLN models is unduly computationally complex. In the first application considered by Coull and Agresti, an identification problem results in highly unstable, highly correlated parameter estimates and large standard errors. A probit-normal version of the specified BLN model is demonstrated to be underidentified, whereas the BLN model is empirically underidentified. Identification can be achieved by constraining one of the parameters. We show that a one-factor probit model is equivalent to the probit version of the specified BLN model and that a one-factor logit model is empirically equivalent to the BLN model. Estimation is greatly simplified by using a factor model.

Explore related subjects

BiometryData Interpretation, StatisticalDisease OutbreaksHumansInfluenza, HumanModels, StatisticalMultivariate AnalysisCommentComparative StudyJournal Article

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

S Rabe-Hesketh, A Skrondal. 2001. Parameterization of multivariate random effects models for categorical data.. https://doi.org/10.1111/j.0006-341x.2001.1256_1.x

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed2004

National biometry audit.

PURPOSE: To determine compliance with the Royal College of Ophthalmologists' (RCOphth) biometry guidelines. METHOD: A structured telephone questionnaire of individuals who perform biometry in all eye departments in the United Kingdom (UK). RESULTS: A biometrist was interviewed in 107 of the UK's 178 eye departments. Nurses alone run the biometry service in 58% of departments, orthoptists alone in 13%, junior doctors alone in 6%, optometrists alone in 3%, and a combination of staff in 20%. Of the staff interviewed, 37% had been on external biometry training courses. One intraocular lens (IOL) calculation formula was used for all eyes in 61% of departments with 17% using the obsolete SRK II formula, 36% of departments used two or more formulae and only 4% adhered to the RCOphth guidelines to use Hoffer Q in eyes with axial lengths <22.0 mm, an average of all three formulae in eyes between 22.0 and 24.5 mm, Holladay in eyes between 24.6 and 26.0 mm, and SRK/T in eyes >26.0 mm. Audit of refractive results was claimed by 71% of units but in only 17 (16%) did the biometrist know the percentage of eyes with a prediction error <or=1 D. CONCLUSION: This study demonstrates poor awareness and/or implementation of the RCOphth biometry guidelines and indicates that audits are either not highlighting poor results or are not resulting in a change in practice. The guidelines should be updated to emphasise the importance of customising A constants and to set benchmark standards for prediction error.

Biometry↗
PubMed2004

Intraocular pressure associations with refractive error and axial length in children.

AIM: To assess whether intraocular pressure (IOP) is associated with refractive error or axial length in children. METHODS: Of subjects from the Singapore Cohort Study of the Risk Factors for Myopia (SCORM), 636 Chinese children aged 9-11 years from two elementary schools underwent non-contact tonometry, cycloplegic autorefraction, and A-scan biometry during 2001. For analyses, refractive error was categorised into four groups; hypermetropia (spherical equivalent refraction (SE) > or = +1.0D), emmetropia (-0.5D<SE< +1.0D), low myopia (-3.0D<SE< or = -0.5D) and high myopia (SE< or = -3.0D). RESULTS: Of the 636 children examined, 50.6% were male. The mean IOP was 16.6 (SD 2.7) mm Hg. There were no significant IOP differences between low (mean IOP = 16.4 (2.8) mm Hg) or high myopes (16.7 (2.5) mm Hg) and emmetropes (16.7 (2.9) mm Hg), p = 0.57. IOP was not correlated with spherical equivalent refraction (Spearman correlation, r = 0.009) or axial length (r = 0.030). In regression analyses adjusting for diastolic blood pressure, neither spherical equivalent (regression coefficient = 0.014) nor axial length (regression coefficient = 0.027) were significantly associated with IOP. CONCLUSION: These findings do not support an association between IOP and refractive error or axial length in children. This questions postulated roles of IOP in the pathogenesis of myopia.

Biometry↗
PubMed2004

Advanced statistics: linear regression, part I: simple linear regression.

Simple linear regression is a mathematical technique used to model the relationship between a single independent predictor variable and a single dependent outcome variable. In this, the first of a two-part series exploring concepts in linear regression analysis, the four fundamental assumptions and the mechanics of simple linear regression are reviewed. The most common technique used to derive the regression line, the method of least squares, is described. The reader will be acquainted with other important concepts in simple linear regression, including: variable transformations, dummy variables, relationship to inference testing, and leverage. Simplified clinical examples with small datasets and graphic models are used to illustrate the points. This will provide a foundation for the second article in this series: a discussion of multiple linear regression, in which there are multiple predictor variables.

Biometry↗