zbMATH — the first resource for mathematics

On errors-in-variables for binary regression models. (English) Zbl 0566.62093
The authors consider binary regression models when the predictors have errors. Assuming that nuisance parameters are independently and normally distributed, the conditional likelihood was derived. When the measurement error is large, the usual estimates are unreliable and in this situation, the authors examine the conditional maximum likelihood estimators.
An illustration is presented using a Monte Carlo study related to the Framingham study, see T. Gordon and W. E. Kannel, The Framingham study. Introduction and general background in the Framingham Study §§1,2. Bethesda, Maryland National Heart, Lung and Blood Institute (1968).
Reviewer: P.W.A.Dayananda

62P10 Applications of statistics to biology and medical sciences; meta analysis
62J99 Linear inference, regression
Full Text: DOI