On Inference in Parametric Survival Data Models

From MaRDI portal





The present paper discusses aspects of maximum likelihood and related estimation methods (such as the Bayes and the \(M\)-type ones) applied to parametric survival data models which are of the following form: Some parametrically specified hazard rate \(\alpha(s,\theta)\) is assumed for possibly censored random life times \(X^ 0_ 1,\dots,X^ 0_ n\); one observes only \(X_ i=\min\{X^ 0_ i,c_ i\}\) and \(\delta_ i=I\{X^ 0_ i\leq c_ i\}\) (where \(I(A)\) is the indicator of the event \(A)\), \(i=1,\dots,n\), for certain censoring times \(c\) that either are given or come from some censoring distribution. These aspects include: (i) large sample behavior when the parametric model is a non-perfect approximation to the true model; (ii) distance measures from true to parametric model; (iii) model-based and model- robust estimation of the approximate covariance matrix of the maximum likelihood estimator; (iv) measures of influence; (v) natural alternative estimation procedures; (vi) similar questions for hazard rate regression models.




Cited in
(47)








This page was built for publication: On Inference in Parametric Survival Data Models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4024535)