leverage analysis vs. case deletion

From: "Nele Plock" nplock@zedat.fu-berlin.de Subject: [NMusers] leverage analysis vs. case deletion Date: Tue, 02 May 2006 17:28:53 +0200 Dear nmusers, I was wondering if anybody could comment on the difference between leverage analysis and case deletion procedures for model evaluation? From what I have read so far it seems that they can be treated as synonyms, but is this really the case? Thanks a lot! Nele

From: Leonid Gibiansky leonidg@metrumrg.com Subject: Re: [NMusers] leverage analysis vs. case deletion Date: Tue, 02 May 2006 12:21:37 -0400 I think, case deletion procedure (i.e., delete some observations and see what happens) is the particular implementation of the leverage analysis (i.e., investigation of influence of particular observations on the parameter estimates). Leonid

From: "Bonate, Peter" Peter.Bonate@genzyme.com Subject: [NMusers] leverage analysis vs. case deletion Date: Tue, 2 May 2006 12:26:41 -0400 They are not synonyms. You can do a leverage analysis using case deletion but with linear models you do not need to do so. You can use the HAT matrix in a linear model to get the leverage a subject exerts in a dataset. With a nonlinear model with individual level data (no repeated data), you can get the same measure of influence using the Jacobian matrix. With a linear mixed effects model there are ways to do this but they are not easy and require access to software that allows matrix calculations, like MATLAB or SAS IML. But with a nonlinear mixed effects model, I am not aware of how you could get this data without doing case deletion. Whenever I can, I do a case deletion analysis on my data (time permitting) since you can also get the jackknife estimate of the standard errors. Hope this helps, pete Peter L. Bonate, PhD Genzyme Corporation Director, Pharmacokinetics 4545 Horizon Hill Blvd San Antonio, TX 78229 USA peter.bonate@genzyme.com phone: 210-949-8662 fax: 210-949-8219

From: "A.J. Rossini" blindglobe@gmail.com Subject: Re: [NMusers] leverage analysis vs. case deletion Date: Tue, 2 May 2006 18:46:33 +0200 Depends on whether you are thinking of using leverage analysis as a generic term or as an analysis technique, where the "see what happens" is particularly defined. Of course, case-deletion is morally the same as both jackknifing (variance estimate) and cross-validation (for a particular k, k-fold...). Lots of names, similar algorithms, different goals.... -- best, -tony blindglobe@gmail.com Muttenz, Switzerland. "Commit early,commit often, and commit in a repository from which we can easily roll-back your mistakes" (AJR, 4Jan05). _______________________________________________________