A geometric approach to nonlinear regression diagnostics with application to matched case-control studies (Q802251)

The paper gives a discussion of the problem of how deletion of observations changes parameter estimation in the fitting of statistical models to observations. A new method is proposed for computing an approximation to the numerical value of this change. The method is compared to older methods of such a computation. As an illustration the methods are applied numerically to a matched case-control study of endometrial cancer in Los Angeles. The results are shown in graphs.