Date: Wed, 2 Jun 1999 16:31:46 -0500
Reply-To: "Nichols, David" <nichols@SPSS.COM>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: "Nichols, David" <nichols@SPSS.COM>
Subject: Re: GLM, lack of fit test
Khuri, A. I. (1985). A Test for Lack of Fit of a Linear Multiresponse Model.
Technometrics, 27, 213-218.
Levy, Martin S., and Neill, James W. (1990). Testing for Lack of Fit in
Linear Multiresponse Models based on Exact or Near Replicates.
Communications in Statistics, Part A-Theory and Methods, 19, 1987-2002.
The formulas are far too complicated to try to reproduce here. The null
hypothesis tested is that the data are generated by the fitted model (all
error is pure error, with no error due to misfitting of the model). The
alternative is that necessary terms are missing. The tests require exact
replications among predictor or covariate patterns, and that a non-full
factorial model be fitted.
Principal Support Statistician and
Manager of Statistical Support
> -----Original Message-----
> From: Helena Truszczynska [mailto:htrus1@POP.UKY.EDU]
> Sent: Monday, May 10, 1999 9:02 AM
> To: SPSSX-L@UGA.CC.UGA.EDU
> Subject: GLM, lack of fit test
> I would appreciate any information concerning the
> exact hypothesis, the
> formula and the references for the lack of fit test in GLM
> (options menu,
> SPSS versions 8 and 9).