Kate,

There is a test for determining whether the elimination of predictor variables in a model results in an improvement to the model (i.e., more parsimonious). It's called the Wald test and is fairly easy to perform by hand (to the best of my knowledge, it isn't available in SPSS or any other package). The test determines whether the Restricted Model (i.e., the model with predictor variables removed) as compared to the Unrestricted Model has a significant increase in the Error Sums of Squares.

The statistic is described in Ramanathan, R. (1998). Introductory econometrics with applications. Fort Worth, TX: Dryden Press. (p. 169ff).

Hope this helps. Edgar

P.S. If you're unable to locate the statistic, I can send you an inelegant excel file with it.

_____ Edgar F. Johns <EdgarJohns@obik.com> Obik, LLC 2906 River Meadow Circle Canton, MI 48188 Tel. 734.495.1292, Fax 734.495.1981 http://www.obik.com

-----Original Message----- From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU]On Behalf Of Kate Riordan Sent: Sunday, May 13, 2001 4:54 PM To: SPSSX-L@LISTSERV.UGA.EDU Subject: F-test

Hello: I have a question regarding the F test in SPSS. I doing a regression analysis and am looking at three different models which build upon each other. I entered race in block one, socioeconomic status in block two and parental involvement in block three. I am trying to figure out if the change in the betas (the beta is reduced in each model) is a significant change-I was told that the F test tests whether an effect is significant as compared to the null hypothesis- it just indicates that somewhere there is a signicant difference- but I would like to know if the change in the beta is significant (as compared to the other models). Does anyone know if there a test in SPSS for this? This may be a joint F test?

Thank you for any help.

Kate Riordan