LISTSERV at the University of Georgia
Menubar Imagemap
Home Browse Manage Request Manuals Register
Previous messageNext messagePrevious in topicNext in topicPrevious by same authorNext by same authorPrevious page (March 2005)Back to main SPSSX-L pageJoin or leave SPSSX-L (or change settings)ReplyPost a new messageSearchProportional fontNon-proportional font
Date:         Mon, 14 Mar 2005 16:43:14 -0500
Reply-To:     Fritz Consulting <fritzconsulting@patmedia.net>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Fritz Consulting <fritzconsulting@patmedia.net>
Organization: Carl Fritz
Subject:      Re: GLM TEST subcommand - specify error term
Comments: To: 'Marta García-Granero' <biostatistics@terra.es>
In-Reply-To:  <858689484.20050314181707@terra.es>
Content-Type: text/plain; charset="iso-8859-1"

Marta,

Thank-you for your quick response. If I had only a few dependent variables, I could do as you recommended and repeat the analysis and specify the df and MSE in the /TEST subcommand to get the test that I want. However, the program is going to be used in situations where there are 20-30 dependent variables and the user would like to automate the process so that the program would be run once and the output would be sent to a file for post-processing into a table.

Is anyone familiar with a way that the MS(Error) can be specified as the error term on a /TEST subcommand within GLM without having to run the analysis twice and 'manually' entering the df and value for the MS(Error)?

Thanks, Carl Fritz

-----Original Message-----

Hi Fritz,

Run the analysis once (without /TEST subcommands), and copy the value for error SS and df, then repeat the analysis with the /TEST subcommand indicating the SS and the df in parenthesis. For example:

UNIANOVA sibs BY sex /METHOD = SSTYPE(3) /INTERCEPT = INCLUDE /CRITERIA = ALPHA(.05) /TEST = sex vs 13892.577 df (1503) /DESIGN = sex .

This code will test the SEX effect against an error term=13892.577/1503=9.243

HTH

Marta

FC> I am using GLM/UNIANOVA with a model that contains fixed and random FC> effects (there are some specific reasons why I don’t want to use the FC> MIXED procedure). I would like to force SPSS to use the MSError as the FC> error term for testing the significance of some model terms. I have FC> tried to use the syntax shown below, but I get an error message that the FC> word ‘Error’ is not allowed in the /TEST subcommand, possibly because FC> ‘Error’ is not a term listed in the /DESIGN subcommand.

FC> Can anyone tell me if it is possible to force SPSS to use the MSError FC> term for testing hypotheses in GLM/UNIANOVA.

FC> UNIANOVA FC> Sour BY Product Rep Judge FC> /RANDOM = Rep Judge FC> /METHOD = SSTYPE(3) FC> /INTERCEPT = INCLUDE FC> /POSTHOC = Product ( TUKEY ) FC> /CRITERIA = ALPHA(.10) FC> /TEST = judge VS Error FC> /TEST = rep VS Error FC> /DESIGN = Rep Judge Product Judge*Rep Product*Rep Judge*Product .


Back to: Top of message | Previous page | Main SPSSX-L page