Date:         Mon, 22 Nov 1999 17:05:32 -0600
Reply-To:     "Nichols, David" <nichols@SPSS.COM>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Nichols, David" <nichols@SPSS.COM>
Subject:      Re: Nesting

By reference text do you mean the simple.txt article available at
http://www.spss.com/tech/articles.html? If not, check that, because the last
example in that article is one with one between subjects and two within
subjects factors, and it lays out the general rules one uses to get simple
effects in MANOVA. I would recommand using the newer GLM procedure though,
which makes this really easy in most cases with the EMMEANS COMPARE option.

David Nichols
Principal Support Statistician and
Manager of Statistical Support
SPSS Inc.

> -----Original Message-----
> From: Michael Mallen [mailto:mjmallen@boombox.micro.umn.edu]
> Sent: Monday, November 22, 1999 10:39 AM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: [SPSSX-L] Nesting
>
>
> Hello,
>
> Using the reference text and your model code, I 've gotten to
> the point
> where I get results with a full factorial model.
>
> However, I think I need to use "mwithin" to get the nesting
> correct. As you
> may recall we have one between subjects factor, group (or
> radiation) and
> two within subjects factors, location and drug. However to
> look at the drug
> effect I only want to compare L1D1 to L1D0 and L2D1 to L2D0,
> i.e drug is
> additionally nested within location. If I want to look at the location
> effect within subjects, I compare L1D1 to L2D1.
>
> Finally, in the full model, the test of between subjects
> effect gives me a
> sig of 0.11 for a group (or radiation) effect. But when I use
> /WSdesign
> statements of mwithin location or drug I get differest results.
>
> Do you think the following code shows that extra level of nesting?
>
>
> * Split Plot with I believe proper nesting.
> TITLE
>  'Peters Group 1 and 2'.
> SUBTITLE
>  '11/20/99'.
> Manova L1D1 to L2D2 by Group(1,2)
>      /WSFactors=Location(2) Drug(2)
>         /WSDesign = mwithin drug(1)Location(1) mwithin
> drug(2)Location(1)
>         /METHOD UNIQUE
>         /ERROR WITHIN
>         /Print=CellInfo(Means) SigNif(AvOnly) Homogeneity
>         /Residuals=Plot
>         /Plot = normal
>         /Design.
>
> More questions along this line:
>
> Will this statement show me group effect while holding drug constant?
>
> /WSDesign = mwithin drug(1) mwithin drug(2)
>
> Will this statemetn show me group effect while holding
> location constant?
>
> /WSDesign = mwithin Location(1) mwithin Location(2)
>
>
> How come I can't run:
> *won't run 'mwithin drug(1)Location(2) mwithin drug(2)Location(2)'
> * or 'mwithin Location(2)drug(1) mwithin Location(2)drug(2)'
>
> Will this statement somehow give me the simple effects of
> drug with the
> levels of group?
>
> /Design = mwithin Group(1) mwithin Group(2)
>
>
> Finally, if I have to run multiple /WSdesign statements, how
> do I string
> them all together like they did in the example on p 465 of
> the advanced
> statistics manual?
>
> Thank you very, very much,
>