Date:         Wed, 3 May 2000 14:52:35 -0500
Reply-To:     "Nichols, David" <nichols@SPSS.COM>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Nichols, David" <nichols@SPSS.COM>
Subject:      Re: SPSS 10 Log Linear

SPSS has three loglinear modeling procedures. The older LOGLINEAR procedure
is now available only through command syntax. The three options in the menus
take you to dialog boxes either for the GENLOG procedure or the HILOGLINEAR
procedure.

The Model Selection option takes you to a dialog box for the HILOGLINEAR
procedure. HILOGLINEAR fits only loglinear representations of models, in
which there are none denoted as independent or dependent. It will only
produce parameter estimates for saturated models. It's major attraction is
that it will automatically start with a saturated model (or something less
complete) and sequentially test terms for exclusion.

The General and Logit options take you to different dialog boxes for the
GENLOG procedure. The General option also fits a loglinear representation of
the models, where no factors are specified as independent or dependent (I've
seen this construed as all factors dependent or as all independent with the
logs of frequencies as the dependent). The Logit option takes you to a
different dialog box that allows specification of some factors as dependent
and others as independent. Throughout GENLOG, you can get parameter
estimates for any model.

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

> -----Original Message-----
> From: Michael Mallen [mailto:mjmallen@BOOMBOX.MICRO.UMN.EDU]
> Sent: Wednesday, April 26, 2000 12:04 PM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: [SPSSX-L] SPSS 10 Log Linear
>
>
> Hello,
>
>   I'm attempting to run statistics using Log linear.  I have
> one dependent
> variable and four factors.  I'm specifically wanting to see
> if there is any
> relationship among the four factors.  When I look at the
> three options under
> log linear I have General, Logit and Model as my analysis.  The Model
> appears to be the analysis I need to run since I'm looking
> for parameter
> estimates and standard error for each factor.  However when I
> attempt to use
> this analysis I'm not getting the correct stats.  Please provide any
> clarification/assistance to help me!!!  Thank you in advance.
>
>
> Michael Mallen
>