help in GLM repeated syntax, pls

Discussion in 'SPSS' started by Erkki.Komulainen, Oct 31, 2011.

  1. Hello SPSS profis!

    I have four measures all before and after.

    I use the following syntax:

    GLM c1 c2 c3 c4 BY time
    /WSFACTOR=chan 4
    /EMMEANS=TABLES (chan*time)
    /METHOD=SSTYPE(3)
    /CRITERIA=ALPHA(.05)
    /WSDESIGN=chan
    /DESIGN=time .

    I do not manage to create post-hoc -tests which
    would follow "/EMMEANS=TABLES (chan*time)"

    I would like to compare especially before-after in c2
    and before after in c4.

    Help would be appreciated!

    Erkki


    Mobile +358 40 5024491 <http://www.helsinki.fi/~komulain/>
     
    Erkki.Komulainen, Oct 31, 2011
    #1
    1. Advertisements

  2. Erkki.Komulainen

    Bruce Weaver Guest

    Hi Erkki. Your description suggests that the data file layout and
    syntax are incorrect. The syntax you posted suggests that you have two
    rows per person. If you want to use GLM, you need one row per person
    (with 8 variables for the Condition x Time combinations), and syntax
    something like:

    GLM c1t1 c1t2 c2t1 c2t2 c3t1 c3t2 c4t1 c4t2
    /WSFACTOR=cond 4 time 2
    /EMMEANS=TABLES (comd*time) compare(time)
    /WSDESIGN=chan time .


    Alternatively, you could restructure the data to have one row for each
    Condition x Time combination, and use MIXED. You would need an ID
    variable, of course, because the Condition x Time repeated measures
    would be clustered within individuals.

    HTH.
     
    Bruce Weaver, Oct 31, 2011
    #2
    1. Advertisements

  3. :On 31/10/2011 5:03 AM, wrote:
    :> Hello SPSS profis!
    :>
    :> I have four measures all before and after.
    :>
    :> I use the following syntax:
    :>
    :> GLM c1 c2 c3 c4 BY time
    :> /WSFACTOR=chan 4
    :> /EMMEANS=TABLES (chan*time)
    :> /METHOD=SSTYPE(3)
    :> /CRITERIA=ALPHA(.05)
    :> /WSDESIGN=chan
    :> /DESIGN=time .
    :>
    :> I do not manage to create post-hoc -tests which
    :> would follow "/EMMEANS=TABLES (chan*time)"
    :>
    :> I would like to compare especially before-after in c2
    :> and before after in c4.
    :>
    :> Help would be appreciated!
    :>
    :> Erkki

    :Hi Erkki. Your description suggests that the data file layout and
    :syntax are incorrect. The syntax you posted suggests that you have two
    :rows per person. If you want to use GLM, you need one row per person
    :(with 8 variables for the Condition x Time combinations), and syntax
    :something like:

    :GLM c1t1 c1t2 c2t1 c2t2 c3t1 c3t2 c4t1 c4t2
    : /WSFACTOR=cond 4 time 2
    : /EMMEANS=TABLES (comd*time) compare(time)
    : /WSDESIGN=chan time .


    :Alternatively, you could restructure the data to have one row for each
    :Condition x Time combination, and use MIXED. You would need an ID
    :variable, of course, because the Condition x Time repeated measures
    :would be clustered within individuals.

    :HTH.

    :--
    :Bruce Weaver
    :
    :http://sites.google.com/a/lakeheadu.ca/bweaver/Home
    :"When all else fails, RTFM."

    Hi Bruce!

    Thanks for the syntax. After correcting the typos it works. In post
    hoc -comparisons it produces different results when compared to the
    MIXED procedure.

    The post hoc -comparisons in this REPEATED syntax are same as separate
    pairwise t-tests while the MIXED procedure uses standard errors based on
    all variables not just the two compared.

    The MIXED model is a very flexible one. It needs the long format,
    though.

    All the best!

    Erkki

    Mobile +358 40 5024491 <http://www.helsinki.fi/~komulain/>
     
    Erkki.Komulainen, Oct 31, 2011
    #3
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.