Date:         Wed, 6 Jun 2007 18:33:46 -0700
Reply-To:     David L Cassell <davidlcassell@MSN.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         David L Cassell <davidlcassell@MSN.COM>
Subject:      Re: How to compute the effect size (glimmix output)
In-Reply-To:  <200706051202.l55AoOfX026021@mailgw.cc.uga.edu>
Content-Type: text/plain; format=flowed

mp2370@COLUMBIA.EDU wrote:
>
>Thank you all for the help. As I am struggling with new to me area of
>statistics, I have a question. How do you compute an effect size of the
>treatment effect (or interaction) in case of this output:
>
>  Solutions for Fixed Effects
>
>                                              Standard
>Effect        RAN002B  MONO  TIME  Estimate     Error     DF  t Value
>Pr<|t|
>
>Intercept                            0.9011    0.1728     90     5.21
><.0001
>logUSO_bl                            0.4845   0.05644    177     8.58
><.0001
>MONO                   No           -0.2934   0.05903    177    -4.97
><.0001
>MONO                   Yes                0         .      .      .       .
>RAN002B       HE                     0.2956    0.1674    177     1.77
>0.0791
>RAN002B       SSB                         0         .      .      .       .
>TIME                         3      0.04754   0.04051    177     1.17
>0.2421
>TIME                         4            0         .      .      .       .
>RAN002B*TIME  HE             3      -0.4141   0.05054    177    -8.19
><.0001
>RAN002B*TIME  HE             4            0         .      .      .       .
>RAN002B*TIME  SSB            3            0         .      .      .       .
>RAN002B*TIME  SSB            4            0         .      .      .       .
>
>
>                         Type III Tests of Fixed Effects
>
>                                 Num      Den
>                Effect            DF       DF    F Value    Pr > F
>
>                logUSO_bl          1      177      73.68    <.0001
>                MONO               1      177      24.70    <.0001
>                RAN002B            1      177       0.29    0.5927
>                TIME               1      177      39.70    <.0001
>                RAN002B*TIME       1      177      67.11    <.0001
>
>
>Do I need something else in my output or the informations are sufficient?
>
>Thank you a lot,
>
>Martina

Why are you after an effect size?

What is your Y variable?  Is it continuous, binary, ordinal, nominal, or ???
This matters, since there are a lot of ways to compute effect sizes, and
you usually need to choose an appropriate one based on your data.

Why are you using glimmix?  What about your data is forcing you to use
this tool?  It's fairly complex, and if a simpler procedure could be used,
you
might find it easier to get your results.

HTH,
David
--
David L. Cassell
mathematical statistician
Design Pathways
3115 NW Norwood Pl.
Corvallis OR 97330

_________________________________________________________________
Get a preview of Live Earth, the hottest event this summer - only on MSN
http://liveearth.msn.com?source=msntaglineliveearthhm