2x2 anova/post hoc

[Debra Morley]

Is there a way to run a post-hoc (Tukey's) on a 2 x 2 ANOVA? I have the SPSS
graduate package on my PC and also have my data in SPSS on the UNIX system at
school.

Thanks in advance for any help.

Debbie

[Debra Morley]

Is there a way to do a post-hoc (Tukey's) on a 2 x 2 ANOVA in SPSS?

[William B. Ware]

I wouldn't recommend doing a Tukey HSD in this context. If you have an
interaction, you should be doing some sort of simple main effect test. If
there is no interaction, you don't need any follow up on a two-level
factor.

__________________________________________________________________________
William B. Ware, Professor and Chair Psychological Studies
CB# 3500 in Education
University of North Carolina PHONE (919)-962-7848
Chapel Hill, NC 27599-3500 FAX: (919)-962-1533
http://www.unc.edu/~wbware/ EMAIL: wbw...@unc.edu
__________________________________________________________________________

On Thu, 12 Mar 1998, Debra Morley wrote:

> Is there a way to run a post-hoc (Tukey's) on a 2 x 2 ANOVA? I have the SPSS
> graduate package on my PC and also have my data in SPSS on the UNIX system at
> school.
>

[Marcelo Costa Ferreira]

Debbie

I am not sure, but it should be possible in new versions of SPSS for
Windows

On 12 Mar 1998 19:34:52 GMT, dsmo...@bu.edu (Debra Morley) wrote:

>Is there a way to do a post-hoc (Tukey's) on a 2 x 2 ANOVA in SPSS?

[Burton L. Alperson]

Kim Ki Won's statement is true only for tests of the main effects. Tests of
the cell means for the interaction effect does require a post hoc
procedure. One possibility is to create a new variable called Groups which
contains the following levels:
1 = a1b1
2 = a1b2
3 = a2b1
4 = a2b2

You can then run post hocs or contrasts in the 1-Way procedure.

Burt

Kim Ki Won <h...@chollian.net> wrote in article
<351933E8...@chollian.net>...
> I can't understand.
> Why do you run post hoc procedure on 2X2 ANOVA design?
> When you have 3 or more levels in your variable, you need post hoc
> procedure.
> And there are no way to do post hoc on factorial design procdure in
> SPSS.

[Kim Ki Won]

[Jim Tromater]

Marcelo Costa Ferreira wrote:
>
> Debbie
>
> I am not sure, but it should be possible in new versions of SPSS for
> Windows
>
> On 12 Mar 1998 19:34:52 GMT, dsmo...@bu.edu (Debra Morley) wrote:
>
> >Is there a way to do a post-hoc (Tukey's) on a 2 x 2 ANOVA in SPSS?
> >
> >Thanks in advance for any help.
> >
> >Debbie

You can fake SPSS into doing the 6 possible pair comparisons by
identifying the cells and using that as the iv in the oneway anova.
rountine.
Jim
--

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
^ Jim Tromater ^^ Ph: 804-289-8131
^
^ Dept. of Psycholoygy ^^ Fax:
804-289-8943 ^
^ Univ. of Richmond, VA 23173 ^^ Email: jtro...@richmond.edu ^

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

[Brigitte Beauvais]

I am currently looking for a way to compute significance and confidence
intervals to test the difference between each individual cells of a
crosstab and the column and row maginals of this table.

Any suggestion?

Brigitte

[Kim Ki Won]

Jim Tromater wrote:

> Marcelo Costa Ferreira wrote:
> >
> > Debbie
> >
> > I am not sure, but it should be possible in new versions of SPSS for
> > Windows
> >
> > On 12 Mar 1998 19:34:52 GMT, dsmo...@bu.edu (Debra Morley) wrote:
> >
> > >Is there a way to do a post-hoc (Tukey's) on a 2 x 2 ANOVA in SPSS?
> > >
> > >Thanks in advance for any help.
> > >
> > >Debbie
> You can fake SPSS into doing the 6 possible pair comparisons by
> identifying the cells and using that as the iv in the oneway anova.
> rountine.
> Jim

I think not 6 possible pair comparisons but 4 possible pair comparisons.
Am I wrong?