How to perform simple main effects analysis in SPSS?

How can I perform a simple main effect analysis on SPSS? I shall use it as post-hoc test but am not really sure. I know I can somehow work in the Syntax with it. Could you help me to find out about the next steps?
Thank you very much :)


Super Moderator
It would help to know a bit more about what you are trying to do and what type of data you have. More than this, SPSS have excellent manuals available online. If you would like further help, please provide our members with some more information.

Thanks bugman for your advise!
So I did a two-way unrelated ANOVA with 4 groups which revealed that the interaction was significant and both main effects too. The groups have unequal sizes. Now I would like to find out between what groups the sig. mean differences exist.
If you have a tip for me - maybe a link - where I can find a easy understandable SPSS manual for SPSS 15 that would be great!
Thank you


No cake for spunky
It is dangerous to interpret main effects with signficant interaction. You should look at simple effects which SPSS will calculate.

Page 9 covers post-hoc tests

If you want a post-hoc test (Tukey HSD tends to be used a lot) this might help (it's an older form of SPSS, but it should still be roughly the same). ANOVA.pdf

The post-hoc example is actually covered in the one way anova portion.

This does not show how to activate post-hoc test, but it analyzes them and two way anova generally in the context of SPSS output.
Thank you noetsi, I did a Scheffe's range test originally but my professor was not satisfied with it. He said its not adequate. I chose it because I believe its the one you use when having unequal groups.
I shall do simple main effect analysis instead and interpret it in terms of relative effects. Why do you think its dangerous to interpret main effects with sig. interactions?


No cake for spunky
Scheffe is not accepted by most researchers these days. It's very conservative, meaning signficant effects won't be detected. SPSS does not do it the way it was actually intended either.

The problem with interpreting main effect when you have interaction is that main effects impact on the dependent variable will vary at different levels of the second main effect.

For example, say you have an independent variable with three levels and it interacted with a 2nd indendent variable with two levels. Then level one of the first independent variable might be lower on the dependent variable than that (independent) variable level three - at level one of the 2nd independent variab le. But at level two of the second independent variable level one of the first independent variable might be higher on the dependent variable than the first independent variable's level three.

So you can only interpret main effects at specific level of the other independent variable - not globally.