2-way ANOVA or repeated measures ANOVA ?? please help for my thesis

[Mareli]

Dear all,

I am working on this for a couple of weeks now and found several guides and examples on both of the methods above, but no example really fits my topic. I tried both of them and lots more with my data but just dont know, which procedure produces the RIGHT results....

I have 90 participants in my experiment and all of them were watching 12 different commercials. There are 4 treatment groups - one was watching a funny movie, one was watching a love movie and the other 2 were exposed to a horror movie and a documentary movie. The commercials were inserted in the movies.

My dependent variables are attitude variables (measured on Likert scales) regarding brand and spot itself.

Is it true that the 12 spots can be seen as repeated measures?
In this case I guess a repeated measures ANOVA would be best.

(In SPSS I did fill in the exposed spots as the within-subjects factor, defined the 12 levels as the outcome attitude variables and selected the movie group (1-4) as the between-subjects factor;
When I do this the Mauchly's Test of Sphericity is highly significant though)

Or is this only the case when there have the same measures been taken (e.g. for the same spots) on different time points? Then I would have to use 2-way ANOVA, right?

I would really appreciate your help as I am really confused with all the different examples I have seen.. I would be happy if you could also help me with which specific test to use in SPSS, in case the procdeure I described above is incorrect.

Many thanks in advance!
Mare

 

[Karabiner]

My dependent variables are attitude variables (measured on Likert scales) regarding brand and spot itself.

Were these really Likert scales (consisting of several Likert-type items which are summed up), or were these merely single Likert-type items?

Is it true that the 12 spots can be seen as repeated measures?

They could be considered as 12 categories of a repeated-measures factor, yes.

In this case I guess a repeated measures ANOVA would be best.

In case that you only have Likert-type items (ordinal scaled), not interval scales, this could perhaps be denied by some reviewers.

With kind regards

K.

 

[Mareli]

Thanks for your quick response, Karabiner

Were these really Likert scales (consisting of several Likert-type items which are summed up), or were these merely single Likert-type items?

These were all multi item scales. I used only the summed up total value.

In case that you only have Likert-type items (ordinal scaled), not interval scales, this could perhaps be denied by some reviewers.

I have only Likert-type items, yes. But aren't they usually treated as interval scales in Marketing science..?