Factorial ANOVA: decomposed significant interaction terms
I have a question regarding the Factorial ANOVA lecture notes. The last page of the notes (the penultimate slide) displays the descriptive and inferential statistics for the decomposed significant interactions. The cexpbydiv row is identical to the Corrected Model row in both that table as well as the table on the 3rd page. What, exactly, does the Corrected Model tell us, and why is it the same as cexpbydiv (the decomposed significant interaction terms)?
posted by Grace @ 11:29 AM
2 Comments:
At 8:29 PM,
David said…
The short answer is that you never have to look at the Corrected Model row in this class.
The longer answer (if you're interested) is that it is a composite of the main and interaction effects. That is, SS of the Corrected Model represents the variability that can be explained by all effects at the same time. Page 174 in the textbook gives a bit of a description of those "extra" terms. In table 5.15, you can see this is true by adding the SS for profess, length, and profess*length. You should get the SS for the Corrected Model.
In the factorial notes (p. 3), adding the SS for coachexpdv, divisiob, and coachexpdv*divisiob is close to, but not exactly 31.312 because of unequal cell sizes.
For the decomposed test, there is only one variable you're testing: cexpbydiv. So you'll get the exact same values as the Corrected Model row.
Hope that was helpful. If not, then you don't need to worry about it.
At 11:57 AM,
Grace said…
Thanks, David. That answers my question. :)
Post a Comment
<< Home