Orthogonality
We didnt go over this very much in class and from I what remember the discussion was simply that it was bad and why it was bad, with some examples of what the correlations would be but we did not go into any ways to fix it or determine the severity of the non-orthogonality, both of which we need to answer question 2F. I've read the pertitent sections in the textbook over and over, but I am still really unsure what you are looking for here. Can I get a clue, a hint, anything? The only thing I can figure out is to run some sort of correlation on the number of participants in each cell but I'm unsure how I would do that. Looking back over the notes, our little league example had unequal n and we never discussed that being a problem, so I am really confused.
1 Comments:
At 6:28 PM,
Mari said…
I'm afraid that I must disagree with you about this. We did discuss how to fix this--said exactly what we would need to do to have orthogonal estimates.
This is a design issue, not a printout or SPSS issue (although there are somewhat controversial approaches about what options you could take in running the ANOVA).
Post a Comment
<< Home