Every year polls dozens of SEOs asking what they see as the most important factors for SEO. This year the also added a correlation study. They used over 10,000 keywords in the study which is an ample sample size.
i think over all their ranking factor reports are excellent and always give me something to think about and try out. But the correlations seem not so useful. As they point out, a high correlation doesn’t explain cause and effect.
For example: Sites that has the keyword being tested in their domain name may not rank well because the keyword is in the domain name but because poeple who buy domain names with keyword orientation tend to put a lot more effort into SEO than the ones that don’t.
So part of the problem here is separating out each element out and assign importance. The factors tend to play off of each other and to be influenced by many human factors.
I do commend SEOmoz for trying to better quantify results. As much as I appreciate the input from all the SEOs and take what they say seriously, it can be even more flawed. Ever read “Money Ball”?
But have a look. If you do SEO you will surly see something here that will be useful to you.