I love it, folks. The reward challenge is not the only thing making my head spin. So, your contemplating an Analysis of Variance on four identified ways in which winners emerge. Would you take the 17- so-far winners and assign them to one of the proposed categories (I guess that is a big enough group?)... and then factor away? I've got an old merchant desk calculator with a crank handle that you can have if you want to do that by hand, Green Coffee.I started making a file to do a logistic regression on Edgic, but gave up after realizing it wouldn't be valid (the fact that multiple winner patterns - at the very least 2, plus a "loser" condition - wouldn't work very well in measuring a dichotomous DV).
I'm not sure how any sort of stats could work when you mix qualitative and continuous variables in predicting a multi-level qualitative DV. While I've had quite a bit of stats, that's beyond anything I've ever done. We would probably need some sort of statistics master to get into Edgic for something like that.
That's like the ultimate disaster lurking. If you dig deep enough and use too fine of an analysis, that's exactly what you'd find. So even if we move toward more complexity, it's important we don't go too far.
Ever feel like there are 17 different winner constructs and that Tocantins will only give us a 18th variation?
