It simply isn't possible to assign meaningful percentages of sexual attraction, and because of that I'd say the numbers would harm a study. The margin for error would surely be at least 30/40% either way, and that to me invalidates the study.
Where are you getting the 30/40% error? You're already making a quantitative analysis on something you just got through saying you can't.
Furthermore, it's not especially uncommon to see large variances. It's a characteristic of how your data is distributed, just like the mean or sample density, and can even itself indicate some other significant factor. You can come up with two classes of data whose intervals overlap a huge chunk of their intervals, but still have notable differences in their centers and endpoints (muscle mass between sexes). You can come up with two classes of data with similar means and variances but drastically different distributions that can mean very different things (concentrated in a bell curve, in off-center clusters, or fairly constant throughout?).
Even for really weird results, we have a shitload of tools to analyze, classify, and look for patterns. If you'd like, I can link you slides from a Data Mining course here, very interesting stuff. A lot of it deals with how to handle and weight error in various circumstances.
As for a meaningful measurement for this topic, frequency and severity of genital engorgement is a good place to start, with the understanding that we also suspect it might be caused by other things and would need to triangulate results somehow via other criteria and studies. We've previously been using vaginal lubrication, which is starting to seem like it had problems because upon comparing it to other factors we found it could be caused by too many other things and that may have skewed our results. That's not a failure of the analytical process, it's a success
— even if we might have to chuck some old data, we know why
, and we came out of it having learned something we didn't know before.
edited 25th Feb '11 7:31:55 PM by Pykrete