I left out an assumption, a different interpretation of the 2008 poll results, in which Obama, Paul, and Kuccinich respectively received 99, 74, and 66 top-rating marks.
Maybe some of the Kuccinich-preferrers top-rated Paul too, but none of the Paul-preferrers top-rated Kuccinich. Maybe, really, the Paul faction was no larger than the Kuccinich faction. So maybe, if everyone had top-rated only their favorite, Paul and Kuccinich would have both only had 66 top-rating marks. In that case, to make Paul outpoll Obama, the Kuccinich-preferrers would have to give Paul 1/2 max. That's convenient, because it could be done in 3-slot Score, as well as in 0-10 Score, since 3-slot Score's Middle rating is worth 1/2 max. Of course these SFR are based on guesses. Which guess you prefer to use is a matter of subjective preference based on intuition. Intuitive judgement about which previous-poll-interpretation, &/or which predicted change from last time, is more likely, &/or how much you like the other candidate. For interpreting the previous election or poll result, calculations could improve on the guesses. Maybe sometimes the solution of a system of equations could give better estimates of the faction-sizes, based on the numbers of points ratings each candidate received, at each points-level, and the total numbers of points received by each candidate. I'm not inclined to do that, and so I just based my SFR for Paul on one of the guesses. Mike Ossipoff ---- Election-Methods mailing list - see http://electorama.com/em for list info
