In my previous post about less biased divisor methods, I spoke of a Taylor or McLaurin polynomial approximation of the "complicated function" (the log-normal function, or the exponential of a fitted sum of log-normal functions that Kristofer spoke of).
Actually, the McLaurin form of the Taylor polynomial isn't what is needed here. The McLaurin version is centered on zero. It's used a lot, but, for the current purpose, we'd use, for each interval's R determination, a Taylor polynomial centered on the middle of that interval. Also, there was a typo: In the paragraph that starts with the words "That means that a term logarithmic in R will remain.." ...that sentence should read: "That means that a term logarithmic in R will remain, in the integration-result, along with a polynomial function of R." Mike Ossipoff ---- Election-Methods mailing list - see http://electorama.com/em for list info
