[EM] Correlation testing and maybe improvement thereby

[Michael Ossipoff]

Whether an allocation is gotten by Webster or Weighted Webster, it could be
tested for Pearson correlation between p and s/p. Then, (by coimputer of
course) some seat transfers could be tried. Maybe every possible seat
transfer, and every possible combination of two seat transfers. The new
allocations resulting from each of these transfers and combinations of
transfers could likewise be tested for Pearson .correlation. If any o them
resulted in a lower correlation coefficient, then that would be the final
allocation.

Just a suggestion for possibly improving a Webster or Weighted Webster seat
allocation.

I consider Webster to be unbiased enough, but maybe there could be demand
for the very best achievable unbias, and these are possible ways of
attempting that.

Mike Ossipoff
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120703/2f700196/attachment-0003.htm>