Arne Babenhauserheide writes: > Ian Clarke writes: > >> On Wed, Sep 28, 2016 2:56 PM, Arne Babenhauserheide arne_...@web.de >> wrote:This, written in a thread where I show a cleaner method of evaluation >> >> along with an implementation of a way to see which reasoning can >> >> actually be taken from the poll as it was conducted. >> >> >> What is your specific proposal, because apparently I missed it. All I saw >> were >> you mentioning a variety of alternatives that are not suitable because they >> do >> not retain the ordinality of people's estimates. > > My specific proposal is to use several different methods of ordering the > tasks by the votes given (value divided by cost estimate) and taking a > subset which is highly ranked in all the different methods of evaluation > (except for the ill-defined one which divides the mean by the spread of > the votes). This is the set for which it is possible to give the robust > answer that it is preferred by the people casting the votes. > > When I do that with the top 10, I get 6 which are ranked in all the > methods. Even just using mean and median should at least make this a bit > more resilient.
Clearer, with the preliminary evaluation, out of the top 10 (by mean value), 6 entries are robust in that position and 4 of the 6 are non-controversial. Taken the other way around: * For 4 entries there is *non-controversial* agreement that they deliver the largest value per cost. Their position in the top 10 is robust. * For 2 further entries, there is a *majority* agreement that they deliver the largest value per cost, but the values differ widely. Their position is robust, but there is no consensus. > I do not find the word ordinality in my dictionary. What does it mean exactly? Best wishes, Arne -- Unpolitisch sein heißt politisch sein ohne es zu merken
signature.asc
Description: PGP signature
_______________________________________________ Devl mailing list Devl@freenetproject.org https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
