At 01:28 PM 7/27/2007, Elisabeth Varin wrote: >Hi Abd, and M. Smith if I understood well,
The spreadsheet and this particular idea was entirely my own. mr. Smith had refrained from comment on it. He is not at all responsible for my errors. >I checked your spreadsheet. > >You are missing the -1 cases for A. >Thus the impact on the sincere ballot compared to the approval one >is overestimated. >Try again considering -1 , 0 and 1 as the three rests after removing >your offset. >You should find different results. > >The emperor is not naked, he is just not well dressed... Thanks for checking. I fixed the problem, I think. The results, however, are still quite interesting. I now have an increasingly general purpose calculator for determining utility expecations. Right now it is set up for three-candidate elections, you can input the utilities and the voting pattern and see the utility. It's exact, it is not a simulator. It's very interesting to see what the utilities do. There is still an error in the spreadsheet I hope to correct tomorrow. When the utilities are changed, the expected utility for not voting changes, whereas I just assume that it is 1, the utility I previously had for the middle candidate. In particular, it is simple with this to compare the utility of Range 2 (CR-3) with Approval (Range 1 or CR-2). I don't think I've seen this comparison done elsewhere. It is not necessarily what one would expect, though, in hindsight, it should have been obvious. Still, it's nice to see the exact numbers. I've been hinting at it, and I announced a result before. However, that was with the erroneous numbers. When the numbers are fixed, what happens? The new spreadsheet is at http://beyondpolitics.org/Range2Utility.htm. (or .xls for the actual spreadsheet file.) It's cleaner and better explained. And hopefully, correct this time. ---- Election-Methods mailing list - see http://electorama.com/em for list info
