Dear all, Im new to the Electorama list and also a begginer to voting methods. I would like some guidance from the list on how to guess the results of an election, that recently took place in my country, if different voting procedures, other than Plurality, were used. The story follows.
The portuguese national television RTP organized an election to select the 10 Greatest Portuguese of all time. These were previously selected from a list of 100, I supposed using the Plurality method. Those final 10 personalities were then ranked by 159.245 telephone voters, during a period of a couple of months which ended last 25.03.2007. Again, Plurality was chosen to find the final results, which were: 1st - AS (Antonio de Oliveira Salazar, politician, portuguese dictator) - 41,0% 2nd - AC (Alvaro Cunhal, politician, comunist party leader) - 19,1% 3rd - AM (Aristides de Sousa Mendes, ambassador) - 13,0% 4th - DH (D. Afonso Henriques, first king of Portugal) - 12,4% 5th - LC (Luis de Camoes, poet) - 4,0% 6th - DJ (D. Joao II, king) - 3,0% 7th - IH (Infante D. Henrique, navigator) - 2,7% 8th - FP (Fernando Pessoa, poet) - 2,4% 9th - MP (Marques de Pombal, politician) - 1,7% 10th - VG (Vasco da Gama, navigator) - 0,7% The problem is, besides the Plurality criteria, that in the last weeks before the end of the telephone voting there were some newspaper rumors that AS and AC were going to win the election, so everyone believes this caused a rush of AS and AC suporters in the last days to vote against each others candidate, and at the same time distort the overall result. In the end the Plurality winer was AS, a dictator that ruled the country, in the last century, for almost 50 years, and the second Plurality winner was AC, the comunist party leader that opposed the AS dictatorship and was in prision some time during that period (notice AS and AC voters are on quite opposite sides). This result, for most portuguese, seem illogical, and for some even unacceptable, in a country where the majority is very proud of its history, of its navigators and poets, and, I believe, not as much of its recent politicians, like AS and AC. The main question is: does this election procedure, considering the Plurality prcedure and the quoted problems, has any mathematical value? Popular will was well represented in the final Plurality result? I truly believe NO. If a better voting method was chosen, certainly the final results would been very different. Based on this assumption, and since no ranked voting information was collected, I asked 5 honest friends to guess how 10 typical voters would rank their candidates (typical voter i would vote 9 in candidate i) and considered the average of these 5 opinions. Of course this exercise has its weaknesses but, anyway, the average rankings if found were: 10 candidates: AS AC AM DH LC DJ IH FP MP VG 1000 voters, distributed according with the Plurality results. 410: AS>DH>MP>IH>VG>DJ>AC>AM>LC 191: AC>AM>DH>IH>LC>FP>VG>AS>MP 130: AM>LC>FP>VG>IH>AC>DH>MP>AS 124: DH>VG>DJ>IH>MP>LC>FP>AS>AM 40: LC>FP>VG>IH>AM>DJ>DH>MP>AC 30: DJ>DH>VG>IH>MP>LC>FP>AS>AC 27: IH>VG>DH>LC>MP>DJ>FP>AS>AC 24: FP>LC>AM>VG>DH>IH>MP>AC>DJ 17: MP>DH>VG>AS>DJ>IH>LC>FP>AM 7: VG>IH>DH>LC>DJ>MP>AM>AS>FP Initially I tested these using the Borda method, and got the following result: DH>IH>VG>AS>MP>AM>LC>AC>DJ>FP. So, the fair winner would be DH (D. Afonso Henriques, the first king of Portugal, also called the Conqueror). I also used the Voting Calculator available from http://rangevoting.org/VoteCalc.html Again DH was elected in several realiable voting methods. Now my main doubts: 1) Which voting method would be best for this specific election?; 2) Which Bayesian regret values can be considered excelent, acceptable or very bad? How to justify, based only on BR, that Plurality is unreliable and should be discarded?; 3) What would be the results of BR using IEVS for NumVoters=159245 or I dont need to used such a large NumVoters?; 4) Is there a recommended Honfrac value I shoud use? What percentage of honest voters are reasonable in this specific election?; 4) Shoud I use only BR to compare the quality of voting methods or Condorced winners also provide a good measure? How these two relate and should be interpreted?; 5) Some voting methods presented in IEVS are quite obscure to me. Which provide the possibility of ranked winners? Are they document somewhere with examples? When will IEVS option 4 will be implemented?; 6) I couldnt find any books or papers focusing on these doubts. Any recommendations? Sorry for so many questions. Any help is very much appreciated. Yours truly, Ricardo Carvalho (Lisbon, Portugal) ------------- Clix ADSL até 24 Mb: a partir de 29,90/mês A Internet mais rápida do mercado, agora com chamadas grátis e downloads ilimitados! Saiba mais em http://acesso.clix.pt/
---- election-methods mailing list - see http://electorama.com/em for list info
