Final poll numbers ranked by 2011 success. Eligible/decided voters
Forum (Jun 11) Libs 41v... PC 35=... NDP 20^... Grn 3v
Abacus (Jun 11) Libs 35^... PC 32^... NDP 26v... Grn 6^
Angus (Jun 10) Libs 36... PC 32.... NDP 26... Grn 5
Ekos (Jun 11) Libs 37=. PC 31^... NDP 19v.. Grn 8^
Ipsos (Jun 11) Libs 33v... PC 31v... NDP 30^... Oth 5^
There seems to be a real divide by the different formulas. The IVR/telephone polls (Forum, Ekos) has the Libs up by 6 and the NDP in the low 20s. The online eligible polls (Abacus, Angus, Ipsos) seems to have a slight Liberal lead by 3 and the NPD over 25. But their likely voters formula seem to not be able to agree with each other that it's everything from tie, small PC lead and large PC lead. Which kind of makes me wonder about the credibility of the likely voter system any of the three online/PC leaning can't agree by how much. My Final data analysis of eligible/decided voters Libs 38, PC 33, NDP 23 Grn 5. If you looked at the "likely" voter numbers I got Lib 36. PC 36, NDP 23 and Grn 5.
The same formula I did with averaging percentage, I just did with projection sites. I took 5 (supposedly) independent projection site. Ranked them from their 2011 projection success and average it out with weight on the more successful site (I think some will be surprised who had the most success in 2011) The data analysis of the 5 projection site come out for me in seats. Lib 48, PC 38, NDP 21
Projection sites ranked by success of 2011
Election prediction: Lib 48, PC 39, NDP 20 (they had a great total seats success in 2011, off by just one seat, giving PC one seat that the Liberals actually won)
Cndian elxn almanic: Lib 46, PC 38 NDP 23
Lispop: Lib 49, PC 38, NDP 20
308: Lib 49, PC 36, NDP22
2close2call: Lib 47, PC 38, NDP 22 (I ranked them last because it's their debut this election)