Had all polling firms in 2012 accepted the PARTISAN SELF-IDENTIFICATION of the voters, their results would have been more accurate. Same thing for 2014. The problem appears to be the modeling that most all firms used. A reminder: many of the margins were off. But 8 of 10 aggregates for the marquee races in the Senate still correctly pointed toward the winner. And one of the two aggregates that was off (NC) was off by 2.8 (currently), which is well within the MoE. That being said, off still means off, only, in a closely contested race, one can almost forgive this one. The one aggregate that showed a statistical tie but pointing toward the Independent and the margins were way off was: KANSAS So, when we are talking about firms being off, we are now talking about the actual calls, we are talking about the margins. And we are not talking at all about the 26 "safe" Senate races that also happened, nor are we talking about the polling for the 435 HOR races that happened. However, the marquee races are the ones where we measure voter intensity, and rightfully so. But just a foretastes of how my analysis will be in about 3 months: NC, PPP (D), final poll: Tillis +2 Current end result, NC: Tillis +1.7 PPP = bullseye NH, UMass and Boston/Suffolk polls: Shaheen +3 (but these polls are just outside the 7 day timeframe and were not in the end-polling aggregate) NH, PPP (D) final poll: Shaheen +2 Current end-result: Shaheen +3.1 UMass and Suffolk = bullseyes PPP = off to the RIGHT by 1. Close to a bullseye, but no bullseye PPP came the closest of all the end-polls. AR, The Arkansas Poll: Cotton +13 AR: PPP (D): Cotton +8 Actual margin at current: Cotton +17 Even the best poll, from a hometown pollster, was off by 4 points to the LEFT. And that was the best pollster. PPP, the second best pollster of the end-polls, was off by 9 points. That is inacceptable. What a mess. KS, PPP (D), final poll: Orman +1 Actual result (current stats): Roberts +10.4 PPP (D) was off by 10.4. That is absolutely inacceptable. Of the 5 end polls, according to the strict time frame of 7 days (this is the usual method), not a single one got closer than 9.4 points away from the truth. They were ALL off. I just quoted PPP (D) four times to show that even a very good, solid pollster like PPP, which nailed 11 of 12 battlegrounds in 2012 and 22 of 23 states overall in 2012, had at least one bullseye, was off to the Right by a little and in at least one poll, was off to the LEFT by a lot in at least two polls, and I mean, a lot. We will discover that for this cycle, even some of the best hometown pollsters were off. Not all, but some. This is going to be a fun one to analyse.