Well if you grade it on a curve PPP did a good job on the Georgia runoff since we showed Saxby Chambliss winning by more than any other pollster. But we did, like everyone else, underestimate the size of his victory.
It doesn't take a ton of imagination to figure out why we did that. The Georgia electorate is easily the most racially polarized of any state we polled regularly during the 2008 election cycle. Since we had Chambliss winning by 43 points with whites and Martin winning by 77 points with blacks it essentially meant that every point we over or underestimated the African American share of the electorate by was worth 1.2% on the margin. We don't have an exit poll and I haven't seen any hard data from the Board of Elections yet but my guess is that black voters accounted for only 22-23% of those who turned out for the runoff, consistent with their share of early voting numbers. There was no rush to the polls yesterday as we had anticipated.
Chambliss' overwhelming support from voters over 65 was also a death blow for Martin. He had a 68-31 lead with that demographic and I wouldn't be surprised if that most dependable group of voters ended up accounting for almost a quarter of those who voted yesterday, basically twice as much as their share of the November 4th electorate.
The next question: what were the implications of yesterday's results? I think the only implication is that you can't expect to get a failed Lieutenant Governor candidate elected to the US Senate in a conservative state, and certainly not without a drawing card at the top of the ballot to ensure high turnout from strong Democratic demographic groups. Beyond that I don't think it means much of anything.