So a few days ago I wrote a post about how the polls can be wrong. I showed how over the past ten years, polls have more often than not overestimated Republican performance. And I warned that we usually see the polls underestimate Republicans in conservative states, and we had a lot of close elections in conservative states.
And yeah, we sure did have some polling errors. Errors like nothing we've seen in the past ten years. And almost all erring on the same side of the partisan divide.
Points below the diagonal line show polling averages that were too friendly to Democrats.
When circles are below the diagonal line, the average of the margins of polls overestimated Democratic performance. That is, Democrats did
worse in the election than the polls said they would. For instance, Pryor lost in Arkansas by about 17 points. Even Rasmussen only had him down 7 points. The polling average was off by almost 13 points in total.
Of all the circles on the graph above, only four are above the diagonal line—and only slightly. The rest vary from below the line, to way below the line.
All the races are shown below the fold.
This shows every Senate and governor contest. We see the same picture—almost every contest is below the diagonal line. Why? Failure to turnout? Pollster overcompensating for 2012 errors? Horribly reduced response rates? Hopefully we'll find out as pollsters dig into their data in depth.
Note: these are simple polling averages which include all polls from October 1 to Election Day, unless a clear trend is seen, in which case only the polls from the previous 10 days are used.