Which Swing States Were Toughest to Poll?

Picking at some remaining morsels in our pollster rankings, I wanted to set out which of the 17 designated swing states were easiest to poll and which ones were the hardest to poll. Here is a list of all 17 states in order of T2L Pollster Score, starting with the lowest -- or best -- scores, and working our way down.

(1) Missouri (3.16)
(2) North Carolina (4.12)
(3) Georgia (4.2)
(4) Virginia (4.48)
(5) Colorado (4.56)
(6) Ohio (4.74)
(7) Montana (5.3)
(8) Pennsylvania (5.62)
(9) Indiana (5.65)
(10) New Hampshire (5.69)
(11) Florida (5.7)
(12) Michigan (5.97)
(13) New Mexico (6.07)
(14) Arizona (6.58)
(15) Minnesota (7.23)
(16) North Dakota (8.24)
(17) Nevada (8.52)

Remember: the score provided represents how close all the final states polls were to the two candidates' final percentages each, combined in that state. Therefore, the lower the score, the closer the polls were on average.

This list is both interesting and telling. Missouri and North Carolina had the best polling, and not coincidentally, they were the two closest states in the country. Everyone seemed to agree the two would be tight, and that sentiment turned out to be correct. Georgia, while not as close as the other two, was tighter than usual, with Obama hitting around 47% in the end. The polling here too was strong.

I want to take special note of Colorado and Virginia. For one, these two states are similar in that they were both strongly controlled by the GOP as late as 5-6 years ago, and today both have undergone nearly complete political transformations which have led the Democrats to holding most of the key offices in the states. Obama's wins in both may have been the final culmination in turning the two light blue.

What is interesting about both of them is that Obama held early, but very small leads in both. However, by around late-September we began to see him pulling away in Colorado as well as Virginia. Given that Virginia had not voted Democratic in a presidential since 1964, and Colorado had been reliably red too, this certainly stood out, but the pollsters were resolute that Obama was indeed opening up sizable leads in each, and sure enough he ended up winning both decisively. They turned out to be right.

The same thing applies in Ohio, the classic swing state, with the only difference being that many pollsters in October did not accord Obama 50%+ support, not buying that he was there yet. When Obama finished above that magic line, I think it was reflected in the T2L Scores for Ohio. The polls were pretty good, but not as sharp as in Virginia and Colorado.

While Indiana was about as close as Missouri and North Carolina, its fairly high score reflects that a lot of pollsters did not believe that Obama would get to 50% (he ended up at 49.9). Only about a third of late polls had him hitting 50%.

The rest at the bottom are interesting, and I think they reflect what we were talking about regarding the difficulty of polling blowouts. New Hampshire, Michigan, New Mexico, Minnesota, and Nevada were all inital swing states that ended up as blowouts. A lot of firms did not seem to believe that Obama would end up with the wide wins in these states that he got. Michigan is a little more understandable since there was a bit less polling done there after McCain pulled out. But it was certainly the case in Minnesota.

Take Nevada. This was the worst swing state for the pollster, and I think the Hispanic vote may have had something to do with it. In 2004, Latinos made up 10% of state voters. This year, that number shot up to 15%, and the percentages of blacks in the voting pool going from 7% to 10%. I don't know for certain, but I wonder if pollsters expected those kinds of drastic jumps in the Silver State.