Does it seem like Obama has performed better in the actual votes than in pre-primary polls thus far? Sure it does. But I wanted to see it for myself, in one image.
(Click for a much larger and clearer version, which can be made clearer still by clicking the image that opens in the new window.)
Obama’s results are on the horizontal dimension and Clinton’s are on the vertical. A point in the positive area means that the actual primary vote for the candidate exceeded the final poll average, as reported at Pollster.com.
Most state results appear in the upper right, as we would expect, because polls include an undecided (or third-candidate support, in earlier contests) that splits between the two main candidates in the final vote.
The most striking results are those in the lower right. There are six states in which Clinton’s final vote was less than her polling average, while Obama’s was greater. These include three of the states where Obama exceeded his poll average by more than ten points: Alabama, South Carolina, and Georgia.
On the other hand, there are only two states in which Obama came in under his poll average, and in neither case by even two points: New Hampshire and Illinois. You already know the New Hampshire story–also the state where Clinton most exceeded her final poll average1 –and Obama’s mere 64.6% in his home state as opposed to his 66% polling average presumably does not mean much.
Interestingly, even in much-touted Clinton victories like Tennessee and Oklahoma, Obama did better relative to the final polling trend than Clinton did. In fact, the same can be said about New York!
For whatever it might be worth, as of today, the polling trends in the two big(ish) states2 that vote on 4 March show the following (which I will update on the 29th and maybe on 2 March and voting day before results are known):
—- 49.8 49.5 49.4
—- 42.4 43.4 43.6
(My 28 Feb. error for Ohio has now been deleted.
Updated again on 3 March. New batch of polls may suggest Obama’s momentum in Texas has peaked.
Clinton’s lead in Texas vanished days ago, as the trend line in recent days shows her now dropping even more steeply than Obama had been gaining since the start of the year. In Ohio, the gap has been narrowing significantly recently. Clinton is going to need at least one of these states to turn out more like New Hampshire or California than like so many other states, or she is in real trouble.3
I included only primaries, not caucuses, and if a primary is not included in the graph, it is because there were not enough polls available to have a meaningful average trend.
I should also note that the Pollster.com crew, to their credit, have posted some caveats about the aggregated-average methodology. (I did not find that post on a quick search.)