Obama Surge? Or Polling Error?
Chris Sillizza believes Barack Obama “received a real New Hampshire boost from his win in Thursday’s Iowa caucuses.” His evidence for the surge:
In a new CNN/WMUR poll, Obama leads Sen. Hillary Rodham Clinton (N.Y.) 39 percent to 29 percent, a major change from the 33 percent tie that the same survey showed yesterday. A new USA Today/Gallup poll confirms the Obama bounce, putting him 13 points ahead of Clinton. A new Franklin Pierce/WBZ survey showed a more modest gain for Obama. He led Clinton 34 percent to 31 percent in the latest poll; four days ago he trailed Clinton 32 percent to 28 percent.
Most of these fluctuations are within the margin of sampling error in these polls. Remember, in a poll with a margin of +/-3 percent, 33 percent is really just a shorthand for a range of 30-36.
The outlyer is the CNN/WMUR poll. But I’m rather dubious of a
shaft shift that dramatic in one day. If one looks at the poll results themselves, though, we see that they are using very small samples and tracking daily, resulting in a margin of +/-5 for the Democrats and +/-6 for the Republicans. And that’s with a confidence level of 95%, meaning one poll in 20 is likely to be a complete anomaly.
If one looks at more polls, though, and the trends over time, we see a pretty steady movement towards Obama that started well before Iowa voted.
Here are the most recent polls tracking Democrats, all of which screen for likely primary voters:
And here are the longer term trends:
So, yes, there has been an Obama surge and it has come mostly at the expense of Hillary Clinton. But, no, it doesn’t seem to be a reaction to the Iowa Caucuses.