Saturday, May 5, 2012

How Rasmussen and Gallup distort national popular vote polling

I'll admit this article gets pretty wonkish on the topic of understanding political polling. But I think its worth talking about so that we can understand what's going on. I hope you'll bear with me and take a look.

People are starting to notice a wide discrepancy about how this presidential race looks depending on whether you focus on the national popular vote polls or the electoral college. Here's some examples:

Real Clear Politics

Popular vote polls: Obama +3.2
Electoral college: Obama 253 Romney 170

Huffington Post

Popular vote polls: Obama +1.2
Electoral College: Obama 288 Romney 170


Popular vote polls: Obama + 1.4
Electoral College: N/A

So the question arises whether or not we're currently seeing a big Obama lead or a neck-and-neck race with Romney.

The folks at electionate have an interesting response to that question. Keep in mind that the three sites I linked to above are showing averages of all polling. The popular vote polls are national samples and the electoral college projections are based on polling in individual states.
But wait, polling says the race is close! Well, maybe. I understand why you think the race is close. After all, Rasmussen and Gallup say so every day. But for the moment, let’s ignore Rasmussen and Gallup, and look at the rest of the polls since March 1. During that period, 25 national polls by groups other than Rasmussen and Gallup found Obama leading 48.16 to 43.48. Of those 25 polls, just two found Romney leading. Nearly half – 11 – found Obama leading by more than 5, and one fourth – 6 – found Obama leading by more than 8%. Rasmussen and Gallup should not be ignored, but the frequency of their polling shouldn’t overwhelm the other data, either.
In the rest of the article they point out some of the questions that have been raised about the methodologies of both Rasmussen and Gallup. But the issue is that both of them do DAILY national tracking polls. So when they are included in polling averages, they outweigh the information from other polls. Since no one does daily tracking polls in states, the averages are more distributed between various pollsters...thus the discrepancy.

No comments:

Post a Comment