Tuesday, September 09, 2008

For everyone who is panicking over polls

Please also read this:

But, interestingly, all three polls were also conducted using a higher sampling of Republican voters than in July, thus raising a question of methodology.

In a year in which Democrats have a lead of 11 million registered-voters over Republicans, and have been adding to that advantage through a robust field operation, are pollsters over-sampling Republicans?

While McCain is definitely getting a bounce in the polls, pollsters seem to be--at the same time--increasing the number of Republicans that they poll and decreasing the number of Democrats.

Now weighting polls is standard operating procedure, but making this change, while McCain is getting his convention bounce is amplifying the magnitude of this bounce. What is less than clear is if this is a real effect, or artificial.


Lisa Colton said...

Do you know if they include cell phones, or adjust sampling to make sure they are getting all age brackets? Could the polls be leaving out the 25 and under set who don't have landlines, but will likely vote democratic?

Sood said...

They generally don't--estimates are varying on how much they are underestimating Obama's support--estimates seem to be a few points as a result.

Also, "likely voter" screens, by definition, underestimate new voters which should pretty obviously favor Obama.