I don’t pay much attention to the 24-hour news stations, and I frankly haven’t heard directly all that much questioning of the current national polls. But apparently there have been many people — especially Romney supporters — saying that the partisan breakdown of the polls has skewed the numbers.
But the partisan breakdown — the number of people who say they identify with a certain party — is itself determined by the polling questions and typically is not weighted differently by the pollsters. The number of Americans who identify as Democrats has generally been higher than those who identify as Republican; that has been the case even in years when Republican presidential candidates have won.
Nate Silver has a great discussion of this at FiveThirtyEight, Poll Averages Have No History of Consistent Partisan Bias:
Pollsters will re-weight their numbers if the demographics of their sample diverge from Census Bureau data. For instance, it is typically more challenging to get younger voters on the phone, so most pollsters weight their samples by age to remedy this problem.
But party identification is not a hard-and-fast demographic characteristic like race, age or gender. Instead, it can change in reaction to news and political events from the party conventions to the Sept. 11 attacks. Since changes in public opinion are precisely what polls are trying to measure, it would defeat the purpose of conducting a survey if pollsters insisted that they knew what it was ahead of time.FiveThirtyEight’s current state-by-state predictions based on its model that aggregates all available polling.
From Political Perceptions: Can the Polls Be Trusted? in today’s Washington Post:
By polling too many Democrats, the critique goes, polls are finding an overly large number of Obama supporters.
We’ve seen similar hullabaloos before. Democrats screamed loudly in the fall of 2004 when polls began to show President George W. Bush opening up a lead over Sen. John Kerry, a shift in sentiment many liberals found hard to believe.
In the end, the rolling average of national polls—as tallied on realclearpolitics.com and similar sites—nailed almost exactly the actual spread once all votes were tallied on Election Day. The same aggregate of polls also offered an uncanny prediction for the final tally in 2008.
As Silver notes, his statistical model’s final average of national polls in 2008 gave Obama a 7.3 point lead — the exact margin of victory.
Any single poll might show serious flaws or be a statistical outlier. But taken together they have a strong track record.
Silver’s model currently predicts that Obama will beat Romney by 4.1 points, although if the election were held today, the model predicts 5.9 points.