|
Then the Mason-Dixon numbers (especially at the state level).
With other polls, the problem is more methodology than partisanship. FoxNews, for instance, is obviously "partisan", but their recent numbers don't really reflect it (being one of the few firms to show a definable "bounce" out of the convention).
How you determine "likely" voters, how you balance your sample, how you weight for party affiliation (or WHETHER you weight for party affiliation), how you make the calls (automated or live), WHEN you make the calls, how you ask the questions... etc.etc.etc.
All of these things impact your results (without necessarily being due to partisanship).
Zogby, for instance, runs stunningly hot and cold. He weighs quite a few more factors in his final numbers than most do and often comes out spot-on the final results. Sometimes he's embarassingly far off.
The totaliy of the polls can be worth looking at. Certainly the direction of movement in the polls is useful. But in general they don't mean nearly as much as people would like to think they do.
|