For a while we heard this election would be a drubbing for the Democrats, with the Republicans taking back the House and possibly the Senate. The pundits and pols call it a “wave” election.
Then something funny happened. People began to look at the Angles, O’Donnells, Bucks, Tancredos, Paladinos, etc. and started thinking “as bad as things are now, will they improve if we elect these nut cases?” The answer to this question is now starting to be reflected in the polls.
There were three mistakes in predicting a “wave” election. First was the election of teabagger wingnuts. Primaries bring out the activists, particularly in off-year elections. Second was the inability of the media to understand what the margin of error means in polling results. If there is a seven point difference between candidates and the margin of error is four percent, the polling results indicate a statistical tie, not that one candidate is ahead of the other. Third is the polling sample population. The statistics used rely on an independent and identically distributed sample population. Pollsters are prohibited from using auto-dialing for cell phone numbers. Few younger voters have land lines. Consequently the sample population of the polls does not match the statistical assumptions (for those interested, see the central limit theorem).
Will the Democrats lose seats? Sure. This always happens when either party wins big in the previous election. But, to paraphrase Mark Twain, the report of their demise is premature.