Tuesday, October 24, 2006

Elections and Polling

The Democrats get surprized every year, because most polls overstate their support by 3-5 points. Rasmussen and Zogby tend to be closest; ignore the others. The Republicans are going to hold the House and Senate, despite the fact that Nancy Pelosi is already eyeing the drapes and carpets at the Speaker's office.

To greatly oversimplify what happens, the MSM overlooks the basic numbers. About 30% of Americans consider themselves conservatives of one stripe or another, 20% liberals. The remain 50% also split 30-20 in favor of conservatives, though single issues, personalities, and the weirdness of general events can affect them more powerfully. The play of the MSM on those personalities, single issues, and weirdnesses moves a lot of easily-misled people leftward, creating the appearance of an evenly divided country. That's Newsweek's estimate, not mine, BTW. They estimated in 2004 that the dominant media was worth 10-15 points overall.

Democrats gain only when the numbers are artificially suppressed, by biased reporting, bad polling, or dumb luck. Without the constant reporting of how badly Iraq is going, and the constant ignoring of how well the economy is doing, the electoral slides back toward its natural divide of 60-40. The Foley scandal artificially suppressed the Republican gains which started in September. Unless the Democrats have another October surprize, the points will creep back.

How the change in media fits into this I don't know. There is rampant speculation on the blogosphere that this is all poised to change soon, but the results aren't in. And when I traveled last week, or when I sat in the hospital waiting room today, the news available is all MSM. Nightly network news and major daily papers remain the default position for credibility, no matter how many times they get shown up.

1 comment:

cakreiz said...

The self-identification polling numbers I've seen are 33-20% (cons to lib), and have remained constant over the last 30 years. This seems to suggest a structural bias for conservatives. Some, including Tigerhawk, question the reliability of these numbers. I've asked myself how these self-identification stats stack up against party registration, where Reps and Dems are more evenly split. I tend to agree with your view that there is an advantage to more conservative candidate (Dem or Rep- usually the latter), although I'm not sure how large it is.