Why Not Belive The Los Angeles Times/Bloomberg Polls?
The Los Angeles Times/Bloomberg polls released today had both good and bad news for the GOP.
Talent and Corker are up in Missouri and Tennessee, for example, with Kean well within the margin of error. According to this poll DeWine is behind by 8 but there is a large number of undecideds and a margin of error of four points.
But here is why the poll (and most others) don’t deserve much credibility.
The Times Poll contacted 3,630 adults in five state samples – Ohio, Missouri, Tennessee, New Jersey, and Virginia. Ohio sample has 653 registered voters, including 507 voters deemed likely to vote; Missouri: 768 registered, including 593 likely voters; Tennessee: 666 registered, including 515 likely voters; New Jersey: 495 registered, including 314 likely voters and Virginia: 529 registered, including 385 likely voters. Among all registered voters in each state, the margin of sampling error is plus or minus 4 percentage points; and for likely voters in Ohio, Missouri and Tennessee it is 4 points; for Virginia it is 5 points and for New Jersey, it is 5.5 points. Likely voters were determined by a screening process which included questions on intention to vote, certainty of vote, interest in the campaign, and past voting history. All interviews were conducted by telephone October 20 through 23, 2006. Telephone numbers were chosen from a list of all exchanges in the nation, and random digit dialing techniques allowed listed and unlisted numbers to be contacted. Multiple attempts were made to contact each number. Adults were weighted slightly to conform with their respective census figures for sex, race, age, and education, and in New Jersey, by party registration. For certain subgroups in all samples, error margins may be somewhat higher. Poll results may also be affected by factors such as question wording and the order in which questions are presented. Telephone interviews in New Jersey and Virginia were supervised by Interviewing Services of America, Van Nuys.
There is no reason not to release the specifics of the cross-tabs and the turnout model. None. Not to do so is tohide the ball from the professionals who could then examine the underlying assumptions.
In this day where online publishing removes the old argument about space limitations any poll that does not provide detailed information on the methodology concerning the sample and the model is immediately suspect.
Just publish the data and stop with the secret sauce. No wonder the Los Angeles Times is cratering. The paper has a huge credibility problem with its readership, so it addresses that problem by keeping secrets about its late-in-the-game polling.
Will old media ever wake up to the fact that nobody believes anything about them or their methods or motives?