Enough with this Post Poll

The Washington Post ran another story yesterday on how their poll shows immutable evidence of something:

Former NAACP president Kweisi Mfume leads U.S. Rep. Benjamin L. Cardin in what is shaping up to be a racially polarized Democratic Senate primary in Maryland, even as roughly a third of the electorate has not settled on a candidate, according to a new Washington Post poll.

This is the same poll the allegedly has O'Malley up 19 points. The problem is that we really don't know that to be true or not. Here is the key sentence:

The Post's telephone poll was conducted among 902 registered voters, including 494 registered Democrats, and has a margin of sampling error of plus or minus three percentage points. The margin is four points for questions about the Democratic primary.

On top of that, the poll was supposedly done over a nine-day period, during the timeframe when Duncan dropped out of the race. Why is all of this a problem? Traditionally accurate polls poll the number of likely voters, not registered voters, over a three-day period, not a nine-day period. The Post's methodology is a shaky, and because of it its results are as unreliable, as the Sun poll that was so much maligned last year.