WaPo Poll, Say that three times real fast

WAPopoll-WAPopoll-WAPopoll-WAPopoll It's the sound my wife's car was making the other week as she drove it up to the house with a flat tire. I explained about driving on a flat... again. But what the heck, the rim was ok, and at least the flat tire didn't smell of fish.

The other WaPo poll does smell somewhat of fish. It's the America Wants Out of Iraq poll from the Washington Post/ABC. What's fishy, the main thing anyway, is the number of Republicans vs the number of Democrats in the polling sample. A 2002 Gallup poll shows the nation pretty much evenly divided between R's, D's, and I's. 32% say Democrat, 31% Republican, and the rest independent, unsure, unresponsive, or asleep. And that is pretty much in line with what I've always heard, that we're divided among Republicans and, you know, the other guys.

So why are there usually more Democrats in the WaPo poll:

clcikabiggen

Notice how many R's vs D's? Seems the pollsters just couldn't get Republicans to answer the phone. But a partial clue is at the top of the results:

This Washington Post-ABC News poll was conducted by telephone September 27-30, 2007, among a random national sample of 1,114 adults, including additional interviews with randomly-selected African Americans, for a total of 212 black respondents.

What struck me was, why the extra interviews specifically of African Americans? I can see it if they found they had an under-sampling of blacks; they could include extra interviews to bring the overall sample up to the percentages found in the general population. And that would make sense. But their extra interviews bring the percentage of blacks up to 19%. Why a higher percentage of blacks than the general population? Could it be because most African Americans are Democrats?

Polls are of pretty limited value in the first place, but when you stack the deck, they become pretty much useless.