Wednesday, April 11, 2012

Lying With Numbers

Recently the ABC News/Washington Post poll came out showing President Obama with a comfortable lead, particularly among women. What do you suppose would be the impact on the results if it turned out a healthy majority of those polled were Democrats?

You think the results would be biased? You'd be correct. Jay Cost of The Weekly Standard shows the sample was biased in Obama's favor:
The poll has an inexplicably large Democratic advantage – the party breakdown in the poll is 34 percent Democratic, 23 percent Republican, and 34 percent independent. As a point of historical comparison, the party spread in four of the last five elections since 2002 has basically been an even split between the two sides. In 2008, a “perfect storm” of bad news for the GOP, the party ID advantage was “only” +7. So, a Democratic advantage of +11 is an unjustifiable number, at least in terms of what the electorate is thinking.
As Mark Twain once said, "figures don't lie but liars figure." Take the WaPo numbers as more of an indication of what they'd like to be true than as showing what the electorate is feeling.