Lies! Damn Lies and Statistics!

As an avid reader of the news, I've recently taken to following the constant train of polls on this election. Now polling is tricky. Some polls, like the national McCain v. Obama or McCain v. Clinton polls are certainly meaningless at this point in the election. And other polls that should have been meaningful, like the NH and IA polls, just plain sucked. But at least they sucked consistently. What drves me nuts is the polls coming out of PA. For those of you not quite as crazy as I am on this, here is a rundown of the last few days of polling in PA, courtesy of pollster.com (the table is date/ sample size/ Clinton/ Obama/ undecided):

InsiderAdvantage	4/8/08	681 LV	48	38	13
PPP (D)	4/7-8/08	1124 LV	46	43	11
Rasmussen	4/7/08	730 LV	48	43	9
SurveyUSA	4/5-7/08	597 LV	56	38	2
ARG	4/5-6/08	600 LV	45	45	6

So let's see Clinton is up by 10 points, 3 points, 5 points, 18 points, or not at all. Now look: I'm as big of a fan of the margin of error as anyone. But these are absurd! Especially since the "undecided" column really doesn't differ that much between the tied poll and the 18 point lead poll, nor between the ten and three point poll.

I think maybe all of Pennsylvania is just fucking with us. Like they had a big meeting and they said: "hey y'all. Know what would be hella-tight? If, like, a fifth of the state decided to answer pollster's questions differently just for the hell of it. That'll keep 'em guessing!"

(So I don't know what PA slang is like. So what?).