Friday, October 17, 2008

Polls and Stuff

Lately I had been feeling a bit down because of the poll numbers showing Obama so far ahead. I mean, I know that you can twist polls any which way but they all seemed *so far* ahead as to even dispute that.

Then I read an excellent discussion on polls.

At least now I can have HOPE that there will be a CHANGE from what the polls predict to what actually happens.

That part that impressed me the most is that the sample group for the polls is mightily out of whack. Instead of asking 50% Democrats and 50% Republicans who they would vote for they instead give things a twist. For instance if Ohio had 35% of people saying they were Democrats in 2004 and 40% saying they were Democrats in 2008 then it only makes sense that this trend should continue and that the sample pool for the 2008 poll should have 46% of those questioned be Democrats. Right? So in this case, if they had 1000 people, 460 of them were self-identified as Democrats, 330 self-identified as Republicans and 200 self-identified as Independents. So the 460 Democrates vote for Obama and the 330 Republicans vote for McCain and the 200 Independents split. The results show 56% of the vote going for Obama and only 43% for McCain, or thereabouts, Obama over McCain by 13 points! Except that the very basis, the pool of participants, was stacked in favor of Obama.

A poll came out today with a headline of Voters souring on McCain, Obama stays steady. Read the article and it sounds pretty conclusive. Click on the Full Poll Results(PDF) link and skip all the way down to page 6. "If the 2008 general election for President where being held today... who would you vote for?" 42% Obama, 44% McCain. A statistical tie.

So why all these headlines of Obama being so far in the lead? Simple insurance. In the case that McCain wins the sheeple who have believed that Obama was and had been clear and away the leader will all then buy the allegation that the election was stolen.

1 comment:

Evan said...

Have you seen this analysis from Zombie yet? It's pretty good, I think. It goes deep into some of psychological reasons behind the polls looking the way they are, and the reason this isn't necessarily a good thing for the Obama camp.