If anything could cause bigger headaches than campaign ads, it's the up-and-down, back-and-forth nature of polls. The conventional wisdom is that Barack Obama has a solid lead against John McCain. Today he's up by more than 7 points in the Real Clear Politics poll averages. (Those numbers feed the Yahoo! News political dashboard.) The Illinois senator has had a similar lead for about a month now. [align=center][b]The blue is a Democratic (Obama leading) state and the red is a Republican (McCain leading state). The states that are blue with dark blue stripes are considered "toss up" states, meaning it's too close to judge who will win.[/b] [img]http://l.yimg.com/us.yimg.com/p/us/news/editorial/1/ed/1ed53d6d33aa6645d7857aa4fe324882.png[/img][/align] [b]For those who want to view a larger electoral map, with the updated poll percentages state by state, refer to this link.[/b] [url]http://news.yahoo.com/election/2008/dashboard[/url] But just as journalists nationwide were starting to write their landslide stories, in comes this week's AP poll ... with McCain behind by just one point. That might as well be a tie. If you factor in the margin of error, then McCain could actually be ahead. If the AP poll was the 1 in a McCain 1-2 punch, then in came the IBD/TIPP Poll with it's left hook. It too had McCain down by 1, with the momentum among key voting blocks going to the Republican. This from the polling firm who says they were the "Most Accurate Pollster" of the 2004 election. Weird thing is, these polls came out at the same time other major polls showed Obama with a double-digit lead. So, what gives? The short answer: The devil is in the details. Polling isn't an exact science. Not even close. It's really an informed estimation game that leans on some speculation about who is showing up on Election Day. Pollsters can't ensure that they talk to a perfectly proportional representation of the people who will vote. So they weigh the responses they get to match the demographics of who they think will actually show up and pull the lever. Confused yet? While the execution can be messy, the concept is pretty simple. We'll use that AP poll as an example. Marc Ambinder at The Atlantic gives a specific example about keeping healthy speculation when debating poll findings: "44% of the likely voter sample in the AP/GFK-Roper poll... are self-identified evangelicals. That's about double the weighted average that pollsters generally assume." Ambinder is saying most pollsters don't assume that many voting people are evangelicals. But that doesn't mean AP is wrong -- no one knows for sure who is showing up on Election Day. Nate Silver at Five Thirty Eight gives the IBD/TIPP poll the same speculative treatment, only in this case Silver questions the number of young people they say plan to vote for McCain: "IBD/TIPP has John McCain ahead 74-22 among 18-24 year olds. Who knew the kids were groovin' on J-Mac these days? IBD/TIPP puts an asterisk by this result, stipulating that 'Age 18-24 has much fluctuation due to small sample size.'" Conventional wisdom (backed by Gallup) says young people will vote for Obama by an incredible margin, so as Silver points out, these numbers don't make sense in a conventional way. While we're on it, there's one more important thing to note on the youth vote. No one knows if the youth turnout is going to be as dramatic as most people say it's going to be. This fact explains another reason for the AP poll discrepancy. AP does not assume a high youth vote on Election Day. And they aren't the only ones. Bloomberg did a story about polling today that quotes Republican pollster Ed Goeas: Some pollsters have an ''unrealistic'' expectation that the youth vote will dramatically increase this year. Since we've got another week until we know who's showing up to vote, what are we poll-obsessed folks to do in the meantime? Well, the answer, as my mother would say, is to be patient and wait for Election Day.

Last edited by Infamous J (2008-10-24 17:18:11)