NP: Primus, various (Rhapsody)

With recent presidential polls bouncing all over the place, I'm intrigued by this emerging theme, which is basically that traditional phone polling is losing accuracy as more and more people use cell phones.

Granted, I fall into the same biases of my immediate peer group as the next guy, so I've always felt there was something missing in these polls that show Bush to be anywhere from breaking even to leading, and this could be it. I know a lot of people who don't have land lines. This sets up a great story in that younger voters could end up being the "deciding" factor in the election, because no one had any notion of what they were going to to.

This theory resonates because I've seen similar inherent biases in data collection before. When I was working at RollingStone.com, we would always look like we were weak among 18-24 year olds, which was massively counterintuitive. The Onion suffered the same trend, but when you consider that a huge portion of college-aged web surfers use public computers in libraries and computer labs, and these computers aren't counted by Nielsen or comScore, suddently the data collection methodology has wreaked all sorts of havoc on your measurements.

Kos makes the argument that since (a) traditional polls are still useful for changes over time, and (b) younger Americans are the least likely to vote, this effect should be minimal in 2004, but what if it isn't? What if there's a whole groundswell of outrage that's going unreported? Zogby might be catching it with his "untested" methodology, but then again, maybe not.