OK it is nearing the end of Election 2008 and depending on what polls you are reading you'll get a different forecasted view. Some of you readers might think it looks a little bleak for my client of the past few years, Senator McCain. So why doesn't there seem to be panic with me? Well I believe some of those polls are crappy and here's why...
- The most important thing about any sampling is the population they pull the sample from. Sure most casual observers can figure out polls based on registered voters are not as good as likely voters, but I go a little further. If I see a NYTimes/CBS poll I don't pay any attention to it. I see a MSNBC I don't even get past the M before I discount it. For me, polls run by MSNBC I wouldn't trust at all because the universe of people they pull from. Same as a survey of New York Times readers. Zogby, Gallup, Reuters, etc are ones that I look forward to reviewing.
- Even when it is a Zogby or Gallup based on likely voters the numbers are very different. Heck maybe the campaigns own internal polling numbers are different. Wonder why? Well maybe you don't know how they are weighting likely voters or what that population looks like. Maybe polls overweight youth voters, or women voters, or Democrats, etc. The numbers based on the sample of likely voters therefore are different because they have their own methodology. Me? I look forward to the Gallup polls because I believe they have the best profile of likely voters.
- Once we get past the national polls which are interesting, but more for looking at overall brand favorability and momentum, I put a lot of time into the state polling. And my favorite state polls are the ones conducted by the local news organizations because I believe they have a better handle on their own universe of voters. The sample problems apply with who the likely voter is in their samples.
Think I'm living in a bubble? Think I'm delirious like other members of the campaign? Why not read this article from an expert who is far away from the campaign? In today's Wall Street Journal Karl Rove wrote the following article called Don't Let The Polls Affect Your Vote. The polls were wrong in 2000 and 2004 and will probably be wrong in 2008. Hence the October Surprise on November 4th.
P.S. Think I've lost my mind? Read this story American Journalism Review
PardonMyFrench,
Eric
Comments