The Phantom Pollbooth

A CNN/USA Today/Gallup/Starbuck's/Gap/McDonald's poll released yesterday says Kerry is entering the Democratic nation convention pretty much tied with George Bush.  However, this poll, like so many others, reflects the popular vote, and as we all know popular votes don't win presidential elections.  If you don't believe me, ask President Gore.

Taken to it's broadest exaggeration, 50% of the people polled may be voting for Kerry, but suppose that they are distributed such 49% of each state has Kerry supporters, with the exception of one state which has 99% Kerry supporters.  Kerry would win a handful of electoral votes with the vast majority going to Bush.  (generally -- Maine and Nebraska use modified systems to determine their electoral votes)

Prior to the 2000 election, modern politics had little need to notice the difference between popular vote and the electoral college. But with that tight race, and the close race that 2004 is shaping up to be, popular polls based on how 100 average Americans are likely to vote are virtually meaningless. It's close? Thank for the journalism!

The linked poll, by the way, was of 1000 respondents, fewer than 800 of whom identified themselves as likely voters. Why would one count the non-voters' responses?  It seems to me that a more effective tool -- especially for organizations with the resources of CNN, USA Today, and Gallup, would to be to take a slightly larger sample, and apply it to the electoral votes available in each state. It would be a little more complicated, but it would also be a little less . . . you know . . . pointless.