Noam Scheiber’s recent commentary in The New Republic mentions two terrific points:

1) He points out how some of the polls in the country may be deceptive, pending on what sample population is being used. He mentions a poll in Ohio which had Kerry ahead of Bush by a two point margin. HOWEVER, when the pool was widened to include not just those who voted in 2000 but to include ALL registered voters, Kerry’s lead expanded to a 10 point margin. One would think that this difference would have extra significance for this election as many indications point to voter turnout being greater than in 2000 (highly polarized country, many feel much is at stake, many learned from Florida 2000, etc.). In that case, this and any other poll should likely use ALL registered voters as the more accurate sample pool. One could also make the case that net-net, a person voting this November that did not vote in 2000 is most likely to be a Kerry voter (if they chose not to vote for Bush in 2000, what are the odds they will vote for him in 2004?).

2) Noam unfortunately points out another grim fact: for those states with an anti-gay marriage initiative to be decided, such an item on the ballot will almost certainly motivate more resolute right-wing conservatives to get out and vote than it would more moderate types. Ohio has an anti-gay marriage initiative on the ballot.