For those who don't know, most of what I write here (but not all, so keep reading here) is cross-posted at Race42008.com -- a great site for following polls and election-related commentary (/plug). In the comments over there, BobH -- who has been posting at sites I've been writing and/or posting at for four years -- writes with regard to my post about the movement in the tracking polls below:
Sean: I find it interesting to compare the results of the LV and RV polls. Using the polls currently in the RCP average, the three RV polls have Obama leading by 11, 7, and 6 (an average of 8). The LV polls have the lead at 6,2,1,4,3,8,3 (an average of just under 4). The question appears to be how accurate the likely voter screens are this year. If McCain is really within 4 in mid-October, then this thing is very winnable. If he’s behind by 8, things are a good bit dicier.
That's spot-on, and I hadn't noticed the pattern (and had completely forgotten that Gallup Tracking is an RV poll). It also brings up some good points for discussion.
As I've said before, the million dollar question this cycle is what the electorate will look like. No one really knows for sure, and I think it is less certain than even 2004 (when increased Kerry enthusiasm was supposed to lead to a blowout). I have heard all the arguments for why turnout will favor Obama this time around, and I credit them (except for the "cellphone effect" argument).
But there are also countervailing arguments. In 2004, the improved Kerry GOTV, vote registration machine was about as hyped as the Obama registration and GOTV machine. It succeeded -- but so did a much less hyped Republican GOTV machine. In short, Kerry met targets that the campaign thought if they met would make it impossible for them to lose. The GOP did better.
So when I see analysis indicating that early voting is not as record-breaking in Ohio as many would expect, I start to wonder about the polls. And when I note that Party ID among the electorate has been astoundingly level over the past twenty years, in elections that have been good, bad, and ugly for both parties, I wonder how much I can expect them to be different this year:
In 1988, Democrats had a three-point party ID advantage over Republicans (38-35). In 1992, Democrats still had a three-point party ID advantage over Republicans (38-35). In 1996, that advantage increased to four - a shift of one point (39-35). In 2000, Democrats were steady, up by four (39-35), and in 2004 they dropped to even (37-37). . . . But note that in 2006, when Democrats clearly found enormous success at the ballot box, that the advantage in party ID was only three points (38-35).
Yes, Obama has registered tons of voters. As I note above, that is an argument to certainly credit in his favor. On the other hand, likely voter screens typically ask if you have voted before for a reason. Every cycle we hear about how likely voter screens are going to be off because of an anticipated surge in Democratic turnout, and it is usually wrong, even in very good Democratic years. They tend not to be. It is one thing to meet a college kid on Franklin St. and get him to sign up to vote; it is another to get him to show up. I agree wholeheartedly that if there is a year when these voters are going to turnout, this is that year. But I also would have said 2006 was arguably that year, and young voters actually made up a smaller portion of the electorate than 2004.
And yes, Democrats had great turnout in the primaries. The likes of which have not been seen since 1988 and 1984 (the previous record-breaking years).
So the question is this: Are the likely voter polls already anticipating a surge in Democratic party ID? Some certainly are. It is suspected that there reason Battleground went from showing McCain winning to showing Obama winning is that they re-weighted from their earlier D+3 electorate. What if, as in 2004, the ballyhooed Democratic turnout machine doesn't materialize, or only adds a point or two to the electorate? How many of those new Democratic voters were Democratic-leaning independents, or voters (either R, D, or I) who registered to vote AGAINST Obama?
On the other hand, what if these likely voter polls are anticipating a 1996 electorate, and we end up with something much different? What if Hotline/FD's +5 Dem electorate (with Obama leading by only 2 in what would be the most heavily Democratic electorate in almost a generation) is actually generous to Republicans?
The million dollar question continues to revolve around what the makeup of the electorate will be. If the likely voter screens still work and McCain is only down 4, I'd feel pretty good heading into November. If the screens don't, then we're really dun-zo.