Back to Basics on Horse Race Polling

I know, I know, I've said this before, but perhaps one last reminder with several new polls on the GOP nomination out today: don’t take national polls too seriously, especially the horse race preference numbers.

Remember that only in Iowa and perhaps New Hampshire will the voters be choosing from among the current field of candidates. Even if the race remains as open as possible, once the first contests are over it will look enormously different from what it looks like now. Recall that at this point in 1984, Democrats were choosing between Walter Mondale and John Glenn; after New Hampshire, it was Mondale and Gary Hart. In fall 1999, George W. Bush and Steve Forbes were getting most of the publicity in the GOP race; after New Hampshire, the choice was Bush or John McCain. All anyone could talk about in fall 2003 was Howard Dean, not John Edwards and John Kerry.

What we’ll all be talking about by mid-January will be a consequence of what happens in the early states along with how GOP opinion leaders react to what happens. What Republican voters will then do will depend only very slightly on what they say to pollsters now. Again, most voters, even many of those dedicated enough to follow nomination politics this early, just don’t know very much about the candidates yet. And for those who are just voters (as opposed to activists who may be giving money or time to candidates), it makes a lot of sense to defer learning about the candidates until later, because there’s no point in assessing Rick Santorum or Michele Bachmann or Rick Perry or Herman Cain when you don’t have any way of knowing which of them will even be still in it when their state comes around.

If you want to know how the nomination fight is going, pay attention to whether party insiders are starting to agree on a candidate, and start glancing at the Iowa and New Hampshire polls (but expect those polls to remain volatile; candidates have only just begun to advertise in Iowa). The national polls are interesting, but they just aren’t very good at predicting how people will actually vote – because they can’t yet know what most voters will be asked to choose once the field narrows and candidates become better defined.