With a large enough universe, a good enough screening process, and reliable enough samples, polls can yield accurate results. This is particularly true in higher-profile races where candidates (and their positions) are well known. In local races, of course, the obverse is true. It makes polling much more difficult when the universe is unfamiliar with the candidates and/or the office(when I ran the Sheriff’s race, an early poll we did had Undecided leading the pack. Ditto in my council fight).
The polling thus far in the presidential race has been a mixed bag. The record turnout in most states has helped confuse the pundits (obviously the sample size used in polling becomes less reliable when more people vote). Yet some trends have been accurately predicted and analyzed.
The general consensus for the next “Big 3” Democratic contests predicts that Clinton will win PA, Obama will win North Carolina, and Indiana is a toss-up. As a Clinton supporter, I am heartened that the website par excellence Real Clear Politics gives Hillary a lead of 4.7% among an average of all of the recent polls in Indiana. After the IN and NC double-header, 7 final states and territories will vote and each state will be seen in play (although certainly some favor one candidate over another). The polls in each state yet to vote will certainly matter and offer the punditocracy fodder for the approximately 3,752 daily tv shows devoted to the 2008 campaign.
Yet, most reputable and large polling organizations are still polling a national Democratic primary race. In fact almost daily, Matt Drudge (and countless imitators and detractors) breathlessly announce the latest numbers. Let’s look at today’s Gallup Daily (the gold standard): Obama Maintains Lead Heading into Debate.
This is perplexing to me for one key reason: it does not matter in the least.
46 states and territories have already cast ballots in the 2008 Democratic presidential primary. The next time that they vote in a Presidential election will be the general election on November 4th. It is completely irrelevant as to who they support now, as they have already had their say in the matter. You cannot change your vote, nor vote again. It’s completely appropriate to poll each state yet to vote and/or conduct a polling among all those states that have not yet had their say; but these national polls are meaningless.
I remember early on in President Bush’s first term (pre-9/11), some polling firm found that significantly more people supported Al Gore for president (more even than his popular vote victory). This may have been nice (or crushing) for the ego of the former Vice President, but it served no purpose; President Bush had already been inaugurated.
There are certainly some Democratic trends to poll nationally, such as the reported quarter of Obama or Clinton voters that will defect to the GOP should their preferred candidate not win the nomination (that number will drop precipitously once the nominee is chosen and the hard feelings begin to thaw). Yet, I still do not understand how else a national Democratic head-to-head match up can be polled, other than as a means of giving Chris Matthews something to foam at the mouth about.
I, and the vast majority of Democrats around the country, have already voted. Our preference was declared when we cast our ballot. If our preferred candidate has changed now, our vote hasn’t.
Cross posted at: http://www.matt-omalley.com
I have actually seen Clinton comfortably ahead in IN, and ditto for KY and WV; Obama is ahead in NC and OR. She is still favored for PA as well. I tend to agree that national polls are virtually meaningless, though they might sway any remaining undecided superdelegates. These guys will undoubtedly poll nationwide for the general election regardless of the fact that it is really 51 state elections.