On the Polling and Political Wrap, I'd argue that (and this is a conservative estimate) a quarter of the comments emanate from some criticism of a particular poll based on the demographics inherent in that poll.
This is a fairly common practice for political junkies like us, and it almost always accompanies data that, quite frankly, forecasts outcomes that the critics of the data really do not like.
Looking at a number of recent polls in the 2010 cycle, we can see an ample reservoir of eye-popping demographic assumptions that could fuel these kinds of criticisms from the Left.
Do these samples have demographic characteristics that are markedly different from previous cycles?
Yes.
Do those palpable differences seem to favor more positive outcomes for Republicans?
Yes.
Does this mean that the "books are cooked" to the detriment of the Democrats in this year's polls?
Actually, I'd doubt that a great deal. Clearly, pollsters are making very clear assumptions about who will comprise the 2010 electorate. Like any assumption, it could be wrong. But they could just as easily be right, and while comparing the demographics of 2010 polling to previous cycles might feel like solid evidence, it is worth remembering that the polls in 2008 and 2006 didn't look like their predecessors, either.
Curious Demographic #1: The Enthusiasm Gorge
Our friends and polling partners over at Public Policy Polling are often described in the press as "Democratic pollsters", because they do have a handful of partisan clients (including Alan Grayson). However, one of the most appealing things about the firm is that it is evident that they play it straight, which has actually led to some pretty hilarious allegations of right-wing bias.
PPP, as a matter of clarifying their samples, asks a question that few pollsters ask, but is incredibly useful: they ask respondents for their 2008 presidential preference.
As a result, we are able to note that one of the most consistent characteristics of the samples of PPP polls in this cycle is an electorate that was far less pro-Obama than 2008.
And, as Tom Jensen spelled out earlier this week:
If the folks planning to turn out this year matched the 2008 electorate:
*Alex Sink running for Governor in Florida and Alexi Giannoulias running for the Senate in Illinois would have double digit leads.
*Elaine Marshall running for Senate in North Carolina and Pat Quinn running for Governor in Illinois would have small leads instead of trailing.
*Ted Strickland running for Governor in Ohio, Lee Fisher running for Senate in Ohio, Joe Sestak running for Senate in Pennsylvania, and Robin Carnahan running for Senate in Missouri would all be within three points rather than trailing by 7-10 as they do now.
This year isn't getting away from the Democrats because voters are moving toward the Republicans en masse. But the enthusiasm gap is turning races that would otherwise be lean Democratic into toss ups, turning toss ups into leaning Republican, and turning leaning Republican into solid Republican.
Despite the best wishes of some Democratic faithful, however, there is absolutely no reason to believe that the 2010 electorate will look anything like the 2008 electorate. When you go from a presidential election to a midterm election, you are shedding about one-third of that leap year electorate. And, as it happens, a lot of that one-third has historically come from demographic groups that would tend to support Democrats. Therefore, assuming that PPP and other pollsters are being unduly pessimistic because they are not asking enough Democrats could be an errant assumption.
That is why comparing sample demographics in 2010 to the most recent election (2008) is probably not the wisest comparison point to make. History tells us that the demographics are going to be different for a much simpler reason than some grand conspiracy: the polling samples are different because the electorate will be different. However, the next curiousity from 2010 polling I noticed not only flies in the face of 2008 polling samples, but previous midterms, as well. Therefore, this next one is worth a much closer look.
Curious Demographic #2: The Older Vote
Anyone that has ever walked a precinct on Election Day knows that the electorate skews far older than the population at large. If your walk sheet has fifty names and addresses on it, it seems as if half of them are over 50 years of age.
That said, a number of polls in this cycle have shown an electorate that has a surprisingly high percentage of older voters, and a considerably reduced number of younger voters.
Nowhere is this distinction more pronounced than in the recent spate of polls in key House races conducted by GOP pollsters Ayers McHenry on behalf of the right-wing think tank American Action Forum. Those polls, which had Democrats flailing quite a bit in 31 key districts from coast-to-coast, have received a ton of media attention. They also have a demographic characteristic that is dramatically different from the exit poll data from prior midterms.
According to the exit polls conducted during the most recent midterm (2006), roughly 63% of the electorate was aged 45 or over. A look at a previous midterm exit poll (1998) suggests that roughly 12% of the electorate is between the ages of 45-49. Doing a little simple subtraction, then, we could make a fair estimate that, in 2006, somewhere between 50-55% of the electorate should be aged 50 or older.
The Ayers McHenry polls had demographics that were, it is safe to say, skewed a tad older than that. Indeed, of the 31 districts surveyed, twenty-six of them had samples in which two-thirds of the sample were in the 50-or-older cohort. In Arizona's 1st district, for example, only 16% of their sample was under the age of 50.
One thing that the 2006 exit polls made clear was that the older the voter, the more likely they were to support Republicans. Even in heavily Democratic 2006, the 65-and-older demographic was split down the middle between Democrats and Republicans. Therefore, Ayers McHenry's older electorate could easily explain at least part of the Democratic underperformance in those polls.
But, it doesn't mean that they are wrong, as a cursory look at the 1998 and 2006 exit polls would seem to attest (for those wondering why I am ignoring 2002, please note that VNS recalled their exit polls for that particular midterm).
The electorate may well be getting older. Ayers McHenry polled in eighteen different states. Eleven of those states had exit polls in both 1998 and 2006. Of those eleven, the proportion of voters over the age of 45 increased in nine of those states. While that wouldn't begin to explain the chasm between the Ayers McHenry polls and the 2006 exit polls, it might mute the impact of that discrepancy if the electorate simply got older in the interim, which it almost certainly seems to have done.
Does the size of that discrepancy between the Ayers McHenry numbers and the past electorate matter? Indeed, it matters a great deal. If they are off the mark by a handful of points, then even adjusting the age cohorts to appropriate levels would not change the overall trial heat numbers a whole hell of a lot. However, if they really are off by twenty points or more, than their trial heat numbers could be off by an order of magnitude that would really start to matter.
Of course, the obvious drawback is that we won't know if they are wrong until November.
* * * * * * * * * *
There are certainly other quirks to be found in polls throughout the cycle. Whether it is SurveyUSA suddenly finding that young voters loves them some Republican candidates (which flies in the face of recent convention), or a California poll which shows Hispanics embracing a GOP candidate who posed for ads in the primary in front of the border fence, there will always be some statistic for those who want to seek to invalidate the outcome.
Understand that this is nothing new, and in fact is a pastime for the party that appears to be on the short end of the polling stick. Consider some very recent history, courtesy of the Washington Post:
There appears to be an undercurrent of worry among some polling professionals and academics. One reason is the wide variation in Obama leads: Just yesterday, an array of polls showed the Democrat leading by as little as two points and as much as 15 points. The latest Washington Post-ABC News tracking poll showed the race holding steady, with Obama enjoying a lead of 52 percent to 45 percent among likely voters.
Some in the McCain camp also argue that the polls showing the largest leads for Obama mistakenly assume that turnout among young voters and African Americans will be disproportionately high.
And that's not even getting into the whole Bradley Effect debate from 2008, although I must confess that the highlight of 2008 for me may well have been Heidi Montag and Spencer Pratt on some absurd celebrity talk show explaining that their guy (McCain, of course) was gonna win on Tuesday, because people didn't want to tell pollsters that they wouldn't vote for the black guy. Because, y'know, they don't want the pollster to think that they are racist, and stuff.
Conservatives jumped all over polling in both 2006 and 2008, convinced that they had gone under the hood and found that the numbers were all wrong (here is one such example). Indeed, the one recent poll that was decent news for Democrats was immediately assailed from the Right (and noted by Pollster's excellent Mark Blumenthal, who proceeded to execute a devastating takedown of the critique in question).
Could the polls be wrong in 2010? Sure, they could. Pollsters, as always, make assumptions about who will show up at the polls. Those assumptions could be in error. But hanging hopes that the critical mass of polls are in error because of this demographic quirk or that deviation from 2006/2008 is probably a glorified method of "shooting the messenger." Certainly, there's a great temptation to shoot the messengers this cycle--certainly I am guilty of it, and I'd probably be convicted on multiple counts of doing so.
That said, it is infinitely more likely that the route to avoiding a Republican resurgence lies in being able to change the composition of the electorate (i.e. attract base voters to the polls) or by the ability of the Democrats to change a few hearts and minds between now and November. That's the ticket, because counting on the myriad of polling firms to uniformly be off of the mark is almost certainly a faith that will be proven to be misguided.