Measuring the electorate vs. projecting it… UPDATED
The simplest way I can sum up the current controversy about the polls is the way I did it on twitter:
they claim party ID of electorate is something they measure, not project. I think '12 will be end of that idea.
Now let's break that down. They (the pollsters) claim that party ID (proportion of Democrats, Republicans and Independents) of the electorate (the actual people who will cast ballots) is something they measure (is a result of their poll), not project (not something they put into their weighting calculations). I think '12 will be end of that idea (when many of the polls are proven inaccurate in this cycle the pollsters will have to rethink their methodologies).
Here's an example of a wacky poll, courtesy of CNN/ORC.
The poll's result is a 49-49 tie among what CNN/ORC believes are "likely voters." Here's an excerpt from the poll's statement on methodology from page 29 of the report:
Among those likely voters, 41% described themselves as Democrats, 29% described themselves as Independents, and 30% described themselves as Republicans.
Whoa, Nelly! Did you catch that? The president is tied with Romney in a poll that contains 11% more Democrats than Republicans. But here's the thing, in 2008 the Democrats had a 7-point advantage in turnout. Without any statistical heavy lifting we can see that if the Dems have the same turnout advantage that they had in 2008 that the president is in trouble according to this one poll.
But wait, there's more! NOBODY and I mean NOBODY expects the 2008 turnout ratio to hold in 2012. Simply put, the Republicans are going to turn out somewhat better than they did in 2008 and the Democrats are going turn out somewhat less. The exact numbers are anybody's guess but you can bet your sweet bippy that the difference between the two parties on election day is going to be less than 7 points (there's numerous data points out there in the polls about enthusiasm, etc. to back up my claim in addition to the anecdotal evidence we are seeing).
And here's the kicker. Independent voters are going to be a large segment of the electorate and in this poll Romney is winning among independents by a margin of 59-37 (that's a whopping 22-pt spread with a margin of error of +/-6.5%).
The ONLY reason this poll even LOOKS close is because of the incredibly lopsided number of Democrats the poll captured. If party ID were something that the pollster projected and weighted for (like they do with geographic distribution, race and other demographic factors) it would show Romney winning because the most they could legitimately project is a D+7 turnout as in 2008 and even that would be very, very questionable based on the facts on the ground.
I would argue that the results of this poll are accurate. It's what the pollster did with them (actually didn't do) that are creating an inaccurate picture. There is simply now way that Obama could be tied with Romney if Romney leads among independents by 20+ points.
UPDATE: This morning Ed Morrissey at Hot Air reaches the same conclusions about this CNN poll that I reached late last night:
So we are expected to believe that since 2008, (a) Obama has lost thirty points in the gap with independents, (b) Obama has lost fifteen points in the gender gap, and (c) Obama is still just four points below his 2008 share of the electorate? Only in a world where 41% of the voters will be Democrats and only 30% Republicans, and that world won’t be what we see tomorrow.