National Polls, State Polls, and Democrat Turnout

National Review ^ | November 1, 2012 | Jim Geraghty

[lasereye]

NRO is back this morning! Hurrah!

Also this morning, I’m scheduled to appear on Chuck Todd’s “Daily Rundown” on MSNBC around 9:40 or so.

So, headlining the Morning Jolt is something I wanted to post all day yesterday…

Why Do So Many Pollsters Use Samples Where Democrat Turnout Tops 2008?

One of the theories floating around about why some pollsters show such a wide disparity in the percentage of Democrats and percentage of Republicans in their samples this year is that since 2008, a portion of right-leaning voters who once identified themselves as Republicans now identify themselves as independents.

Thus, the argument goes, Mitt Romney’s not really doing that much better among independents than John McCain did; Romney’s strength is an illusion fueled by some voters who were previously in the Republican pile in 2008 shifting over to the independent pile.

This notion is not hard to prove or dispel. Let’s stop looking at the margin between the Democrat percentage and the Republican percentage. Instead, let’s examine, key state by key state, what percentage of the final vote identified themselves as Democrats in 2008, and see how that compares to some recent poll samples.

Poll results in italics have Democrats with a larger share of the electorate than the party had in 2008.

Colorado 2008 exit poll: 30 percent. ARG: 33 percent. Purple Strategies: 34 percent. PPP: 37 percent. NBC/WSJ/Marist: 45 percent (with leaners, 34 percent without).

Florida 2008 exit poll: 37 percent. Quinnipiac out this morning: 37 percent. Survey USA: 42 percent. CNN: 33 percent. PPP: 43 percent.

Iowa 2008 exit poll: 34 percent. Gravis Polls: 41 percent. PPP: 34 percent. NBC/WSJ/Marist: 45 percent (with leaners, 34 percent without).

Michigan 2008 exit poll: 41 percent. EPIC-MRA: 42 percent.

Minnesota 2008 exit poll: 40 percent. Survey USA: 45 percent. St. Cloud University: 32 percent.

Nevada 2008 exit poll: 38 percent. Survey USA: 43 percent. NBC/WSJ/Marist: 39 percent. Gravis Polls: 45 percent. PPP: 42 percent.

New Hampshire 2008 exit poll: 29 percent. PPP: 30 percent. ARG: 29 percent. University of New Hampshire: 45.6 percent.

North Carolina 2008 exit poll: 42 percent. Survey USA: 41 percent. Elon University: 47.9 percent (with leaners, 38 percent without). PPP: 46 percent (10/25) PPP:45 percent (10/31).

Ohio 2008 exit poll: 39 percent. Quinnipiac out this morning: 37 percent. Survey USA: 38 percent. Purple Strategies:34 percent. PPP: 43 percent.(10/28) PPP:45 percent (10/30) Gravis Polls: 40 percent.

Pennsylvania 2008 exit poll: 44 percent. Gravis Polls: 45 percent. Muhlenberg College: 46 percent. PPP: 48 percent. Quinnipiac: 39 percent.

Virginia 2008 exit poll: 39 percent. Quinnipiac out this morning: 35 percent. Purple Strategies: 36 percent. Gravis Polls: 41 percent.Washington Post: 31 percent.

Are pollsters projecting Democrats will make up more of the electorate in 2012 than they did in 2008? In quite a few cases, the answer is yes. Not every pollster has Democrats exceeding their 2008 share of the electorate, but you do notice Public Policy Polling, Survey USA and Gravis Polls cropping up on this list again and again. (In case you’re wondering, I didn’t list Rasmussen because they only release their internal party split to subscribers.)

Now, the polling profession’s defenders will insist that they don’t weigh poll results by party, they never have weighed poll results by party, and they’re never going to start weighing poll results by party, because we can never know what the partisan makeup of the electorate is going to be before Election Day.

We may not know the exact composition of the electorate on Election Day, but do have a historical perspective and a ballpark sense of what is realistic. If you consider 2008 to be the perfect storm for Democrats, and the “ceiling” for the percentage of voters who self-identify as Democrats, then most of the poll samples in swing states in recent weeks will strike you as weighed far too heavily in favor of Democrats.

Then again, perhaps more voters identify themselves as Democrats today than they did in 2008, when the country was gripped by Bush fatigue, Wall Street was melting down and taking the economy with it, the country was swept up in the euphoria of hope and change, the historical nature of voting to elect the first African-American president, and so on. I haven’t seen a really compelling argument as to why that would be the case, but it is theoretically possible.

The news isn’t all good for Republicans, however. Quinnipiac’s samples from Wednesday morning aren’t as bad as some on the Right might presume, although their samples does presume Democrats make up as much of the electorate as they did in 2008 in Florida, and just slightly below that in Ohio.

By the way, if you’re wondering how the national polls compare to last cycle’s national partisan split:

2008 exit poll: 39 percent. CBS News/New York Times: 36 percent. Pew: 35 percent. Washington Post/ABC News: 33 percent. NPR: 31 percent, 41 percent with leaners. IBD/TIPP: 38 percent. Battleground: 35 percent, 43 percent with leaners. National Journal: 36 percent.

So none of the national polls are presuming 2008-level turnout for Democrats (at least without throwing in Democrat-leaning independents), while a large chunk of the state-level ones are. In this light, it’s not all that surprising that Romney’s showing a lead in most national polls while trailing in a bunch of the key swing states.

[lasereye]

Seems like the total percentage identifying themselves as Democrats compared to 2008 is what should be the focus rather than the difference between D's and R's. Many of the state polls are implausible by that criterion as well but maybe not all.