Wednesday, August 1, 2012

Latest Quinnipiac/New York Times/CBS Swing States Poll — Overstates Democrats in the Electorate

My first reaction was, "No, it can't be!" Then I remembered, "Oh yeah, this is the New York Times."

So there had to be some bogus sampling issues. Here's the report, "New Polls Show Obama Has Edge in 3 Battleground States." (Via Memeorandum.) I checked the internals, "The Latest Quinnipiac University/New York Times/CBS Poll." And boom! There's way too many Democrats for these numbers to be meaningful. Ed Morrissey has this:
Now let’s take a look at the partisan breakdown (D/R/I) in the sample data for each state, and compare them to 2008 and 2010 exit polling:
* Florida: CBS/NYT 36/27/32, 2008 37/34/29, 2010 36/36/29
* Ohio: CBS/NYT 35/27/32, 2008 39/31/30, 2010 36/37/28
* Pennsylvania: CBS/NYT 38/32/26, 2008 44/37/18, 2010 40/37/23...
In other words, these polls are entirely predictive if one believes that Democrats will outperform their turnout models from the 2008 election in Florida and Ohio. That would require a huge boost in Democratic enthusiasm and a sharp dropoff in Republican enthusiasm — which is exactly the opposite that Gallup found last week.
Here's Gallup from last week: "Democratic Voting Enthusiasm Down Sharply From 2004, 2008." And check Take My Liberty, "Enough with the bogus polling data."

It's not just voter enthusiasm, by the way. The New York Times consistently oversamples Democrats, for example, "New York Times: Oversamples Democrats, Push-Polls Respondents in Health Care Survey."

And check Pew's findings for national-level partisan identification: "Partisan Polarization Surges in Bush, Obama Years." Dems are about 32 percent of the electorate nationally, with independents comprising the plurality of party ID.