For the Stupid, By the Stupid: Conservative Poll Deniars Can't Losing Their Minds

You may have heard by now that some conservatives have found an elegant solution to Mitt Romney's nose-dive in the polls: stick their fingers in their ears and yell, "La la la la I can't hear you!" Actually, I want to thank the people at the ironically named UnskewedPolls.com; since it will give me the opportunity to have some fun with them. Well, here's the first piece of said fun. I was curious about the numbers, and of course, the polling methodology they are using to "unskew" the data that the "liberal media" is putting out in a conspiracy to depress Republican turnout.

As I read through their website, hilarity ensued almost instantly. Of course, their methodology is far skewed to the right - in both partisan and ideological breakdowns. This stuff jives with neither the latest registration data (D +12), nor the 2008 exit polling (where Democrats enjoyed a 7-point advantage and conservative made up 34% of the electorate). But that is to be expected. The hilarity, though, comes from something different yet. The conspiracy theorists can't seem to decide just which skewed numbers to use! (yes, I saved screenshots).

They describe one partisan breakdown in one page, and another one in a different page. I have found no less than three separate ideological breakdowns their numbers either directly state directly calculate. Let's have a look.

The site, UnskewedPolls.com is run by something called the QStarNews, whose Quick Poll page describes their methodology:
The QStarNews poll works with the premise that the partisan makeup of the electorate 37.6 percent Republicans, 33.3 percent Democrats and 29.1 percent independent voters. Additionally, our model is based on the electoral including approximately 41.6 percent self-described conservatives, 32.6 percent self-described moderates and 25.8 percent self-described liberals.

Republicans are 68 percent conservative, 27 percent moderate and 5 percent liberal. Among Democrats, 14 percent are conservative, 36 percent are moderate and 50 percent are liberal. Independents include 39 percent conservatives, 36 percent moderates and 25 percent liberals.
Wow, you say. Are these people nuts or what?  But wait. We'll have plenty of opportunity to ridicule their methodology. Right after they decide what their weighing methodology actually is! Here's the same site describing their weighing methodology, in another one of their pages, which is supposedly the official page to describe their methodology:
The QStarNews poll works with the premise that the partisan makeup of the electorate 34.8 percent Republicans, 35.2 percent Democrats and 30.0 percent independent voters. Additionally, our model is based on the electorate including approximately 41.0 percent conservatives, 20.0 percent moderates and 39.0 percent liberals.

Republicans are 89 percent conservative, 9 percent moderate and 2 percent liberal. Among Democrats, 3 percent are conservative, 23 percent are moderate and 74 percent are liberal. Independents include 33 percent conservatives, 49 percent moderates and 18 percent liberals.
Ahem. So according to the GOP bubble-world, either 37.6 or 34.8 percent of electorate are Republicans, of whom either 68 or 89 percent are conservatives, either 27 or 9 percent are moderates and either 5 or 2 percent are liberals. Likewise, of the either 33.3% or 35.2% who are Democrats, either 14 or 3 percent are conservatives, either 36 or 23 percent are moderate, and either 50 or 74 percent are liberal. Of the remaining either 29.1% or 30% who are independents, either 39% or 33% are conservative, either 36% or 49% (!) are moderates and either 25% or 18% are liberals.

Making your head spin yet? Take a breather, there's more. Also according to their ONE AND ONLY methodologies the Right wing bubble sees as the correct model, here's how the ideological breakdown of the electorate breaks down:
  • Conservatives: either 41.6% or 41.0% (this is probably the smallest discrepancy).
  • Liberals: either 25.8% or... you know... 39% (!!!).
  • Moderates: either 32.6% or 20% (you know, same difference).
They even go on to helpfully describe how their samples are ALWAYS weighted. Either this way:
Our polls about doubly-weighted, to doubly insure the results are most accurate and not skewed, by both party identification and self-identified ideology. For instance, no matter how many Republicans answer our survey, they are weighted at 37.6 percent. If conservatives are over-represented among Republicans in the raw sample, they are still weighted at 68 percent of Republicans regardless.
Or this way:
Our polls about doubly-weighted, to doubly insure the results are most accurate and not skewed, by both party identification and self-identified ideology. For instance, no matter how many Republicans answer our survey, they are weighted at 34.8 percent. If conservatives are over-represented among Republicans in the raw sample, they are still weighted at 89 percent of Republicans regardless.
But wait, that is not all. What if you take their given numbers in the two pages about the ideological breakdowns within each of the partisan identities? The first set of ideological breakdowns seems to have been calculated to match the first set of partisan breakdowns (and the sub-breakdowns on ideology) but on the second mode, they didn't even bother to do that. In the second model, their overall ideological numbers say one thing, but the calculating the ideological subsections in the partisan breakdowns would yield the following overall ideological set:
  • Conservatives: 41.9%
  • Liberals: 32.1%
  • Moderates: 25.9%
This is a disaster. The site wanting to "unskew" the polls has generated, within a day of launch, on its own website, two sets of partisan breakdowns and three sets of ideological breakdowns in the American electorate. I would say these people need a class in statistics, but that's too advanced. They need a class in basic math.