Sep 27 2012
The political blogs are all atwitter about how the polls are slanting their samples and turnout models. Some claim nefarious intentions, others see a rising tide of Democrat voters that would eclipse 2008. Many assume doom for Romney. I doubt any of these situations are true. Something else is in play, and I touched on it in my previous post - see here. Since that last post much of the commentary has bolstered my observations and theory.
Before we dive into the data, let me reiterate the premise of my theory: The Libertarian/Tea Party insurgent voting bloc of 2010, that swept the GOP into office in mind boggling historic numbers, is still out there in force. But this bloc is resistant to being sampled in the polls, and this causes the polls to be skewed towards Obama.
Confidence in any statistical analysis (of which polling is just one instance) is driven by this simple fact – the sample must be highly representative of the system or population under study. If you miss a major element of the ‘thing’ being sampled, the statistical results will be off. The more you miss, the more the sample is off. If you are randomly missing individuals, no big deal. But if you miss a segment that is unified in its characteristic (e.g., is anti-Obama), then your result will be off.
After the GOP was swept into office at all levels of government in 2010, the voters who sent them there expected great things. We expected the nation to turn away from its current disastrous course and move to less government (and we mean MUCH less here), towards fiscal responsibility (or sanity),back to fairness, etc. What we got was the usual DC two-step (huge budget increases and lame promises to behave in the future).
The 2010 electorate never forgave the politicians for failing to exploit the opportunity given them. Thus both parties remain pariahs of the people, and DC is still the enemy of Main Street: where a vast majority of the voters live.
This disappointment has turned into hot, but patient, determination. There will be another round of incumbent culling in 2012 (as we have seen in primaries within the GOP, and we see with new comers easily taking on the old guard in two party contests). Scott Walker is still in the Governor’s office because the 2010 insurgent voter is still out there and active.
But these voters have no interest in playing the game anymore. They are not mindless or desperate groupies running to the GOP for salvation. They are quietly determined to keep sending messages to DC and the entire political industrial complex (note the still falling revenues, ratings, viewers and readers of the old media) – a strong message. And they are not answering the pollsters. They are a strong anti-Obama vote (note: not a pro-Romney vote), and they are resisting the pollsters efforts to sample them. They are not going to play any more. They will live their lives and then show up when it counts. Politics is tuned out right now.
We see tell tales of this in the data. For example, let’s begin here where one person’s “over sampling” actually supports my theory taht major groups are resisting to be sampled, thus making it look like the other groups are over sampled:
But the oversampling of Democrats is only a small part of a larger problem. In addition, pollsters are undersampling groups that are sympathetic to Republicans, like evangelicals and people who make over $100,000 a year, and oversampling groups that are sympathetic to Democrats, like mainline Protestants and people who make less than $50,000 a year.
If someone cares to do the analysis, they will find these under sampled groups probably represent the home of much of the Libertarian/Tea Party movement. The results of this missed sampling are mind boggling:
Here’s what Marist is predicting the 2012 electorate will look like in various swing states, relative to 2008:
Iowa -6.1% Iowa +5.9%
Colorado +3.2% Colorado +13.3%
Virginia -21.2% Virginia -20.5%
Florida +11.7% Florida +10.8%
I have highlighted my home state to illustrate how dumb this sample is. Virginia is deep purple, but started to turn on Obama within the first year, leading to the election of Governor Bob McDonnell (R) by a stunning +20% margin. In 2010 Virginians swept out a lot of Democrats in the House. So is it really going to be pro-Obama this year? The comparison between 2008 and 2012 indicates a serious sampling problem. One that should be noted by the pollsters and reducing the confidence (in other words, increasing the MoE).
The polls are statistics based on samples. Pollsters can only go on what the sample indicates. If a major segment of the population is not willing to be sampled, is resisting being sampled, and they universally hold a strong view one way or the other, then the results will come out wonky.
Here is some proof of this theory, when you focus on a specific voting block, it tells a different story than when you integrate the voting blocks to represent the country. Here is the proof:
For example, the most recent NYT/CBS/Quinnipiac poll of Virginia shows Romney carrying Republicans 95-3 and independents 54-43.
How can Romney be losing when he is taking independents and holding the Republicans? He can’t unless the Democrats are being oversampled, or the Tea Party component of independents and the GOP moderates are being under sampled! The fact is we see this sampling issue across the board (Romney holding his base and independents, but losing when the groups are integrated and weighted by sampling size). This universal feature in the data indicates to me this is not a pollster issue, but an issue with a voting bloc resisting to be sampled.
Here is more evidence that the polls are not getting a square sample, probable out of no fault of their own – the party ID race:
Of all the polls you have seen, there is one poll that has gotten no attention, it is a poll that has been taken monthly, it is a poll that Doug Ross spotted and promoted on his site. It is the Rasmussen Poll of party identification.
They have party identification results online dating back to 2004. Here are the 2012 numbers through August
This poll shows a huge surge in GOP levels. A historic surge. A surge that makes the 2010 results pale in comparison. Never has the GOP seen this large of a lead over the Democrats. Is this an outlier? I doubt it. Rasmussen has an excellent record. Why is the GOP registration surging, but not their intensity in the samples? Obvious voters are preparing to vote, but the polling is meaningless to them.
How is this comparing with current poll samples? Not even close!
The bottom line is there were seven percent more Democrats in the electorate in 2008 than there were Republicans. That’s from the exit polls and that’s about as accurate as you can get….President Obama won by about seven points. Given 90 percent of Democrats vote for the Democrat and 90 percent of Republicans vote for the Republican, every time you reduce the margin between the parties by one point, roughly it’s about one point off the margin.”
Schoen pointed out that the Pew poll was based on Democrats sampled for having an 11 percent voters registration edge over Republicans. He further added, “saying that America has gotten more Democratic than 2008, which is a questionable assumption.”
The sample in polls has become more Democrat leaning because those red hot angry Libertarian/Tea Party voters (who are strongly anti-big government and therefore anti-Democrat) are ducking the sampling. Jay Cost also sees an “over sampling”, which could in reality be an under sampling:
Specifically, many polls have, in my judgment, overestimated the Democrats’ standing right now. I base this conclusion not on a secret, black box statistical methodology or some crystal ball, but rather on a read of American electoral history going back to 1972. If I am right, then some of the polls are giving a false sense of the true state of the race, and will likely correct themselves at some point or another.
One important “tell” in my opinion, is this president’s continued weak position with independent voters, who remain the true swing vote.
Who is right? Is this biased and deliberate over sampling to shore up the lamest of ducks? Or is the data telling us the truth. Romney has the edge with the GOP voters and independents, but the under sampling of a stealthy, large voting bloc makes it appear as if Obama has the edge? We will know soon enough. We will definitely know by November. If it is a blow out (as I suspect it will be) then DC will again be put on notice – especially the GOP. 2012 will never be a pro-GOP vote because the government is broke all over. Main Street wants government to shrink back down to its minimum level, get out of our lives and stop taking our hard earned money. This means neither a GOP or Democrat activist government. We want both parties out of our lives.
The data to date indicates this is what is happening. Now we wait and see.If you want to see what the polls look like if the sample is corrected go here. If I am right, these are the numbers we will see in November.
Update: The above Hot Air article linked to this Breitbart article, that tells me something is truly wrong in the sample, not the polls:
In 2004 the vote was R+4.
In 2008 the vote was D+3
CBS/NYTs is reporting that in 2012 we will see D+9.
Is Florida now 3 times more Democrat than in 2008? Doubtful. It looks to me like the FL 2010 exit polls showed a 36-36-29 DRI – which means it was a tie. I fail to see the case for a D+9 in 2012 except if the 2010 insurgent voters are ducking the sample.