Friday, October 5, 2012

Debate Aftermath: Jobs Report Conspiracy Theories

The day after the day-after Democrats are still picking up the pieces and Republicans are dancing!  Nate Silver councils patience:
So have some patience, poll-watchers. There are some points in each election cycle where the amount of news exceeds the amount of polling evidence that we have to analyze it. Because a number of major economic reports areaccounted for directly by the model, we can sometimes get a head start on those — but that isn’t true for noneconomic events like the debates.  
The big news is the jobs report which despite being fairly anemic in its job numbers dipped the all-important Unemployment % below 8.0 (from 8.1 to 7.8). The Washington Post makes a case that this is even better news than the numbers suggest (the last two really bad numbers were revised up, etc.). And Obama's InTrade bounce was bigger than his debate slump on the jobs news.

So we can see why this isn't good news for Republican strategists who just gain control of the narrative and suddenly may have to deal with everyone talking about a (slow, painful) recovery. That's fair and it's worth noting that the bad jobs report that landed the day after Obama gave his DNC speech was considered to be potentially blunting of his surge (did it? No proof, of course--but it sure wasn't adding extra lift).

What Does It Mean?
What's interesting is the immediate reaction: The numbers were fixed!
Minutes after the report was released, Jack Welch, who famously cooked General Electric’s accounting books when he was CEO, accused  President Obama of manipulating the numbers to distract from his debate performance:
He Would Know ...
Conn Carroll, a Washington Examiner writer doesn't think it's the BLS that did the cooking--it's "a bunch of Dems lied about getting jobs."

There are some posts debunking the claims--but I'm more interested in the role that conspiracy theory is playing in the election. The Monkey Cage emerges with a study of conspiracy theory that suggests that it's based on a power-imbalance: when Republicans control the presidency or government the conspirators are the Right or Big Business. When it's the Democrats, it's the Left or Communists. In other words, as their report is titled, "Conspiracy Is For Losers."

This isn't exactly news--no one wants to see a bad poll for their guy and Democrats did this too last cycle against Bush. Having a conspiracy theory is a way to mitigate the (emotional) damage. However, I think there's something else that's also at work. Let's look at the polling conspiracy theory.

The Polling Conspiracy
The polling-conspiracy theory suggests this: when you look at the composition of a suspect poll you find that the Dem/Rep/Ind split looks something like 42% / 30% / 28%. If your "real" (Rasmussen-based) numbers suggest something more like 32% / 35% / 33% (or whatever) you "realize" the poll is showing DEM +12 (12 percentage points) over Republicans when the real numbers should be Dem -3 (according to the population) or at most Dem +7 from the highly-energized 2008 election exit polls. So clearly this poll is bunk.

The possibilities are:
  1. The pollster made up their numbers
  2. The pollster uses a "random sampling" method that over-selects democrats such as polling outside an instant-check-cashing store (that was an actual "joke" from the comments)
  3. Only Democrats are picking up the phone to answer (the least conspiratorial)
This could be corrected if the pollster just re-weighted their numbers for electorate composition the way they do for minorities or the youth vote. That is to say: if your poll of 100 people only found 3 black households and you know the area polled has a 12% black population you have to do something to weigh those three respondents more heavily. If you weigh by projected electorate split, those Democrat respondents would each count less (or, put the other way, the fewer Republican respondents would count more).

There's a reason very few polling agencies do this: the reason is that the question is not are you a registered Republican but rather something along the lines of "whom do you identify with." That, unlike age, gender, or race--is subject to change. As it is subject to change that means that Dem+12 is correlated (at least theoretically) with Obama winning. 

If you do adjust / weigh for "presumed" electoral-composition you wind up "flattening" the polls--because of the way reality works (i.e. the further ahead a candidate is, the more +X shows up in their category) you would actually lose vital resolution.

So, okay: that's Polling 102, I guess. So let's go to the insight ...

What Do I Think?
I think that the reason why the polling conspiracies have caught fire is because the above explanation seems entirely nonsensical to the on-line commenter demographic because they cannot imagine a circumstance where they would change their vote to the D (or R) column. In other words, it makes more sense to believe that the polling conspiracy is, well, a conspiracy--than the "real" explanation.

It's an easy sell.

In the case of the jobs numbers, again, there is a selling point here--the timing could not be worse--Romney gets momentum and then BANG there's a jobs report that--however bad it actually is (there are a lot of ways to interpret the data)--the optics on the top-line number (less than 8%) are good for Obama. It opens the window for Obama to change the conversation which is breathing room you don't want your opponent to have.

So I think that there is an opening--a key-hole, so to speak--that the "key" of a conspiracy theory fits into. The problem is this: the people promoting this conspiracy are doing it on Fox News and from positions of, reasonable prominence. 

I think that the more conspiracy theories you believe the more apt you are to believe others--conspiratorial thinking is a modality--and the Republicans are doing themselves no favors by serving these up for ingestion by their electorate ... although it does help ratings.

NOTE: We can contrast this to the incredibly stupid "Romney had a cheat-sheet" conspiracy theory that some liberals are propagating. It's the same kind of "there must be some mysterious agency at work" thinking (HOW ELSE COULD ROMNEY BEAT HIM!) but it is not being, for example, promoted by MSNBC or the like. Outlets like the Huffington Post conclude the object is, indeed, a handkerchief. Clearly political partisans--not just Republicans--are susceptible to conspiratorial thinking--but the reach of the jobs-numbers theory is pretty daunting.

No comments:

Post a Comment