RWCG


The (hoped-for) Silent Non-Pollable Majority
November 4, 2012, 10:28 pm
Filed under: Uncategorized

Let’s forget Nate Silver and statistical modeling, and just look at poll numbers. These are all the polls I see of Ohio in the past two weeks (via electoral-vote.com):

No poll in evidence whatsoever shows Romney in the lead. Only two even show him tied (Rasmussen the alleged conservative-biased pollster, and one other from two weeks ago). To one digit anyway.

Now everyone says that if Romney wins, it’s likely to be via winning Ohio. But to think Romney will win Ohio, or is even all that close (the average gap above is ~3points), you have to think virtually ALL of these pollsters have (intentionally or inadvertently) a (D) bias. It is just astronomically-unlikely to get a set of results like above from 18 different pollsters (using whatever different methods) from a genuinely Romney-leaning state without there being something systematically wrong with virtually all of the pollsters.

(Another explanation of course is that Ohio isn’t actually a Romney-leaning state, it’s a state that’s gonna vote for Obama by ~3pts on Tuesday.)

In other words, if Romney wins on Tuesday, via Ohio, we basically have to blow up the industry of polling entirely and go back to the drawing-board. Now, is that possible? Sure. It’s possible. But just so you know, if you’re hoping Romney wins a state like Ohio, that’s the kind of thing you have to hope for.

In the process, the media would have to do a lot of soul-searching and try to explain the curious phenomenon of the Silent, Non-Pollable Majority that lots of people, myself included, are hoping are out there.

About these ads

6 Comments so far
Leave a comment

What are the internals like? (RCP has more polls for the last 2 weeks, incidentally).

Comment by A Lady

Internals – no idea. Would internals help explain why they are all biased in the same direction all the time?

I grabbed this from the downloadable data section of electoral-vote.com. I guess it is not complete; I could add in the RCP polls and see how it looks…would it look better? I’ll check…

Comment by Sonic Charmer

yes, people have been going through the internals (where they can be found) and there are clear trends of reshuffling the internals in many polls when the results start moving away from Obama. as some have noted, you can reweight by group, which will change your D/R/I without explicitly changing your D/R/I. gallup was apparently sued by the obama campaign for having results too favorable to romney pre-debates.

i voted today. i will not vote tomorrow though(i was sent many ballots instead of, you know, one per registered voter in the household).

Comment by A Lady

Oh noes! The polls might be wrong!

You appear to be starting from a large amount of faith in the polling industry’s methods and accuracy. I have very little, so it won’t surprise me if they fall on their faces. It’s obvious that the “silent non-pollable majority” is the 91% of the population that pollsters admit they don’t talk to, and it’s hopefully obvious that their methods encompass far larger sources of error than the statistical +/- 4% in the press release.

At this point it appears probable, even likely, that all 18 got it wrong in OH.

Comment by SkepticalCynical

I certainly don’t have anything like ‘faith’. I assume the polls are error-prone.

But all polls systematically erring in the same direction, virtually all the time, would be remarkable. Wouldn’t it?

If the real sentiment in Ohio is Romney 48, Obama 46 – then by just random error we should be seeing some Romney 50 – Obama 44 results. That is simply not happening,

The 9%-penetration story might be the answer, except then you still need a good reason why that other 91% would necessarily lean (R).

So either way, there are only two explanations,

1. Romney is not really ahead in Ohio, he’s behind, just as it appears.
2. ALL POLLS BY EVERYONE are systematically biased in favor of (D)s.

I am prepared to and in some sense would love to believe #2! Election aside, it would just be a fascinating forensics story, like CDOs/rating agencies! But right now, it’s hard to believe it for me. Is it easy for you? In fact the only viable explanation I’ve seen for it is the Bradley Effect.

Maybe indeed there is such a Bradley Effect. I hope so!

Comment by Sonic Charmer

From the pollster’s standpoint, the challenge is the other way around – prove that the 9% mirrors the 91% so that your sample is in fact random. That’s probably not impossible, but it sounds both time-consuming and expensive. I seriously doubt that any state pollsters have the resources to undertake it.

I don’t have any trouble disbelieving the OH polls because (a) my priors say that these polls are pretty inaccurate, and (b) everything else I see points to a Romney win.
– Early voting has swung +R by more than Obama’s 2008 margin of victory
– Ds have very low voter enthusiasm – did you hear about Obama’s 20,000 person rally in Cleveland yesterday? Me neither!
– Rs have historically won the election day vote in OH, even in 2008.
– The economy sucks. Historically, most undecideds/independents wind up going for the challenger in this situation.

Comment by SkepticalCynical




Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s



Follow

Get every new post delivered to your Inbox.

Join 455 other followers

%d bloggers like this: