Three Feet of Partly Cloudy

Sharing Options

The real losers last night — not that anyone is likely to take real notice — were the pollsters. This is not the case because we had a wave election, because a number of people were predicting that. The surprises all came in the margins.

Pollsters
“What I drink and what I tell the pollsters I drink are two different things.”

Races that were not supposed to be close, like Warner and Gillespie in Virginia, were close. Races that were supposed to be close, like Kentucky, weren’t close at all. This kind of thing happened in state after state, in race after race. Now when this happens from time to time, as it does, the one-off surprise is chalked up to the electorate having wild mood swings. The twenty point spread between the poll results the week before the election and the actual election results is attributed to a whole bunch of people making up or changing their minds. But it is actually the result of the very nature of polling itself.

What happens in a poll is that 2,000 people are asked their views, the necessary “scientific” adjustments are made, and this is then assumed to be the mindset of three million people. The reasoning process is called induction, where you are going from the particular to the general. Whether or not that reasoning is strong or weak depends entirely on whether your sample size is representative. Of course, you don’t know whether it is representative or not until after the election, at which point you should calibrate your methods. But we have now gotten to the point where the poll results are treated as mini-elections, with settled results, and the elections are treated as big elections, also with settled results.

What ought to happen is that our pollsters should be on television this morning, acting like a local teevee weatherman who has had to deal with multiple irate callers who had to shovel three feet of partly cloudy off their driveways. “Folks, this is not an exact science . . .” At its best, polling is educated guessing. At its worst, it is wish fulfillment therapy. At its best, polling is having thousands of conversations with people in the run-up to an election. At its worst, it is little better than telling the king which way he should go because your guild is the best haruspicy firm in the business.

Only God knows the end from the beginning. Mortal men want to know the future and they cannot. Mortal men want to know the future so badly that they are willing to pay big money for any plausible account. And much of the time, it can all seem pretty plausible — because the voting public is following the polls also and many times a reinforcement theme is created. Polls can and do create real momentum, and really can affect the outcome. But of course, if a bald eagle happened to land on a general’s helmet right before the battle, that could affect the outcome as well.

In all this, we should remember Isaiah’s taunt. “Shew the things that are to come hereafter, That we may know that ye are gods. Yea, do good, or do evil, That we may be dismayed, and behold it together” (Is. 41:23). Whatever else we may say about the political results of this election — and I would want to say that the unraveling of Obama’s apotheosis is almost complete — we can also take comfort in the fact that many of our nation’s soothsayers had their pointy cone hat knocked off, the one with the stars and crescent moons on it.

Subscribe
Notify of
guest
15 Comments
Oldest
Newest
Inline Feedbacks
View all comments
Matt
Matt
9 years ago

What ought to happen is that our pollsters should be on television this morning, acting like a local teevee weatherman who has had to deal with multiple irate callers who had to shovel three feet of partly cloudy off their driveways.

You mean like this?

Stephen Donovan
Stephen Donovan
9 years ago

Think you can swing a cat in the senate and hit a few republicans now? :D

Mark Hanson
Mark Hanson
9 years ago

I am fairly well convinced that poll numbers will tend to skew to call elections closer than they actually are. After all, the media loves a horserace.

Joe_WA
Joe_WA
9 years ago

At its best, polling is educated guessing.

Hmm… would the statistics professor at NSA agree with that statement?

Rob Steele
Rob Steele
9 years ago

It’s easy to predict the future. It’s hard to be right.

Joe_WA
Joe_WA
9 years ago

538: Polls skewered toward the Democrats Interestingly, this year’s polls were not especially inaccurate. Between gubernatorial and Senate races, the average poll missed the final result by an average of about 5 percentage points — well in line with the recent average. The problem is that almost all of the misses were in the same direction. That reduces the benefit of aggregating or averaging different polls together. It’s crucially important for psephologists to recognize that the error in polls is often correlated. It’s correlated both within states (literally every nonpartisan poll called the Maryland governor’s race wrong, for example) and… Read more »

Joe_WA
Joe_WA
9 years ago

Skewered=skewed

The kids call that an “eggcorn”

Shayne
Shayne
9 years ago

I’ve run a decent number of polling banks and have a degree in this. Everyone needs to remember that polling is about what’s happening in someone’s head “right now.” It’s entirely possible that people told pollsters an accurate view of what they thought a few days ago, and then didn’t act like that in the voting booth. Detailed exit polls can discover people switching sides and verify the pollster’s science. Polls used to be more accurate when everyone had a landline phone and were in a phone book. We used to be able to run down the the list and… Read more »

David Douglas
David Douglas
9 years ago

would the statistics professor at NSA agree with that statement? Joe…I sure hope so. Doug characterized his terms pretty well. If you have a sample that is random and is unbiased then the mathematics of the sample and the inferences regarding the population it was drawn from are pretty straight-forward. If the sample is biased or non-random, then your conclusions are not going to be correct. An example of the former is a roll of 2 dice (assuming the roll is not controlled or fixed). The total of the two dice averages 7 with a very predictable standard deviation and… Read more »

Jane Dunsworth
Jane Dunsworth
9 years ago

It’s not just that different kinds of people have cell phones, it’s that different kinds of people use cell phones vs. land lines differently. Most people my age and older, and maybe down to a decade younger, though nearly all of us have cell phones in this day and age, still have land lines, still only have our land line phone numbers widely published, and still default to using them whenever we’re inside our houses, unless calling long distance. At least until the land line disappears in another generation or so, sorting all this out has got to be very… Read more »

Robert
Robert
9 years ago

From what I see on TV, Obama is going to double down these next two years. Dictate by fiat.

Joe_Wa
Joe_Wa
9 years ago

Yeah , maybe I just have a lower opinion of the phrase “educated guess” (though I suppose that a comparison between the pollster and the “weatherman” is a fairly accurate one). If the polls are interpreted using the best methodologies then the predictions can be surprisingly accurate, as proven over and over again by Larry Sabato and Nate Silver (and if you really are just guessing or clamoring for your own side, you turn into Karl Rove and Dick Morris). Even undecided voters have predictive traits. It’s important to note that this particular race was particularly bad for the pollsters–not… Read more »

timothy
timothy
9 years ago

@Robert wrote:

From what I see on TV, Obama is going to double down these next two years. Dictate by fiat.

Pastor Wilson’s recent series on legitimate/illigitimate government is looking rather prescient.