Cold-blooded polling

The UK elections have left FiveThirtyEight a bit shaken up:

Perhaps it’s just been a run of bad luck. But there are lots of reasons to worry about the state of the polling industry. Voters are becoming harder to contact, especially on landline telephones. Online polls have become commonplace, but some eschew probability sampling, historically the bedrock of polling methodology. And in the U.S., some pollsters have been caught withholding results when they differ from other surveys, "herding" toward a false consensus about a race instead of behaving independently. There may be more difficult times ahead for the polling industry.

Maybe this isn't such a bad thing. It most definitely isn't for homegrown conservatives; they — backed by empirical evidence, which will be a shocking first for them — can now hoot and jeer whenever polls show Hillary with an invincible lead. For everyone else, though, maybe some lingering mystery as to 8 November 2016's outcome will reinvigorate center-left mobilization, ensuring that the mystery never was.

And there's this: a rather charming story in, as I recall, Richard Moe's Roosevelt's Second Act: The Election of 1940 and the Politics of War — an election which, of course, took place in pre-crackjack polling times. Aides had never before seen FDR so visibly out of sorts, until that election night. As disappointing early returns came in, the president turned grim and irascible. He isolated himself. He suspected a bad night and an even worse morning. The numbers soon reversed, and FDR wound up destroying Willkie, 449 to 82 in the Electoral College.

I wasn't around then, but I nonetheless miss such historical drama. Politics has become too clinical, too predictable, too statistical and computerized. The human drama of it is being — has been? — sucked out. Perhaps a touch of doubt can put it back in.

 

Comments

[RT]

Is FiveThirtyEight aware of the "Shy Tory" polling effect? It's notorious in the UK.

[Bob]

Though not around to see it personally despite the impression I might have left with some, I'm immediately reminded of the "Dewey Defeats Truman" headline in the cranky conservative Chicago Tribune. (I haven't read the thing in years and don't know if their temperament’s changed.) Mathematicians like to say that probability is lumpy. An easy illustration is that even though a flipped coin has a 50/50 chance of being heads or tails it's possible to get ten of either in a row. Lumpiness also applies to statistical political prognostication. Nate Silver and others were treated like oracles after a few hits, but that doesn't necessarily mean their methods work under all conditions; especially quickly changing conditions having to do with human emotions.

[Turgidson]

The Tribune is still the conservative paper, but they're not braindead regressive ideologues like the Journal and others. They endorsed Obama both times, I believe, even while many purportedly moderate or non-ideological papers were furiously huffing the glue the Romney campaign was doling out by the barrel.

[Peter G]

My understanding was that the polls that showed a clear Dewey victory in the offing were based on telephone surveys. Which rather overlooked that in those times only relatively well off people had telephones in their homes. Many of the great supporters of the Democratic party were among those who did not. But they were happy to vote nonetheless.

We may be seeing a similar phenomena today. How exactly do you find a statistically valid sample today when many people have only cell phones or even use only voip systems?

[Bob]

Nice that The Tribune has somewhat joined the real world, but I still probably won't bother with them out of habit alone, Turgidson.

Good point about changing technologies in addition to emotions, Peter.

[Bob]

This must have zapped around some brain cells for awhile before surfacing as a conscious memory, but the only times I read a Tribune were for Mike Royko's column. He swore he'd never write for the Trib but did some great work there. After all these years I can remember him describing the pols downstate as ditch jumping, clod throwing, 16 rpm hayseeds. The man had a way with words.

[shsavage]

Why on Earth would the GOP resort to statistically based polling when they have the expert opinions of political geniuses like Dick Morris and Karl Rove to tell them what they want to hear? That things don't work out the way they "know" they will simply provides more grist for the conspiracy mills and opportunities to fleece their addle-brained flock.