Sunday, January 24, 2016

Polling

I've been thinking for a while that the whole polling culture that has come to dominate our elections is seriously screwed up.  The media is gaa-gaa over them and touts this poll or that poll as if they are the gospel.  But there are so many things that can impact polling, so many different variables, so much hype, so many differing questions and populations, and on and on and on.  I feel really stupid when I hear of some poll and think that it matters.  Because it doesn't.

Frank Bruni has written a really good piece in today's NYT called "Our Insane Addition to Polls".  You can read it here.  He makes a ton of good points!

First is that they have crazy variability.  Someone who is up today is down tomorrow.  What is a voter to think.  And there are way too many of them.
"Over a monthlong period ending Thursday night — a monthlong period, mind you, that included the Christmas and New Year’s break — there were 11 polls in Iowa, 10 in New Hampshire and nine nationally. There were polls focused on 10 different states. ... I’d say that we’re in a period of polling bloat, but bloat is too wan a word. Where polling and the media’s attention to it are concerned, we’re gorging ourselves into a state of morbid obesity."  
And they can be self-fulfilling.  If the media weren't so fixated on how Trump is doing in polls, would he be getting the coverage?
"If Donald Trump wins the Republican nomination — or, heaven forbid, the White House — it will be partly because we in the media justified saturation coverage of him by pointing to polls, which in turn legitimized his fixation on them as proof that he’s up to the job: He must be, because plenty of people apparently picture him in it."
Polling that shows candidates close isn't as good a story as someone getting trounced.  We've come to the place in society where we don't just like winners...we like big winners.  So I suspect pollsters sometimes arrange the questions or the audience so as to show just how badly someone is doing.  Because the the worse someone is doing, the better the story.
"Desperation makes a better story. So the media dwells on the most pessimistic projections, ensuring that polling, no matter how divorced from reality, shapes it. ... Polls determined which Republican candidates participated in which debates, although just a couple of percentage points — the margin of error, really — separated a few of the prime-time debaters from the early birds." 
Then there is the local poll vs the national poll craziness.  This far in front of the general election, national polls make no sense.  And the truth is that the vast majority of folks haven't made up their mind and are potentially open to change.
“A national poll is absolutely meaningless,” said Stuart Stevens, the chief strategist for Romney’s 2012 campaign. “One of every nine Americans lives in California. So one of every nine voters in that poll is going to be in California. When’s the last time anybody read a story about the Republican primary in California?” ... And if you dig below the surface of these national polls — or of polls in the states with the first contests — you find a crucial detail that we in the media blithely gloss over: Many, if not most, voters haven’t made up their minds. In last week’s CNN/WMUR poll of New Hampshire voters, for example, about one in three Republicans said that they had definitely decided on a candidate." 
The thing that is never talked about when the anchor or reporter breathlessly reports that so-and-so is getting trounced (heh, heh, heh) is who is being polled and what the questions are.  That seems like it might sorta matter.
“The entire industry rests on the idea that the people you get are representative of the people you don’t get,” said Jon Cohen, who supervised polling for The Washington Post from 2006 to 2013 and is now the vice president of survey research at SurveyMonkey. “I think that’s an increasingly questionable premise and one that I keep in mind every time I design a survey.” 
So if you're interested in this stuff, read the whole thing.  Polls are one of those things we all take for granted.  We hear what's being said but how often to we evaluate the truth behind it?  Maybe we should!

No comments: