Poll positions

T.A. Barnhart

Poll positions

Polls can't predict this.

There are polls, and then there are polls. The media tosses them all out regardless of their substance or quality, so people assume all polls are created equal. Far from it. Here are a few things to know about polls.

People lie to pollsters. This is dumb, because the two people on opposite ends of the phone call do not know each other. They will never meet. The pollster has too many calls to make to care what any particular respondent is saying. The respondent should be able to speak with candor and honesty knowing that his or her answers will be anonymous data and no one will ever know how they answered.

Still, people lie to pollsters. It’s a human thing, to say what you think people want to hear or expect to hear. Or to not say things that you know might not be popular or socially acceptable. Such as, “Do you think black people are inferior to white people?” Ask someone that in a televised person-on-the-street format, and they’ll say “Of course not” because who wants to be on tv (and then YouTube) saying, “Yup, white people is superior”? So they lie to the camera.

And they lie to the pollster. They have no reason to do so, but they do. The pollster doesn’t care; they have a job to do, and this is today’s set of questions for today’s set of phone numbers. You’re a racist? I have work to do. But voicing unpopular sentiments is difficult to do, so even in this setting, people lie. That’s why you’ll see polls that show support for a ballot measure that would benefit some minority population and then, on election day, the measure fails. People had been lying to pollsters all along.

Good polls are expensive and difficult. Respondents lying to pollsters is just one of my problems facing a good pollster. A dependable poll can be done with a bit over 400 respondents; the trick is finding the right 400 respondents. First, to get 400 people to answer the poll, probably ten times are many phone calls need to be made to get past busy signals, no answers, disconnected numbers, hang-ups, etc. Then the pollster has to spend 20 minutes or so asking questions. This means a team of people on phones, in front of computers, dialing and talking, over and over.

That don’t come cheap.

On top of which, a single poll, however well it’s done, is just that: a single poll. Scientists never rely on a single research study to draw conclusions about a hypothesis; multiple studies need to be done, and it’s from the many different studies that reliable conclusions can be drawn. Political polls are the same: the more polls on a particular campaign or issue, the better our knowledge. Nate Silver and his peers have developed means to aggregate polls in order to get reliable information.

Multiple polls repeated multiple times in order to get good data that leads to strong inferences; again, that don’t come cheap.

(How the questions are phrased, or even which questions are asked, are critical to what data is gleaned. But that’s a topic for not merely another blog post but for hundreds of books. Feel free to read up on the problem if you want.)

People love to be part of a bandwagon. Bernie Sanders supporters have been crowing for some time about how he’s surging in the polls. And yes, part of that is that it took a long time for people to learn anything about him, think about his campaign promises, and then to compare him and his promise to Hillary Clinton. The massive lead she once held was never going to hold up once he started to become familiar to voters.

But then his surge zoomed, and for supporters there was a single reason: voters are learning that he’s the right person for the presidency and people are thrilled to desert Clinton. This is true of many of those telling pollsters they now support him, but so is this: People want to back a winner. They want to be part of the new, exiting, and, they think, winning team. Who wants to back a loser?

The trouble with this viewpoint is that it’s more than transitory; it reflects an emotional response to the campaign that people are feeling today. Will they feel that way in a week or a month? That depends. Clinton loses big in New Hampshire, so Sanders surges nationally. On Saturday, after she wins big in South Carolina, what do they think then? Or a few days later, if one or the other dominates Super Tuesday, which bandwagon do people hop on?

This leads to another truth about polling:

Horserace polls are stupid. It’s one thing to poll voters about two nominees in the weeks prior to an election. By that time, voters know most of what they need to know and they’re getting ready to vote. The best polling firms know how to ask questions that will give them solid information. But more than half-a-year before the election, and with multiple candidates still in the running, and very few primaries or caucuses having been conducted? And the candidates have only visited a few states?

It doesn’t take much to change the horserace. A gaffe, a revelation, a bad debate performance, an unexpected primary result. Hell, maybe a really awesome commercial – Yes, we can – or even an amazing speech – again, Yes, we can. What was amazing about Obama’s “Yes, we can” speech was that it came after a pretty solid defeat. But he took that moment and used it to change his campaign. The chants at the beginning of the speech were “O-ba-ma”; before he was done, his campaign was defined by “Yes, we can”. His defeat ended up sounding like a victory rally.

A single moment can change the horserace. We don’t know which candidate will do something awful or wonderful in the next few hours or days or when. We don’t know what “outside” events might bring new considerations to the minds of voters. So the horserace today is exactly that: the horserace today. For supporters, to be ahead now is exciting and better than not being in first. But what does it mean in the long run?

For many candidates who’ve led the horserace early on, it means that when they bow out of the campaign or end up losing, part of the story will be about their fall from the summit. Just ask Howard Dean how much being the front runner in December 2003 did him when his campaign failed to do the job in Iowa. Ben Carson led the huge Republican pack at one point last year, and a fat lot of good that did him.

Polls are not votes. In the end, polls can inform about what’s going on now, but they do not determine any outcome. A properly done poll can provide a lot of useful data for campaigns to use, but that’s about strategy, not outcomes. Doing well in the polls can lead to increases in fundraising, but Jeb has shown how little can be accomplished with an ungodly amount of money. Campaigns conduct their own polls because knowing what’s going on now is vital, not so the campaign can tell supporters “We’re winning!” (which they’ll do if their poll comes out well for their candidate) but so they know how to tweak the ads and the field and the speeches.

Votes count, not polls. No one has ever taken an oath of office in this country because they were first in a poll. For Sanders supporters, using current polling to urge others to support their guy is legitimate politics, but only within the scope of “We’ve got a great chance to win”. Taking a horserace poll that shows Sanders doing better against Trump than Clinton is nonsensical. None of those three is a nominee at this point, and the number of possibilities for what could change between today’s poll and the reality of, say, September 1st, is huge. The polls are snapshots, not inevitabilities.

Take a deep breath. Every candidate has good and bad polls. Along with the polls showing Sanders doing great are undoubtedly others that are less rosy; perhaps, among the many questions a good poll will have, there will be hints of trouble ahead. Candidates may make noise about the polls – in Trump’s case, half of his campaign is him making noise about the polls – but the good candidates glance at the numbers, ponder for a bit, and then move forward. They and their staff will use the numbers to adjust things, but whatever the polls say, the candidate has a job to do.

And so do voters. Listen and learn is the #1 job of any responsible voter; “go crazy about a poll” is a job no one should take on. Polls won’t win an election; grassroots campaigns win an election. Candidates speaking to as many people as possible. Volunteers going door-to-door. Effective media. Get out the vote efforts. These are what win campaigns. Polls are a tool, but they are not to be trusted.

I like polls. They can be fun, and they can be maddening. The more I’ve been involved in politics, the less I’ve let myself be attached to them. I can’t do a lot about a presidential campaign, so the solid info I know I’ll find at FiveThirtyEight and other top-notch aggregators of polls is interesting but little more. I am trying to help elect a mayor in Portland, and any polling released publicly isn’t going to mean squat to me. The campaign’s field operations concern me more than anything: identifying our voters and then ensuring they turn in their ballot.

Here’s the last thing to remember about polls: They are meant to make money. Why do so many media operations conduct polls? Because they want to be where people come to get informed about an election, and the more people who turn to a particular network or website, the more ads that organization can sell and the more money they’ll make. NBC and Fox and the Wall Street Journal and all the rest aren’t polling to serve the public good. They are trying to be dominant news sources, because that means money.

If you really care about your candidate, let go of the polls. Donate what you can, and talk about the issues. Don’t tell people “your candidate sucks because mine does better in this poll”. That’s not a winning argument; that’s a fight. Your candidate’s great poll today might be a crappy one in two weeks. Will you quit because of the bad poll?

Your passion is what matters. Your beliefs. Your desire to elect your candidate. You know you won’t be swayed by the polls, so why use them as a tool against others? Polls are info, and nothing more. And, taken individually, they often are not very good info.

Unless that’s all you’ve got.

Comments

connect with blueoregon