Q&A: Darrell Bricker, the Toronto guy who knows everything about election polling

Q&A: Darrell Bricker, the Toronto guy who knows everything about election polling

(Image: Claire Foster) (Image: Claire Foster)
 

Polls don’t just measure political campaigns; as wonky as they sometimes are, they can actually shape outcomes. The latest example? David Soknacki’s exit from the mayoral race following a disappointing showing in a survey by Forum Research. Even as the candidates base major decisions on polls, though, many of us don’t really understand how they’re conducted, or what the results mean. We sat down with Darrell Bricker—who is the Toronto-based CEO of the public-affairs branch of Ipsos, one of the top market-research firms in the world—to talk about the nature of polling, and how to tell the good polls from the bad.

Have you always been a numbers guy?
I don’t do sudoku or anything like that, but there is a precision to the way numbers answer questions that I find compelling. There’s less room for interpretation.

Seeing as how the municipal election is coming up, could you explain, in layman’s terms, how Ipsos might determine who is the mayoral frontrunner?
I think everyone does it in a somewhat similar fashion. The questions are pretty standard and straightforward; where the confusion comes is how people are contacted and how representative they are. We tend not to do telephone anymore, because the response rates are so low. Most of our work we do online. So you might get a pop-up question on a particular site if you fit a specific type of profile for that site.

It must be difficult to figure out exactly who is going to vote.
In the last provincial election campaign, 50 per cent of the people that were eligible to participate in our survey didn’t vote. Even though they may say they are going to vote, they don’t vote. In a mayoral campaign, it tends to be even worse, because the turnout tends to be even lower. So we might interview 100 per cent of the public and report that as though it’s going to be the actual voting population, but that’s two different groups.

Who’s going to win the mayoral election?
We know that in municipal elections, the turnout tends to be pretty low. The exception was Rob Ford. The last truly competitive election was John Tory versus David Miller, in 2003, and the turnout for that election was 40 per cent. Rob Ford got 10 per cent more of the public to show up and basically won all of those votes. That’s how he won. The presence of Rob Ford on the ballot really encourages people to show up. Olivia Chow is going to win downtown. Her voters are going to be motivated because Rob Ford is still on the ballot. John Tory has to find a way of engaging suburban voters. To me, that’s the dynamics of the campaign.

It must be an interesting time for your industry, because everyone’s willingly putting so much data online through social media.
It is really interesting. All of these sources of information that confirm similar things are worth paying attention to, but we haven’t figured out a way to calibrate that yet. Nobody’s figured out a way to successfully predict an election based on social media. At some point, that may be where we’re going. It’s a very exciting time to be in research, but it’s scary as hell, because you have to have bucketloads of skepticism and creativity.

Is it easier to poll in a country like Australia, where people are obligated to vote?
Well, they have a higher proportion of people spoiling their ballots as a protest, so you’re still struggling to represent the voting population. We did well in the U.S. presidential election, which is the Olympics of polling. The U.K. is tough, France is tough. We do this all over the world, and Canada is one of the toughest, because the methods of contacting people are in such transition.

You must see terrible polls get published all the time.
Every day.

What makes a bad poll, and what are some ways to spot them?
The first thing is when you see something that jars you, or something that seems counterintuitive. Also, if there’s no disclosure about what the pollster did, you should be automatically skeptical. These days, the media jump on whatever the new thing is, right or wrong.

When a poll is way off, what’s happening?
A lot of the ones that seem to be the spikiest are the Interactive Voice Recordings, or robocalls. They’ll get better over time, I think. The way these work is that you have this voice recording—“press one if this, press two if that”—and that gets blasted out to everyone. You phone as many numbers as you can. Out of that, one per cent might respond. The sample you get is going to be very old homeowners, disproportionately female, because they’re more likely to answer the phone. You’re getting a bunch of stay-at-home women. How representative are they of the voting population? If it’s an IVR poll, I generally don’t pay attention to it. But to be fair, the guys who did the best in the last Ontario election were the IVR guys. If you asked them why, they couldn’t explain it to you.

What happened with the provincial election? Everyone was shocked at the results.
Including the leaders of the parties. That’s explainable by the fact that what used to be a relatively easy thing to do is becoming increasingly complicated.

How so?
The people that we’re interviewing are becoming more complicated. For example, it used to be that everyone watched television, and not everyone watches television anymore. Before that it was radio, and before that it was newspapers. It also used to be that 99.9 per cent of households had a landline. In the U.S., over 40 per cent of households are cell phone only. Contacting someone by cell phone or by landline are very different things. And even the people that you do reach are less likely to participate. The old methodology where you used to just phone people up is pretty questionable for political research these days. We’re still working on perfecting the online method. Most times it works very well, but every now and then you learn something new. We certainly learned some things in the Ontario election.

Are you making any specific changes based on what happened?
I always learn something. Too often, when people get it right, it’s like they’ve discovered the elixir of life. But the truth is, you should be keeping your mind open, you should be learning and you should doubt your own work all the time. From election to election, we change something. Anybody who is in this business is walking a tightrope, and a dose of humility would go a long way.