How 'micro-data' are being used to influence Canadian voters, and how that's changing our democracy

Political parties have been collecting and discerning data from voters long before the age of the Internet and Big Data.

But when sensitive data leaked from social media can influence a presidential election, concerns over digital privacy come roaring to the forefront.

How is our data being used? What do Canadians need to know about "micro-data"?

Jennifer Robson is an Associate Professor of Political Management at Carleton University. Robson has observed the evolution of the political uses of data over the past few decades.

Professor Robson spoke with Michael Enright about this "micro-data" and how Big Data can change democracy. Below is an excerpt of the interview.

Michael Enright: Since you worked in politics in the days of the Jean Chretien government how much have you seen the role of data mining or voter profiling expand in our politics?

Jennifer Robson: I would say a lot and also it's important to recognize that parties have been doing this kind of thing for a long time. What's different now, though, are the tools and the scope and the techniques that they can use. So back in the dark ages when I worked in politics, there wasn't social media. There was no Facebook that you were able to mine. And so parties had to rely on old-fashioned public opinion research polling.

What's different now though is that you don't even necessarily have to ask somebody to to answer a questionnaire. You can introduce an email message a political ad and then you can watch how they respond to that because we've got this online platform now. Because we can do this online this also means that parties can now integrate data from a lot of different sources. So if I know your email and I can find you on Facebook ... Facebook has the ability to track all kinds of things about your online behaviour. That's rich, rich, rich data and definitely a lot more complex to deal with but probably also a lot more useful to parties once they figure out how to manage all of this in terms of actually crafting messages that aren't just segmented by you know maybe your gender, your age or what part of the country you live in, but who you are individually.

ME: And this is information that couldn't be obtained in the course of a campaign by going door to door?

JR: That's a great point. One of the adages of political campaigning, of course, is what's the most important strategic thing you can do? Go knock doors. Want to learn how to win the campaign? Knock doors. And knocking on doors — a lot of what it comes down to is effectively you're trying to do a census of your riding. You're going to try and talk to people and get get a read. And people who have a lot of experience in campaigning and especially in door knocking probably have an intuitive sense of whether or not somebody is likely to be a supporter.

In the last number of years parties have tried to get a little bit more sophisticated about not just having a gut sense of which doors to knock on or what to say, but to actually score people in these big databases that they are assembling in terms of the likelihood that somebody might be a supporter or that you might be able to convert them.

ME: And that's the phrase that I came across: voter scoring. Is it literally like a scale of one to 10 or something?

JR: It literally is. So different parties have what they call voter contact databases. The Liberals have a tool called Liberalist. The Conservatives tried to get something new but they're still using CIMS, as they call it and the NDP have Populus. I don't know what the Green Party has, but I'm sure they have something as well. And each of these databases will have data in addition to some very generic information from the voter rolls. So your name, your address. They'll also have hopefully any contact information that they can get for you like your email. That's super important for parties, and they'll record your engagement with the party — did you ever give money, did you ever sign a petition, did you ever show up to an event, have you ever volunteered, have you ever held a membership — and they use that information, plus external information, to generate a score, and different parties have different scales.

If I remember correctly, the Conservatives are one to 10 and I think the Liberals are one to five, and that indicates their best estimate of how strong a supporter or how likely or unlikely you are to vote for them.

ME: What's the risk that political parties, consultants and even government will wind up overgeneralizing or mischaracterizing the voters, or simply misinterpreting the information they have?

JR: I think it's a bit of both. So if you as a government are trying to get a pulse on what is the mood of the public and an increasingly important input for you for that information is what's happening within the voting segment that supports you, you can actually see that information in real time — how many of your supporters or likely supporters are responding positively to various messages that you're putting out to them. That can confuse decision-making in the sense that it obscures a whole segment of the population that isn't part of that universe. I think it also leaves out and further marginalizes people who are already fairly vulnerable and fairly excluded from the political process. So if you're not in one of the party databases, or if you're not online if you don't have email, if you're not on Facebook, you're kind of left out of this entire ecosystem. And there's also evidence to suggest that the way that Facebook's advertising algorithms work — and parties all rely on them and governments, as well — it serves up really different content to different kinds of people in part based on your financial value. So there's those kinds of risks and last but not least, I would say that there are potential harms in terms of the effects that we have on people when governments and parties send out highly targeted, highly personalized messages.

ME: I think this is key: that the data governments and political parties are getting are going to become so specific. And this information is so easy to get that any ethical considerations are going to fall by the wayside.

JR: I said to another journalist that I do an exercise with my students where I try and scare them a little bit and make them go and download all of the data that Facebook has on them and then asked him the question: "Is it is it still okay for political parties to make use of this or does this change your perception?" And my soul always dies a little bit when I hear one of the students say, "You know this does seem a little kind of sketchy, but it's just too important for electoral success. We can't possibly walk away from it."

I would say on the ethics of it, absolutely there's a privacy concern. There's also this real question of what are the unobserved impacts that the parties and their messaging are having on individual citizens and voters if you send out all these highly targeted messages that, you know, even people in the same house are getting different ads and different emails from the same party. What does that do to our ability to actually have conversations with each other as citizens? If certain forms of political ads are so tailored to people's underlying biases or fears and anxieties, what does that do in terms of our ability to trust anybody in political office? What does that do ability to say, "Hey, you know, that person over there is from a different political party, but that's a good idea that they are raising, and I believe that they are actually acting in the public interest?"

And you know, the worst of all: Does this make us wary of people who are not quite like us? So I do worry about those kinds of harms and measures, because if all parties are doing is looking at how many dollars to be raised and how many people click through the link on our email, they're not tracking those other impacts.

Click the Listen link above to hear the full interview.