We launched Vote Compass today with a company called Vox Pop Labs (no relation). It's a tool that puts your political views on the same spectrum as the presidential candidates'. It's a longer quiz than you would normally take on the internet, but it's the most efficient and rigorous way to map your views onto a spectrum that looks like this:
When I first got my results, I realized that I had never actually quantified where I fell on a political ideology scale. I just tried to place myself somewhere on a scale that looked like this:
But as you can see from the matrix above, Vote Compass — and many political scientists — insist on using quantifiable data and then mapping it onto a Cartesian plane. On this chart, one axis is social views and the other is economic views. It looks like this:
What is compelling about Vote Compass, at least to me, is that it takes this academic rigor and quantifies you and the candidates on these scales. And because these questions are based around policy issues, instead of squishy questions about likability or trustworthiness, it helps us frame poignant conversations with our friends and family.
For example, Vote Compass ask you whether the possession of drugs for personal use should result in jail time — and it also knows how each of the candidates would answer the question, based on their public statements. It does this with 30 policy positions, which gives an incredibly deep (albeit incomplete) insight into where you are on the spectrum. If you and your best friends end up on different parts of the matrix, the two of you can pinpoint where exactly the two of you disagree (maybe it's this drug possession question) and dig into it.
It's easy to conflate our political parties with our actual political beliefs — especially if we use the mental model I was using, since one side means Democrat/liberal and the other side means Republican/conservative.
But there's a chart published on Vox, by political scientist Jennifer Victor, that illustrates why it's important to quantifiably map our own political beliefs onto this two-axis chart. She shows that Republicans and Democrats used to be split along economic lines, but that's shifted in the past 50 years to be along social lines. So if you are in the top-right quadrant, your views were better represented by Republicans in 1960, but by Democrats in more recent elections.
We plan to analyze the survey results and write stories about how different views do or don’t line up with each other. We will not, however, see your individual response. Vote Compass will adjust the results so they accurately reflect a voting population, and will provide us with anonymized data. You should see those stories on Vox in the coming weeks.
To get a better idea of how Vote Compass works, I chatted with the man who created the tool, Clifton van der Linden, a PhD candidate in the University of Toronto's political department. His research interests are technology, electoral politics, "big data," and public opinion; that's why he created this tool for the 2011 Canadian federal election. It has since been used for Canadian provincial elections, Australian and New Zealand federal elections, and the 2012 US presidential election. Vox Pop Labs publishes its methodology, but van der Linden provided further insight in our conversation.
This conversation has been edited for length and clarity.
So I answer these 30 questions and provide demographic information. How does Vote Compass use that to put me on this two-dimensional social and economic map with the candidates?
Clifton van der Linden
The questions the user responds to in Vote Compass are all questions that reflect key points of difference between the policy platforms of the candidates. So our team of political scientists [at Vox Pop Labs], in collaboration with prominent American political scientists — in this case, at NYU's Social Media and Political Participation lab — calibrate the candidates based on the public disclosures they've made during the course of the campaign.
For every question a user encounters, the candidates have also been coded on those questions — what would their responses be if they were using Vote Compass? And what do their public disclosures suggest about how they're positioned on those issues?
Vote Compass takes the user's responses and the positions of the candidates, and it has several different algorithms that calculate the proximity between user and candidate on the basis of their respective responses. It allows you to see how you stack up against each of the candidates.
And all of this reflects a sort of reality in electoral politics, which is: People's vote choice is complex. Whom people decide to vote for can be a function of their party identification, which can be a longstanding party identification. It can be a result of their affinity toward a particular leader or candidate, so people vote based on the perceived character, integrity, and trustworthiness of a candidate. And it can be on public policy issues. So some people vote on the basis of who they think will implement the most effective policies for the country. Vote Compass tries to capture much of that calculus, but it certainly doesn't capture the entire calculus.
How do you ensure the Likert scale translates into the liberal-to-conservative scale accurately? [The Likert scale is the type of rating scale you often see on surveys, where the possible answers start with "strongly disagree" and scale up to "strongly agree."]
Clifton van der Linden
Earlier iterations of Vote Compass would rely on theoretical considerations. Is this a social issue or is this an economic issue? You assign it to that dimension, and which way does it scale? If strongly agree is the most liberal position, then that question scales strongly agree as liberal. And if strongly agree is most conservative, that scales as conservative.
But in recent years, we've actually removed, to a large extent, any theoretical or arbitrary considerations from the scaling of those items. So what we do now is commission a pilot study in advance of launching Vote Compass, and we go out to Americans and we actually put these questions directly to a representative sample of Americans. [Van der Linden later told me this helps them figure out questions Vote Compass uses to find differences between the candidates' positions.]
Their responses tell us which issues are salient for Americans, i.e., which questions should be included in Vote Compass. The original Vote Compass had a much longer set of questions — about 100 questions. So we're able to extract the most important questions — ones that not only differentiating the candidate, but also the positions of the American population.
You've said in the past that Vote Compass isn't a tool that gives voting advice. I know why you're saying that, largely because there are other factors at play, as we're reminded in this election. But I have to think it gives some voters pause, right?
Clifton van der Linden
The short answer is yes.
I think people are often intrigued by their results. Vote Compass results may, for many people, confirm where they're situation ideologically in the American political landscape. But for others, it's something that surprises them.
What it does do — the most important aspect of Vote Compass — is encourage a focus on the public policy position of the candidates. So often elections, not only in US but around the world, are characterized by horse race politics. Public policy considerations are relegated to almost an afterthought.
But if you're going to vote for a candidate to govern you and your fellow citizens for the next four years, then you should have some sense of how they plan to govern you — not only in terms of character or their approach or their personality, but also the kinds of policies that they're going to put in place.
So Vote Compass encourages a relatively substantial interrogation of those policies. But moreover, it's an interesting platform and interesting way to organize information such that users can get a sense of how those policies align with, or don't align with, their individual views.
One thing I know you're aware of is that voters' political stances don't exist in a vacuum. So the political messaging machine of the parties may drive voters to a platform and not necessarily to individual views. In that sense, doesn't Vote Compass serve as an affirmation tool more than an informative one?
Clifton van der Linden
The first point I should make: Not only is Vote Compass not a tool that is directly advising people how to vote, but it's also not meant to predict people's vote choice. That's key. You may get a result in Vote Compass that is not consistent with the candidate you tend to vote for. And that's okay! You may be voting for that candidate for reasons other than their public policy positions. You may be voting for that candidate with respect to a specific public policy or a policy commitment they've made that maybe isn't included in Vote Compass.
For the sake of brevity, we've only included 30 public policy questions. That's still substantive and a considerable number of issues to include, but there are reasons someone will vote for a candidate that Vote Compass will not capture.
We're not predicting votes, but we are trying to spur or catalyze a moment of reflection, individual reflection, about how one's politics square with the candidates that are vying for their vote on Election Day. And for many people — we've run Vote Compass in almost two dozen elections worldwide in almost half a decade, and we heard from many, many users about how the experience impacted them — the most common message we get is that it sparks discussion or conversation between them and people around them.
People will do Vote Compass with their families. I've had people say, "I never realized how much my sister and I differed in terms of our political views, because we never talked about it. We never had this discussion." Or grandparents do it with their grandchildren and see how they differ. So it's a learning opportunity and an opportunity to start a conversation. And it's an opportunity to have a conversation grounded in public policy, as opposed to the more ambiguous discussion of the character of leaders — which, again, is important but not really the focus of Vote Compass.
Vote Compass was criticized for not following survey design standards. But you've said the tool isn't designed to be a poll but to "discriminate between party positions." Can you explain that a bit more?
Clifton van der Linden
The purpose of a poll, in my view, is to collect information from a sample that can be considered representative of a population of interest, so as to make a generalizable inference about that population.
The purpose of Vote Compass is to survey users in such a way as to provide them with a personalized analysis of their position in the American political landscape and their alignment with the candidates running for office. It's meant to be an educational and informative exercise, as opposed to an information or data collection instrument aimed at providing a snapshot of public sentiment at a given point in time.
Because we are able to tap into the opinion and ideas and values of so many voters in a given election, we actually accrue data sets that are orders of magnitude larger than conventional commercial tools. Our hope is to prompt more responsive government by having unique and credible data about how Americans feel about different policies and practices.
Vote Compass certainly doesn't take a traditional approach to this, because our primary objective is voter information and education. We are constrained in some ways by that primary objective in terms of the survey design — in terms of the design of Vote Compass itself. It's not designed as a poll. And there are some aspects to the design that diverge from conventional poll integrity, because that's not a poll mandate. But we do collect an extraordinary amount of insight into attitudes and opinions and values of voters.
You went to school to be a political scientist — into the weeds of high-level discourse — but then you created this tool that simplifies electoral politics for voters who don't think about this with an academic frame. Why?
Clifton van der Linden
I was a visiting researcher in the Netherlands in 2010 during the Dutch parliamentary elections, and I saw a number of these kinds of applications come out helping position users with the various options or candidates for government. And what I noticed was how compelling this was for people.
You and I are political junkies, so we tend to talk about politics whenever there's an opportunity. Not everyone is as interested, but everyone is as entitled to exercise their democratic right to select their democratic officials. When I saw the level of engagement around these types of tools in the Netherlands, I just thought it was a great opportunity to inspire people to really think about their politics and their political representation.
In terms of the application in the Netherlands themselves, I thought they could benefit from some more rigor, methodological rigor. So I essentially adapted that framework for the Canadian context in 2011 when I launched the first version of Vote Compass for the Canadian federal election here.
In the same way I had seen anecdotally that kind of engagement abroad, we saw 2 million Canadians use the application within the five-week election campaign period, which was just an incredible uptake no one could've anticipated. So I think the opportunity to promote democratic engagement is really the motivating force for me and my colleagues.
On the academic side, it's an opportunity for us to learn, as well. We can learn different things from these kinds of data sets than we have from the existing empirical work that's been done in political science. So it gives us a unique opportunity to contribute to the understanding of how democratic societies operate, and hopefully advance the practice of democracy.