When it comes to politics, it's not disagreeing that's the problem — it's that we like each other less

We’re not seeing an exodus from the middle to the extremes of the ideological spectrum. We're seeing Canadians who identify with a party becoming more negative towards other parties.

Polarization has barely changed with respect to policy positions; it's the emotional kind that has grown

Anti-pipeline and pro-pipeline demonstrators are seen in these file photos. Is it policy that divides Canadians the most? Or partisanship? John Santos says the data points to the latter. (Darryl Dyck/Canadian Press, Peter Evans/CBC)

Everyone's talking about polarization. We're more divided than ever, commentators say. Politics is so divisive these days! 

But is this true? And what does "polarization" even mean?

There are at least a couple of ways to understand the term.

Defining 'polarization'

One way to understand polarization is a shift in the distribution of public opinion such that there are fewer people in the middle and more people at the ends of the ideological spectrum.

Think back to the "bell curve" you learned in your statistics class. A normal distribution of opinion would look like a bell with a single peak in the middle. An ideologically polarized distribution would have two peaks on either side and a valley in the middle.

Another way is "affective" or emotional polarization ("affect" being another word for feeling or emotion). This means partisans are growing more negative toward parties other than their own.

This type of polarization can happen regardless of whether the parties disagree with each other. In fact, research shows partisans dislike supporters of other parties, even if those other partisans have the same policy positions as they do. Conversely, they still like partisans on their own team with whom they disagree on policy.

In Canada, we're mostly seeing the latter form of polarization. The parties are being increasingly differentiated by the positions they take and the preferences held by those who identify with them. And those who identify with a party are also becoming more negative toward other parties. 

What we're not seeing is an exodus from the middle to the extremes of the ideological spectrum.

Looking at data

Let's look at some data from the Canadian Election Studies (CES). These are surveys conducted by a team of political scientists during and after each federal election campaign. As public opinion polls go, these are the gold standard, and they are used widely in academic research.

(Note, because politics works a bit differently in Quebec, I only look at results from the English-speaking provinces.)

The CES has a question that asks voters to place themselves on a left-to-right scale, ranging from zero to 10. Zero means very left-wing, ten means very right-wing, and five means centrist. 

If Canada has become more polarized over time, what we should see is fewer people placing themselves in the middle of the spectrum and more people placing themselves on the extremes. This is not really the case, as we see in the following graph.

Between 1997 and 2015, the proportion placing themselves dead in the middle of the spectrum (five out of 10) went from 33 per cent to 30 per cent. The proportion placing themselves at the farthest left position on the spectrum increased from six per cent to 10 per cent. And the proportion placing themselves at the farthest right position of the spectrum decreased from 14 per cent to 11 per cent. 

All of these changes are within the margin of error.

The picture is clearer when we "fold" the spectrum and combine the far left and far right into one category, combine the left and right into one category, and keep the centrists as a third category. See how that looks below.

The proportions of people in each of these three categories is roughly the same over this period. The centre has not collapsed, nor has there been a significant increase in the proportion of people placing themselves at the extremes of the political spectrum.

Where things have become polarized is that partisans of one party dislike the other parties more and more. This is affective polarization.

Partisans don't like each other

The CES also collects data on individuals' partisan identities and evaluations of other parties. 

Someone has a partisan identity when they have an enduring psychological attachment to a party — so much so that their bond to their party is a part of how they see themselves. In turn, this identity colours how they perceive reality, and can even shape their values

Evaluations of other parties is measured through a "feeling thermometer" question. This asks a survey respondent to rate each of the parties on a zero-to-100 scale, where zero means very negative and 100 means very positive. 

We can look at how partisans of the different parties rate each of the parties over time. If partisans are becoming increasingly positive about their own party and/or increasingly negative about the other parties, then we can say affective polarization has taken place. 

(Richard Johnston from UBC and Christopher Cochrane have written about this before in much more depth than I cover here. Anyone interested in polarization should read their work.)

It is no surprise that partisans of all stripes rate their own party higher than they rate other parties. But, the love of their own party does not seem to be increasing over time. What has changed is how partisans feel about the other parties. 

Looking at Liberal identifiers, we see a mixed picture of affective polarization.

Since 1988, Liberals have become more positive toward the NDP (43 points in 1988 to 54 points in 2015) and more negative toward the Progressive Conservative Party and, later, Conservative Party (43 points in 1988 to 34 points in 2015). Liberals were even more negative toward the Reform and Canadian Alliance parties in the 1990s than toward the Progressive Conservatives of that time. And, they have been quite positive about the Green Party since the CES started asking about feelings towards the Greens in 2011.

The picture is similar with NDP identifiers.

They have become more positive toward the Liberal Party (42 points in 1988 to 56 points in 2015) and more negative toward the Conservative Party (36 points in 1988 to 24 points in 2015). New Democrats were also more negative toward the Reform and Canadian Alliance in the 1990s than the PCs. And their feelings toward the Greens are similar to their feelings toward the Liberals in 2011 and 2015.

The picture is a bit different with Conservative identifiers.

They have had consistently negative impressions of the NDP (32 points in 1988 and 31 points in 2015). Their feelings toward the Liberal Party have gone from slightly negative (41 points in 1988) to neutral (51 points in 1993) to negative (33 points in 2015). By 2015, Conservatives dislike the Liberal Party as much as they dislike the NDP.

Interestingly, they were more positive about the Reform and Canadian Alliance parties than their own party in the 1990s. And they are even more negative toward the Greens than either the Liberals or the NDP in 2011 and 2015.

This analysis is not complete without looking at non-partisans.

Their evaluations of the parties form a boring graph with relatively stable lines that hover between neutral to slightly negative (50 to 40 points). Non-partisans rate the parties similarly, with slightly less negativity toward the Liberals.

There is change among the non-partisans, yes. Of course, there would be fluctuations over nine elections spanning 27 years. But the differences are not stark, nor are there clear trends over time.

On one hand, non-partisans' ambivalence is not surprising. On the other hand, it challenges a popular notion that everyone is more negative about politics these days.

One might be tempted to assume non-partisans are non-partisans because they are equal opportunity haters — folks who say, "a plague o' all your houses!" But the data do not show that. Non-partisans are neutral to all the parties. This suggests that the increased animosity in politics these days is mostly a partisan-on-partisan affair.

So what?

OK, so there isn't an exodus to the extremes of the political spectrum and people just like each other less. So what?

For parties and partisans, this should be a wake up call to dial back the rhetoric and resist the temptation to call the people on the other side a bunch of ideologues who are un-Canadian and hate Canada. These are cheap ploys designed to press individuals' psychological buttons for political gain.

Short of calling for people's rights to be taken away, most political disagreements should be within the acceptable boundaries of debate. It should not be considered un-Canadian to have a discussion on whether there should be more private delivery of health care services. Nor should it be considered un-Albertan to question if more should be done to promote industries other than oil and gas.

The whole point of living in a democracy is to be able to disagree and to have a way to resolve those disagreements without resorting to violence.

It should worry all Canadians when politicians receive death threats. Most of these threats do not come to fruition and there is a mental health aspect to many terrorist attacks, but an affectively polarized, us-versus-them political environment can encourage radicalization, which is a step along the path toward politically motivated violence.

For non-partisans, this should be a call to hold parties accountable when they take those cheap political shots. The evidence from the CES suggests non-partisans are ambivalent toward the parties and are not biased for or against them. But that doesn't mean individual non-partisans are not susceptible to having their emotions manipulated for partisan gain on a particular issue.

So, when it comes to polarization, it's not disagreeing that's the problem — it's being disagreeable about it.


John Santos

Data Scientist

John Santos is a data scientist with Janet Brown Opinion Research.