CBC News with Melanee Thomas 17 May 2018
Imagine a spreadsheet with 66 columns and 1,200 rows. We had 700 pages of tables with almost 80,000 data points.
And, over the past few weeks, we’ve pumped out a lot of stories based on this data. There have been stories on Albertans’ values, on our hopes and dreams, on what divides us and what unites us.
But a lot of you are probably wondering: What does this all mean for next year’s provincial election?
We’ve told you the traditional horse-race results — the UCP is comfortably in the lead — but that only captures how voters were feeling when the poll was conducted in the spring. And a year is a long time in politics. So, does the richer data set tell us more about what influences voter behaviour, more broadly? Can it predict how people are likely cast ballots in the future?
The answer is: Yes, to a degree. But it’s complicated.
Nothing is certain, of course, yet the numbers do offer a glimpse into what makes it more or less likely that a voter will support the UCP, NDP or another political party come election time.
Data scientist John Santos, who works with pollster Janet Brown, did some advanced statistical analysis of the polling data.
Those findings reveal some interesting things about what drives voter sentiment and which factors might motivate how Albertans vote in next year’s provincial election.
The model is based on what’s known as a regression analysis. The process is a bit complicated but we recently delved into the details of how it works — and what it all means — with both Santos and University of Calgary political scientist Melanee Thomas.
You can listen to that whole conversation below. We’ve also provided an abridged version, in a Q-and-A format, at the bottom of this story.
In short, the model is based on a variety of voter attributes. This includes how voters feel about various issues — ranging from the carbon tax to balancing the provincial budget — as well things like where they live, how old they are, and their levels of income, education, religiosity, and so on.
It then parses the data to focus on one attribute at a time, while controlling for what Santos describes as “spurious” correlations.
“A good example is that taller people are more likely to vote conservative, which is true in a very weak sense, he said. “But … given that men are typically taller than women and men are typically more conservative than women … once we control for gender, the relationship between height and conservativism sort of disappears.”
Think of this statistical analysis as a big funnel. We put all the data through the statistical tube to figure out what might prompt Albertans to vote a certain way.
Some of the results may surprise you; others might seem entirely predictable. We’ve compiled some of the attributes with the biggest expected impact on voter behaviour in the interactive chart below.
The advanced data work also involved a “cluster analysis” — a broad grouping of voters — based on the numerous questions we asked in the poll about Albertans’ beliefs and attitudes.
Santos then plotted the survey respondents along two different lines: their social values and their economic values.
This method identified four potential clusters of voters: consistent conservatives, consistent progressives, moderates and a fourth group dubbed “prairie populists.” (More on that in a moment.)
Take a look for yourself at the four clusters in the interactive chart below.
Here’s how it works.
Economic values are plotted on the horizontal axis. So the left side is more left-wing, the right side is more right-wing.
Social values are plotted on the vertical axis. Up is more conservative, down is more progressive.
Each marker on the chart represents a different survey respondent and where they fall on these two spectrums. The four clusters are identified by colour and have differently shaped markers.
What does this all mean?
The CBC’s Brooks DeCillia sat down with both Santos and Thomas to talk about that — in detail.
You can listen to their full convesation in the audio player above, or read an abridged version in the Q-and-A below.
Q & A
Brooks DeCillia: Melanee … what values might be underlying how Albertans are approaching politics?
Melanee Thomas: Populism matters. … Here, we’re talking about prairie populism. It has a long history in the 20th century. It cuts across the entire ideological spectrum. This is just about who advocates for the little person … against large vested interests. And where people sit on the ideological spectrum will change who the little people are and who the vested interests are.
But the interesting thing that I find that comes from these data is that the NDP is seen to the more populist party. This came through in 2015. Rachel Notley was much more of a populist and seemed to be advocating for regular Albertans more than, say, Jim Prentice.
One of the things that comes through very clearly [in this data] is that populism is a positive predictor for the NDP. This doesn’t surprise me but I imagine, for others, this looks weird.
Brooks DeCillia: What did you learn about age from this model?
John Santos: So, it’s really important to remember that these models control for a lot of different factors.
And when we look at age, what’s interesting is that it actually isn’t all that good of a predictor. Once we start adding in other controls, we find that the only significant predictor is that those aged 25 to 44 are actually a little bit less likely to vote for the NDP compared to those who are aged 45 to 64.
And what that says is patterns we might see between age and voting are largely a reflection of other things — not of age, per se.
Melanee Thomas: One thing I would forcefully push back on … is this idea that, because Alberta is a young place, it’s skewing to the left. There’s a couple of narratives like this, where people try to explain … the 2015 election based on this idea that we’ve got a lot of youth in the province, that there’s been an in-migration of a lot of young people.
And I can understand, intuitively, why that works. But this idea that age has this consistent, monolithic effect on vote choice? We simply don’t have evidence for that.
Brooks DeCillia: Let’s talk about gender.
Melanee Thomas: We’ve got a modern gender gap in Alberta. This means that women are more likely to support parties of the left than our men. Men are more likely to support parties of the right than are women.
The generalization that we have from broader Canadian research is that this is because women are less socially conservative and more skeptical of economic liberalism than are men. And it turns out that this is exactly what we see in the Alberta data.
It’s explained by gender differences in values and beliefs. Men are more likely to be economically conservative. Men are more likely to be socially conservative.
Brooks DeCillia: And what about religiosity? What stands out for you there?
Melanee Thomas: One of the things that’s worth noting is that a lot of people aren’t very religious in Alberta. I have students almost every year who assume that we are the Bible Belt in Canada. We are not. That’s Atlantic Canada, actually in terms of devoutness.
Brooks DeCillia: [If you] were a political party looking at this data … what would you think?
Melanee Thomas: If I were the UCP, I would say how do we manage to keep things quiet enough that we don’t ruin anything. Basically, the perceptions are in our favour. We just need to make sure that people continue to like the leader and there are no bozo eruptions.
If I were the New Democrats, one of the things that I would find frustrating would be: How is it that we can work on a lot of these issues and then not get credit for them? And this is where issue ownership and those kind of perceptions remain strong. The other thing would be: How do we make the party all about the leader? Because people like Rachel Notley a lot more than they like the party label.
John Santos: The UCP definitely needs to just sort of keep their head down and try not to shoot themselves in the foot. … If they keep that to a minimum I think that’s the best strategy for them.
The NDP, they need to solidify their hold on their base — so, this group of voters who are economically left, socially left. The NDP needs to win on home ice first. And then they need to plot their expansion strategy into this cluster of people who are definitely socially progressive but economically sort of on a range of issues.
And it’s interesting, because that cluster of moderates, a lot of them are voting for the Alberta Party. And I suspect that, if push came to shove and the antipathy those folks have toward Kenney is high enough, they might go to the NDP. But it’s just not there right now.
The random survey of 1,200 Albertans was conducted using a hybrid method between March 13 to April 5, 2018, by Trend Research under the direction of Janet Brown Opinion Research. The sample is representative along regional, age, and gender factors. The margin of error is +/-2.8 percentage points, 19 times out of 20, and larger for subsets.
The survey used a hybrid methodology that involved contacting survey respondents by telephone and giving them the option of completing the survey at that time or later, or completing it online. The response rate among valid numbers (i.e. residential and personal) was 20.8 per cent.