Wednesday, July 29, 2015

New Ipsos poll, Battleground Toronto, and this site's name

After Friday's confusion arising from diametrically opposed polls being released on the same day, Ipsos Reid cleared things up a little yesterday evening by splitting the difference in its poll for Global News.

The CBC Poll Tracker has been updated with these latest numbers, and continues to show a tie between the Conservatives and New Democrats. The riding-by-riding projections have also been put up-to-date.

I identified what was potentially the main reason for Mainstreet Research's divergence from the consensus in my Poll Tracker analysis on Friday: namely, that Mainstreet had some abnormally high scores for the Conservatives among 18-to-34 year olds.

After removing the undecideds myself, I calculated that the Conservatives had 43% support among this age group in Mainstreet's poll. That was very high. You can see how high in the chart below. Mainstreet's poll is the second last in this chart, and is in italics. All the other polls are recent findings from other pollsters among this age group. Ipsos's latest is at the bottom of the chart, showing the Conservatives at just 17% among this group.

18-to-34 year olds, Mainstreet in italics
So, you can see that this is a bit of an issue. If this age group represents about 20% of the weighted sample, the discrepancy between Mainstreet's numbers and those of other pollsters would be worth about four points to the Conservatives. That drops them from the 38% of that poll to 34%, just about where Ipsos has them.

Of course, everybody else could be wrong or this could have been a momentary excitement on the part of young Canadians that lasted a couple of days, but the odds of that would seem rather low.

Leafs not the only blue team likely to struggle this fall in Toronto

In light of the Liberal nomination battle in Eglinton-Lawrence that saw Marco Mendicino defeat Eve Adams on the weekend, and yesterday's announcement that Olivia Chow was returning to the federal scene, I took a look at how the parties's chances are shaping up in Toronto for the CBC.

I hadn't fully realized something before writing this piece: Toronto itself, and not just the GTA in general, was very important to the Conservatives in 2011. They won eight seats there, after winning zero seats there in every election prior to that going back a few decades. Eight seats out of 22 in Toronto is quite good (tying the NDP), but it is very important when you consider that the Conservatives made a net gain of 23 seats between the 2008 and 2011 elections. That means Toronto represented just over one-third of the Tories' gains in the last election. The 905 is important for the Conservatives, but the 416 is no slouch either.

ThreeThirtyEight not going to happen, so please stop asking me

The name of this site,, is in homage to Nate Silver's FiveThirtyEight as well as the number of seats that are currently in the House of Commons and were in the House when the site was launched in 2008. It might as well also represent the number of times I have been asked when I will be changing the name of the site.

Let me clear it up once and for all (not that I've been saying anything different from the last few years). The name of the site is not going to be changed to ThreeThirtyEight or variations of that. And the name of the site is already quite long, so changing it ThreeHundredThirtyEight, as some have suggested, would be a step in the wrong direction too.

A few reasons for why I will not be changing the name of this site to reflect the new number of seats in the House of Commons:

1) It has been this way for almost seven years, which gives this name its own branding strength. It would be confusing to suddenly change the name and I'd likely be correcting people for a year or two. While the site was named after the number of seats in the House of Commons, it was never meant as a running tally of the number of seats in the House of Commons.

2) Four elections occurred with 308 seats up for grabs. You have to go back to 1965 to find a string of at least that many elections with the same number of seats. It is a particularly interesting era to name the site after, with three minority governments and the newly merged Conservative Party. Stephen Harper, Jack Layton, and Gilles Duceppe led their respective parties in all four of thsoe elections, and it is hard to name three federal leaders that have more greatly marked their respective parties or Canadian politics in the 21st century. Harper embodies, in many ways, the merged Conservative Party, Duceppe has led his party for almost 20 years, and Layton set the NDP on the path that might bring it to power this fall. The 308-seat House was also the only House I ever really knew, having started getting interested in politics after the 2000 election.

3) This site was launched in 2008 because of how interesting I found FiveThirtyEight. But to change the name of the site to ThreeThirtyEight (or 338 as many would call it) gets the name much too close to FiveThirtyEight (538). That would also lead to confusion, as well as being discourteous to 538.

4) If this site still somehow exists in another 10 years, I'll have to change the name yet again.

5) Last I checked, the URL was taken anyway so the whole thing is moot.

Friday, July 24, 2015

Welcome to the Pollcast

Read the headline here while channeling your inner Don Newman. "...the P-o-o-o-l-l-l-l-l-l-l-l-c-a-s-t."

Very excited to launch the CBC Pollcast today! You can find it here, or listen to it directly here.

Every week, and perhaps more often once the campaign gets going, we'll be inviting guests to the Pollcast to discuss everything about the polling industry: how polls are done, what challenges the polling industry faces, what influence polling has on election campaigns and, of course, the trends, both national and regional.

Joining me this week is David Coletto, the CEO of Abacus Data. We discussed how polling is done in this new modern world of ours, and we saved a little time to talk about what the polls are showing right now. David had some interesting thoughts on what it will take to move the numbers.

We're going to get at the topic of polls from every angle in future episodes. It isn't just about the horserace.

We'll have more pollsters on to discuss what they are seeing in their numbers, as well as to discuss issues related to the industry and the relationship pollsters have with the media. We'll have members of the media on to discuss that relationship and the role of polling in political journalism. We'll also have them on to give some context to the numbers themselves. And we'll have some party insiders, too. Not to spin the numbers (though they might try!), but to discuss how parties are using polls in their campaign strategies.

I think it will be a lot of fun, and I hope you'll enjoy it. And it's a podcast for the audience. We're not going to shy away from getting into the details - in this week's episode we even get into sampling quotas - like is often the case on other platforms.

Please let me know what you think, and what you'd like to see from the Pollcast!

Wednesday, July 22, 2015

How the leaders stack up, literally (almost)

In my column today for the CBC, I compared each of the party's approval ratings by looking at polls conducted since mid-June. The crux of it is that Thomas Mulcair's numbers are good and getting better, Justin Trudeau's are middling and getting worse, and Stephen Harper's are bad and not improving. Click on the link for the full analysis.

It is interesting to do a regional comparison of the leaders, though, to see how they stack up. In the chart below, I've ordered the three leaders according to their net approval ratings (approval minus disapproval).

The first thing you see is that there is a clear order for the three leaders. Mulcair is on top, Trudeau in the middle, and Harper at the bottom. This is because Mulcair's approval ratings are all a net positive, Trudeau's are mostly breaking even, and Harper's are mostly negative.

But there are a few interesting divergences from this.

Start with Mulcair. His approval rating in Quebec is stellar, and it is also very good in British Columbia. This is important as these are the NDP's two keys to a minority government. But he also has good scores in Atlantic Canada and Ontario.

His popularity drops off in the West, however. He is less popular in the Prairies and Alberta than Trudeau is in Atlantic Canada and British Columbia, while he ties Harper in Alberta. That latter score is actually quite good, relatively speaking. Harper is from Alberta after all.

For Trudeau, he only has good numbers in Atlantic Canada, where he is more popular than Harper is in any part of the country. The same can be said for his modest result in British Columbia. He breaks even in Quebec, the Prairies, and Ontario, but that puts him low on the table.

His very low score in Alberta, though, suggests the party may not have bright chances for a breakthrough. Albertans dislike Trudeau as much as Ontarians dislike Harper. Only in Quebec, British Columbia, and Atlantic Canada, where Harper is deeply unpopular, does anyone score worse than Trudeau in Alberta.

For Harper, none of these numbers are very good. Even his net +2 in Alberta is worrisome for the Conservatives, as that is supposed to be their fortress. His low scores in battleground provinces like Ontario and British Columbia do not bode well for a Conservative re-election, and he is plumbing the depths of unpopularity in Atlantic Canada.

If there is a silver lining for Harper, it is that he leaves few people unmoved. That means that his low scores still give him an approval rating of 30% or more in Alberta, the Prairies, and Ontario. Enough to keep him competitive, if those people who approve of him go out and vote for him. But he's a long way from a majority government at these numbers.

Just for fun, let's calculate how each of the parties would do if they could replicate their leaders' approval ratings in every region of the country. To do this, I've reduced (or increased, as need be) the support of other parties equally to accommodate for the adjustments.

For the Conservatives, they would still be in it - but only just. The Conservatives would win 127 seats, putting them nearly in a tie with the NDP and far from majority status.

That might seem pretty good, considering how bad Harper's numbers look and where the Conservatives stand in the polls right now. But when we compare it to what Mulcair's and Trudeau's approval ratings would produce, we see how far Harper trails.

If the Liberals replicated Trudeau's approval ratings at the ballot box, the Liberals would fall just short of a majority - but close enough that it would still be possible. They'd win 165 seats, taking 65 in Ontario, 35 in Quebec, 25 in Atlantic Canada, and 21 in British Columbia. The NDP and Conservatives would be in a tight race for runner-up spot.

It would not be nearly as competitive in the NDP's scenario, though. If the NDP replicated Mulcair's approval ratings, the party would win a landslide. The model would give the NDP 257 seats, with the Conservatives and Liberals splitting 80 seats between them. The NDP would win all but three seats in both B.C. and Quebec, and almost 100 in Ontario. If Mulcair's approval rating represents a ceiling for the NDP, then the sky is (nearly) the limit.

This simple exercise shows how Mulcair and Trudeau are draws for their party, whereas Harper does about as well as his party is doing. After more than 11 years as leader, Canadians may see Harper and the Conservative Party as inseparable.