Bubble Pop with Rachel Gilmore - April 04, 2025


Demystifying the witchcraft of election polls (with Philippe J. Fournier)


Episode Stats


Length

45 minutes

Words per minute

171.56

Word count

7,781

Sentence count

426

Harmful content

Toxicity

2

sentences flagged


Transcript

Transcript generated with Whisper (turbo).
Toxicity classifications generated with s-nlp/roberta_toxicity_classifier .
00:00:00.000 Right now, I'm just going to tell you if unless there's a big outside event or a major stumble during the debates or during the campaign, all the indicators point to a liberal victory.
00:00:13.320 But this is not a prediction.
00:00:15.540 I'm sorry.
00:00:16.160 I'm very sorry.
00:00:16.900 That's okay.
00:00:17.620 No, I knew as soon as I was looking at my wording of the question, I was like, he's not going to want to give a prediction.
00:00:22.940 That's not what these guys do.
00:00:30.400 Hello, I'm Rachel Gilmore, your least favorite person's least favorite journalist.
00:00:34.300 Welcome to Bubble Pop, a weekly show where we break down the biggest headlines so you
00:00:38.460 don't have to doomscroll. Then we'll dive deeper into something juicy that you should know about.
00:00:42.980 In this episode, we've got the man behind those maps that you're probably seeing everywhere.
00:00:47.920 You know, those aggregations of all of the polls that our politicians probably see in their dreams
00:00:53.700 and nightmares? That's right, Philippe J. Fournier, the creator of 338 Canada, is on the show.
00:00:59.560 But first, the news.
00:01:04.200 Tariffs are here, and it's bad.
00:01:06.120 Not as bad as it could have been, but that's kind of like being forced to listen to a Nickelback album on a really long car ride.
00:01:13.140 Even if it's the best album, it's still Nickelback.
00:01:17.060 And these are still tariffs.
00:01:18.500 There's also been drama on the campaign trail this week, as several candidates' past comments have come back to haunt them.
00:01:24.420 Leading to several, mostly conservative, candidates dropping out.
00:01:28.300 We're also gonna take a little peek at the crumbling democracy next door because ultimately
00:01:32.620 that's gonna impact us too, so. And things are getting weird. Let's dive in.
00:01:39.900 Well, well, well, looks like the entire world is once again getting to enjoy the fruit of the
00:01:44.640 United States' terrible decision-making. And that even includes the penguins. That's right,
00:01:50.240 Trump even slapped tariffs on islands that are totally uninhabited other than by penguins.
00:01:55.500 Those penguins, they've treated us very unfairly, very unjust.
00:02:00.060 Those penguins, they don't have the cars.
00:02:02.380 The penguins don't have the cars.
00:02:03.860 Anyways, Canada could have done worse here, as I already said.
00:02:06.780 But that doesn't mean this is good.
00:02:09.420 Here's where things stand as I'm filming this on Thursday afternoon.
00:02:12.760 Because, as you know, anything could change in a heartbeat here.
00:02:16.080 We're all subject to Trump's whims, which is, like, so fun.
00:02:20.240 So fun.
00:02:20.840 Canada managed to avoid the 10% baseline tariffs that Trump was just throwing at everyone,
00:02:26.060 but there's still a 25% tariff on vehicles going from Canada to the U.S. and on our aluminum and
00:02:32.020 steel, as well as a slightly lower tariff on our energy sector. So we are going to feel pain,
00:02:37.460 and for some, they're feeling it right now. Thousands of people are out of work for two
00:02:41.860 weeks in Windsor, where auto manufacturer Stellantis has said it plans to shut down
00:02:46.360 its plant for two weeks starting April 7th. And they say more changes to the schedule are expected
00:02:51.600 in weeks ahead. You've got to really feel for these workers who are just sitting there waiting
00:02:56.640 to find out if they're losing their livelihood. That's just such an unbearable position to be in.
00:03:01.980 Now, Canada isn't just taking this lying down. Prime Minister Mark Carney on Thursday laid out
00:03:06.780 our plan to punch back at the U.S. over these tariffs. He said these tariffs are unjustified,
00:03:12.340 unwarranted, and misguided. And Canada will respond with a 25% tariff on imported vehicles
00:03:18.300 that aren't compliant with CUSMA, our trade deal. Unlike the U.S. tariffs, though, these tariffs
00:03:23.040 won't impact auto parts, Carney said. So that way there's less impact on our seriously integrated
00:03:28.720 auto sector. And the tariffs also won't impact vehicle content from Mexico. The Prime Minister
00:03:33.840 also said that all of the money they make from these tariffs will go towards supporting workers.
00:03:38.320 Carney, also speaking on Thursday, had some big picture warnings about where our relationship with the U.S. is going.
00:03:45.040 Yesterday's actions by the U.S. administration, well, not specifically targeting Canada,
00:03:49.640 will rupture the global economy and adversely affect global economic growth.
00:03:55.160 The global economy is fundamentally different today than it was yesterday.
00:03:59.040 The system of global trade anchored on the United States that Canada has relied on since the end of the Second World War,
00:04:05.280 a system that, while not perfect, has helped to deliver prosperity for our country for decades, is over.
00:04:13.440 Our old relationship of steadily deepening integration with the United States is over.
00:04:18.280 The 80-year period when the United States embraced the mantle of global economic leadership,
00:04:22.760 when it forged alliances rooted in trust and mutual respect,
00:04:25.840 and championed the free and open exchange of goods and services, is over.
00:04:31.260 While this is a tragedy, it is also the new reality.
00:04:35.280 we must respond with both purpose and force.
00:04:40.860 We are a free, sovereign, and ambitious country.
00:04:44.860 We are masters in our own home.
00:04:47.000 And he also said that if the United States has walked away from its leadership role in the world,
00:04:52.320 Canada will seek out other countries that share our values.
00:04:55.780 And if the U.S. doesn't want to lead, Canada will.
00:04:58.780 Canada must be looking elsewhere to expand our trade,
00:05:02.680 to build our economy and to protect our sovereignty.
00:05:07.040 Canada is ready to take a leadership role
00:05:09.160 in building a coalition of like-minded countries
00:05:11.640 who share our values.
00:05:14.500 We believe in international cooperation.
00:05:18.240 We believe in the free and open exchange
00:05:20.580 of goods, services and ideas.
00:05:23.620 And if the United States no longer wants to lead,
00:05:27.420 Canada will.
00:05:28.700 Meanwhile, Polyev, after laying out a plan yesterday
00:05:31.440 to turn Canada into a quote-unquote economic fortress to withstand Trump's attacks. A plan
00:05:36.920 that includes a whole lot of tax cuts, a business liquidity support program, and lots of focus on
00:05:41.800 oil and gas development to lessen Canada's dependence on the U.S. market. But not a lot
00:05:46.220 of talk on the climate crisis and how he's going to mitigate that. Then, after the tariffs were
00:05:51.140 announced, Pauliev said Thursday that if he's elected prime minister, he'd remove the GST from
00:05:56.080 Canadian-made autos. That's an idea that was first put forward by the NDP, interestingly.
00:06:00.960 Speaking of which, NDP leader Jagmeet Singh suggested we bring back those tax-free victory
00:06:05.980 bonds that we had during both world wars to raise the money needed to fight. It's important to note
00:06:11.420 that the Canadian government stopped selling these bonds years ago, in part due to the high
00:06:15.920 administrative costs. But when pressed on those concerns, Singh batted them down, saying that we
00:06:20.680 are in a, quote, very different context right now. Oh, and the Bloc Québécois called for greater
00:06:25.980 Quebec representation in the delegation that'll be dealing with the Americans. That comes on the
00:06:30.560 heels of the Trump administration listing a Quebec language law as a trade barrier. Which, um,
00:06:37.720 my god, do you know what you've done? Do you know what you've awoken? All I can say on that,
00:06:43.600 les États-Unis, bonne chance. Some other big stuff happened in the election campaign this week,
00:06:50.000 and while Trump's tariffs really overshadowed basically everything else, I still think it's
00:06:54.620 worth mentioning. The big theme this week, again other than tariffs, seemed to be candidates' pasts
00:06:59.680 coming back to haunt them. Multiple candidates were dropped this week over unveiled old comments,
00:07:05.460 all but one being from the Conservative Party. Paul Chiang, a Liberal candidate who Mark Carney
00:07:10.500 initially stood behind in this controversy, which felt boneheaded given the severity of it,
00:07:16.120 suggested folks should hand a Conservative over to China to collect the bounty on him. Something
00:07:22.000 that, if it was a joke, was really not funny. But I feel like I'd be remiss if I didn't mention that
00:07:26.980 the Conservative candidate should be a little careful about what he's saying on his Twitter.
00:07:31.720 But anyways, Carney accepted Chiang's resignation after the RCMP announced it was looking into his
00:07:36.840 comments. Meanwhile, the Conservative Party ditched four candidates over a two-day stretch, which is
00:07:42.780 wild. And most of them were over past comments that these candidates had made. The most recent
00:07:47.820 Conservative candidate to get the boot, Don Patel, got turfed after CTV News reported that he had done
00:07:53.000 a care reaction on Facebook in response to a comment calling for Canadian residents to be
00:07:58.480 deported to India so Prime Minister Modi can, quote, take care of them. And that came after,
00:08:05.420 on Tuesday, the party ditched their candidate Mark McKenzie, who was running in Windsor. After CTV
00:08:10.160 News obtained audio of his right-wing comedy podcast from 2022, which, uh, that always works
00:08:16.160 out well, doesn't it? Anyways, in that podcast, he apparently joked about Justin Trudeau getting the
00:08:21.220 death penalty. Mackenzie also voiced his support for, um, public hangings. So yeah, he got the
00:08:27.320 boot. And then hours later, Stéphane Marquis, who was running for the Conservatives in a Montreal
00:08:31.120 riding, shared that he had also gotten the axe. In a statement, he said that if you, quote,
00:08:36.060 publish a non-vetted tweet on X, you can, quote, expect ostracization from your natural ally.
00:08:42.380 That night, a third candidate, Laurent Sieng, who was running NBC, was kicked out. But the party
00:08:47.280 didn't say why, and Singh has said he plans to appeal it. So yeah, that's a pretty wild two-day
00:08:53.480 run. And honestly, if the Conservative Party actually cares now about the stuff that its
00:08:57.680 candidates have said online, this could just be the beginning of the bloodletting. Because they've
00:09:02.480 picked up candidates like Roman Baber, who has suggested you shouldn't trust doctors or public
00:09:07.420 health officials, and Aaron Gunn, who said the gender pay gap doesn't exist. It does. I'm sure
00:09:13.560 that'll also really help Paliyev with women who were pretty upset with him this week after he
00:09:17.920 made comments about their biological clocks. Although I hate the fact that I had to hear him
00:09:22.920 use those words, it did lead to this very funny clapback from Jagmeet Singh. Do you think it's
00:09:28.440 appropriate for Paliyev to be invoking biological clocks and a woman's ability to get pregnant in
00:09:33.640 this context? I don't think any woman wants to hear Paliyev talking about their body, period.
00:09:38.920 period don't do it meanwhile the big story of not only the election campaign but like
00:09:46.920 in general in our lives is still the tariffs and it's important to note just how bad this is even
00:09:53.060 just within the u.s where wall street is losing its mind americans are terrified rightly so of a
00:09:59.920 potential recession and trump is turning around and comparing it to a successful operation which
00:10:05.380 actually kind of makes sense because getting an operation in the U.S. can leave you financially
00:10:10.580 destitute. Meanwhile, Elon Musk seems to be slightly on the outs with the Trump administration.
00:10:15.540 This comes after his blatant attempts to use his wealth to get his way in Wisconsin blew up in his
00:10:20.620 face. Because despite spending $20 million to try to get a pro-Trump candidate elected to the
00:10:26.680 Wisconsin Supreme Court, the liberal candidate won. And now Trump seems to have a new, worse
00:10:31.880 conspiracy fiddler in his ear, Laura Loomer. This is one of those monkey's paw moments, huh?
00:10:39.000 The White House fired multiple administration officials, including three National Security
00:10:43.980 Council staffers, after a woman who once claimed 9-11 was an inside job said they were disloyal.
00:10:50.240 I'm gonna leave the news there this week because I think we're all very tired. I know I am after
00:10:55.560 this wild week. I've honestly been so overwhelmed and grateful for the public reaction, the huge
00:11:01.860 public reaction, oh my god, that's like huge, to CTV dropping my fact-checking segment after there
00:11:08.900 was a ton of bad faith online right-wing outrage, including from a member of Pierre Polyev's team.
00:11:15.200 It was a really hard week, but it also gave me some hope that together we can make a difference
00:11:21.540 and actually make the world a little bit better. And I hope that CTV learned its lesson, because
00:11:26.100 I'm not going to be the last person that this happens to. This is a playbook that works the
00:11:31.140 online right wing has figured out if they kick and scream loud enough no matter how bad faith it is
00:11:36.220 that a lot of news organizations will capitulate. But if those news organizations don't and they
00:11:41.560 actually have a backbone in the face of those attacks we can not only protect more journalists
00:11:46.680 and allow them to do the kind of reporting that kicks up those kinds of outrages because a lot
00:11:52.980 of the most important reporting does. And in the process I actually think that protects our
00:11:57.020 democracy. Not to be too dramatic, but I'm on to better things. The incredible folks at the
00:12:03.040 National Observer have picked up my fact-checking segment, and I'm going to be doing that work
00:12:07.480 for them. I'll be working with Emily Baron-Kadloff, and we're going to be putting that out regularly.
00:12:12.000 I'll be publishing it on my socials, so keep an eye out for it. I'm super excited. And I've said
00:12:16.520 this before, but I do think it's really important to point it out, that the only reason I was able
00:12:20.520 to stick up for myself and speak out in the face of CTV doing all this was because you guys support
00:12:25.880 my work and that has allowed me to be independent so even if I burn a bridge with someone like CTV
00:12:32.200 that doesn't destroy my career because you guys have given me a career without the need of those
00:12:39.100 mainstream institutions so I'm really grateful for you and I just want you to know that I recognize
00:12:44.200 that this is because of you so yeah thank you so much anyways enough about me after the break
00:12:51.380 we're nerding out. We've all seen those maps and graphs that go viral constantly telling us who
00:12:56.320 is currently winning or losing in the polls amid the federal election campaign. Well, after the
00:13:02.140 break, Philippe J. Fournier, the creator of 338 Canada, which is the thing behind most of those
00:13:08.140 maps and graphs that you're seeing, is gonna join me. This nerd is very excited, and I hope you are
00:13:13.840 too. We'll be right back. You know those maps and graphs you see throughout the election campaign?
00:13:24.740 The ones that tell you who seems to be winning, who seems to be losing, and whether your province
00:13:29.160 or territory is going liberal, conservative, or even maybe NDP? Well, there is a good chance that
00:13:34.920 what you're looking at is from 338 Canada, which is a statistical model of electoral projection
00:13:40.660 based on historical trends, opinion polls, and demographic data. The world of political
00:13:46.140 projections can seem like a nerdier version of witchcraft, depending on what you're looking at
00:13:52.120 and how you interpret it. But luckily, we've got a pro here to help us make sense of it all.
00:13:57.020 Philippe J. Fournier is the creator of 338 Canada, co-host of the Numbers podcast,
00:14:02.940 and has written for Politico, The Walrus, and Lactualité. And he's a regular guest on CBC
00:14:08.360 news where he breaks down the polls. Philippe, welcome to Bubble Pop.
00:14:12.520 Good morning, Rachel. Good to see you.
00:14:14.600 It's great to have you on. I'm so excited. We get to nerd out today.
00:14:19.380 Let's do it.
00:14:20.700 Yeah. So, I mean, first things first, we are recording this on Wednesday morning and obviously
00:14:26.020 anything could change. Today's a really big day specifically because the tariffs are either going
00:14:31.080 to come out or not. You guys, by the time you're listening to this, will know what has happened.
00:14:34.300 but right now, we don't know. And so, yeah, anything could change in the polls, which is
00:14:40.520 always true. But I'm wondering what the polls are telling you right now about this federal election.
00:14:47.960 So since the campaign began, we are not at, what, day nine today or day 10. So the liberals
00:14:55.280 entered this campaign as the slight favorite. And 10 days later, they are the heavy favorite.
00:15:02.300 it. Now, as you said, Rachel, everything can change. We saw from the past two months how fast
00:15:08.560 numbers can change, how opinion, public opinion can turn on a dime. But now with every day that
00:15:14.360 passes, it seems like all the indicators and not just voting intentions are pointing towards
00:15:21.120 the liberals. And now they, the liberal team must wish we could go to the ballot boxes today to end
00:15:29.140 this as quickly as possible. But with 26 days to go right now, the Liberals are in front.
00:15:34.260 They're in front in key areas, Quebec, Ontario, Atlantic Canada, B.C. as well. And the Conservatives,
00:15:40.780 even though they have a good score historically, they're trading the Liberals where they need to
00:15:48.480 win more seats. And so that's why right now, Mr. Carney and the Liberals are the favorite.
00:15:53.620 Yeah, they must be. As much as I think all Canadians hate what's happening with Trump
00:15:57.700 right now it is very helpful it seems like for the liberals i was hearing a lot of chatter about how
00:16:02.500 the uh the trump talk and the response to trump as the ballot box question seems to favor the
00:16:07.380 liberals as opposed to when the ballot box question is about affordability polyeth does a
00:16:11.460 bit better on that um is that a fair assessment like as long as trump stays in the game that
00:16:15.700 seems to be something right now at least is benefiting the liberals it's a fair assessment
00:16:20.580 rachel but i wouldn't say it's a straight line from trump attacks canada it's better for the
00:16:26.100 liberals uh we when we saw the numbers change like really change in late january after monsieur
00:16:33.860 trudeau had announced his resignation we saw the liberals recover a little bit but most polls still
00:16:38.980 had the the conservatives far ahead when the big change the big momentum shift happened was in the
00:16:45.860 last week of january monsieur trump had just been inaugurated and then he became his daily attacks
00:16:51.620 and daily threats on in the economy sovereignty the first 51st state nonsense and it was the
00:16:59.460 reaction to donald trump that i believe changed the momentum uh trudeau and the liberals basically
00:17:06.500 said we're not gonna stand for this this is unfair we're gonna fight back and even though
00:17:11.540 the conservatives have changed their message messaging since their first reaction was well
00:17:17.140 this is trudeau's fault uh he made us weak and even though it's something that you could defend
00:17:23.540 you know that trudeau made us weak it's it's an argument that could be made it really was
00:17:28.820 not in sync with the mood of canadians and from that point on we saw the numbers change
00:17:34.580 and obviously now we have mr carney with great personal numbers that is riding
00:17:39.220 a wave riding a honeymoon and this is where we are right now so it's not trump attacks canada
00:17:46.020 better for the Liberals, it's how you react to those Trump threats.
00:17:49.700 Which is probably why all those conservative insiders are going public, begging Pierre
00:17:55.140 Poliev to pivot. He seemed a little bit this morning in his speech to try to focus a bit more
00:17:59.860 on Trump. But he still is critiquing the current economic state, at least of Canada, in a bid to
00:18:07.300 tie that to the Liberals. But I think that a lot of people just hear Canada is the worst.
00:18:12.260 It sounds like a critique of Canada, even though he intends it as a critique of the Liberals.
00:18:15.780 So I'm curious to see if he pivots that. But I did want to ask you, I mean, you've obviously been doing this for a little while now, watching all these numbers.
00:18:25.780 And this feels almost like a silly question because I feel like we all know the answer.
00:18:30.260 But are you seeing anything weird or unusual in the numbers this election?
00:18:36.220 Like what has stood out to you as the most interesting story they tell so far as someone who watches these professionally?
00:18:42.200 Well, my friend and podcast partner, Eric Grenier, mentioned on our show that the conservatives, if they stand right now around 38% nationally, the last time they had 38% in a general election and lost, we have to go back a century.
00:19:01.920 Like usually 38-39% for the Conservatives is a winning formula.
00:19:07.280 But the entire Conservative vote really, really coalesced behind Mr. Carney and the Liberals to a point that we haven't seen in past elections.
00:19:19.880 The NDP, you're asking me which numbers are the weirder.
00:19:23.540 It's weird to see the Liberals back in front after seeing them trailed by 15, then 20, then 25 points in the polls.
00:19:30.740 But it's seeing the NDP in single digits, and most of that support went to the Liberals.
00:19:37.540 This morning, we had the new Leger poll, of course, Leger, one of the top pollsters in the country, has the NDP national number at 7%.
00:19:46.220 So that means that roughly two-thirds of NDP voters decided to flee, and most of them are going to the Liberals since 2021.
00:19:55.720 one. So I would say that's the weirdest part.
00:19:59.680 But again, a weird poll happens.
00:20:02.360 But when you get 12 polls in a row from, you know, professional posters
00:20:07.000 that know what they're doing and we all have the same bracket, the same range
00:20:11.360 for the NDP, you can put your hand in the sand or you can believe the data
00:20:15.520 and try to do something with it.
00:20:17.160 Totally. And, you know, I find that really interesting, too, about the NDP.
00:20:20.800 And it makes me think about how politics makes for strange
00:20:24.600 bedfellows I mean because as you said like the Liberal support seems to be coming in large part
00:20:30.560 from uh the NDP and even the Bloc Québécois a little bit if I'm not mistaken oh yeah oh yeah
00:20:35.940 right now um yeah and overall the Conservative numbers as you said are in a range that has seen
00:20:41.300 them win elections before so it's not necessarily that the Conservatives are doing something that
00:20:46.280 wrong it's that this whole progressive vote is is coalescing as you said behind the Liberals
00:20:51.060 I know we aren't going to see Paulyov out there campaigning for Singer Blanchet,
00:20:56.480 but is that in a weird way something that he needs to have happen?
00:21:02.480 Well, okay, I'll say this.
00:21:04.860 If we think a bit retroactively, if the conservatives do lose in the election,
00:21:11.860 and I'm not making a prediction, it's way too early.
00:21:14.780 But perhaps in hindsight, the conservative permanent campaign
00:21:19.700 that they have started since Poiliev became leader,
00:21:23.980 rolling ads all the time,
00:21:26.520 not most ads, but many ads in the fall
00:21:29.580 were attacking Jack Mead Singh and the NDP.
00:21:32.700 And now maybe they're starting to regret it
00:21:34.780 because they tore down Jack Mead Singh and the NDP so much
00:21:37.760 that instead of being at 16 to 20%,
00:21:41.760 which has been the NDP's average over the past five years,
00:21:46.480 well, maybe 50 years even,
00:21:47.680 now we have them in single digits and voters who are left-leaning are ready to give that
00:21:55.500 boring banker a shot which is completely strange for me but this as for conservatives who say we
00:22:03.320 don't understand why those NDP voters are flocking to the liberals well some introspection after the
00:22:10.340 election could help because they attacked and, you know, fired their advertising attacks on all
00:22:17.440 sides. And it could be biting them in the behind right now. Yeah, I mean, I said this on my podcast
00:22:23.820 campaign cops with my friend Paris Marx, but I you have to wonder if it seemed like Polyev's intent
00:22:31.160 with tying the NDP to the liberal so strongly, the NDP liberal coalition was to paint the Trudeau
00:22:37.380 government as extremely left. But I think that it also had the effect of painting Jagmeet Singh's
00:22:42.420 NDP as not as left-wing as maybe some progressives would want, which in the analysis or calculus of
00:22:49.960 the election, maybe a lot of NDP voters have internalized that and say, well, if he's not
00:22:55.520 that progressive anyways, might as well go for the other guy. I don't know. It's a lot to think
00:22:59.400 about, but we don't know that, right? We can't read minds. It's part of the thinking that will
00:23:03.740 have to occur after the election how i would say however that before christmas let's say
00:23:10.540 it would have been very difficult to criticize the campaign of the conservatives on a purely
00:23:17.500 strategic point of view whatever they were doing was working they were leading by 25 points and
00:23:22.860 even though sometimes they pierre poiliev and his team did or said something that go like okay that's
00:23:28.220 bit strange i'm looking at the numbers like well clearly it's working uh but right now the the
00:23:33.740 ballot question the campaign completely changed in january and it seems that they were not ready
00:23:40.780 for this and uh monsieur carnegie right now as i said he's benefiting from uh for miss honeymoon
00:23:46.860 yeah and you know philippe i did want to ask you some questions about polls more broadly as well
00:23:53.340 um maybe some of this is me being a big nerd or getting a little too in the weeds so i'm sorry to
00:24:00.540 the listeners if you aren't quite as nerdy as i am but but one thing that i think about a lot and
00:24:05.340 uh forgive my ignorance on this as well is um about online polls because they can't be assigned
00:24:11.980 a margin of error right which my understanding of that is that that makes it effectively impossible
00:24:18.700 to know how confident you can be in it. But I wanted to ask you about that.
00:24:23.020 Like, should we dismiss online polls entirely? Like, how do we interpret them?
00:24:27.820 Okay, so we have to be... Okay, we're going to get in very nerdy here.
00:24:33.100 Yeah, sorry, everyone.
00:24:35.580 The term margin of error in statistics is something very precise.
00:24:41.420 It's an equation that you can use to calculate the uncertainty of a sample.
00:24:48.140 Now, uncertainty and margin of error is not the same thing.
00:24:52.980 If you have a poll whose respondents are not picked randomly,
00:24:58.740 like in an online poll, then the margin of error is not applicable.
00:25:04.380 It's not that the poll doesn't have uncertainty.
00:25:06.860 It's that the classic definition of margin of error does not apply.
00:25:11.780 Now, OK, so I hope I didn't lose half your audience.
00:25:16.260 No, it's fine. You did that very well. It's very succinct.
00:25:19.880 But online polls, they are pulled from a panel.
00:25:23.580 So for the example of Leger, Leger has a vast panel of Canadians.
00:25:28.060 They send out emails and they try to construct a sample that is representative of the population.
00:25:37.920 So it's not picked at random, even though there's some randomness.
00:25:41.340 The whole panel, of course, is polled every time.
00:25:47.740 And so they try to build a representative sample of the population using the data from the census to have enough young people, old people, people in every region, people of both genders.
00:25:58.720 And so they build a sample.
00:26:00.920 And so it's not that it doesn't have uncertainty, it's that we don't really know what it is.
00:26:05.420 we have a guideline by using still the old classic formula that says if it was a random sample it
00:26:13.820 would have an uncertainty of plus or minus 319 times out of 20 or something and it gives you a
00:26:19.280 guideline but if people are confused by that and i understand why uh then you look at the track
00:26:27.380 record if the sample example again from leger if his online panel that i think it's 400 000 canadians
00:26:35.360 If that panel is not representative, it will falter when we compare the final polls with the election results.
00:26:43.880 And in the past decade, federally, these online polls have been very good.
00:26:49.400 And so it's not that we don't doubt them.
00:26:51.520 I always doubt.
00:26:52.380 I'm a scientist.
00:26:53.320 It's part of what we do to doubt.
00:26:56.300 But when you see that it works, and not once, but repeatedly, you go like, okay, they know what they're doing.
00:27:02.220 And so it goes from the big online polls, the Ipsos, Abacus Data, and others.
00:27:08.280 Everybody can be right once.
00:27:09.940 But if you're right or close enough repeatedly, you have to give them a little credit.
00:27:14.740 That is incredibly helpful, actually, because I think that a lot of the things that people think about when they see polls is how to interpret them and how to engage with the information they are seeing, right?
00:27:26.960 and that that's why i also wanted to ask you i mean what are the biggest mistakes that you see
00:27:31.520 people make even when they're interpreting your interpretation of the polls like what are the
00:27:35.600 biggest assumptions that people get wrong when they are talking about polls in an election campaign
00:27:40.720 oh well one would be very personal uh i know i know a lot of people go to my website take a
00:27:47.360 screenshot and then publish it online and say see that poll from 338 and i have to remind them very
00:27:52.800 kindly, I have little boxes that explain this is not a poll. It's a projection based on polls,
00:28:00.640 mostly. So there's modelization in there using demographic data and historical patterns.
00:28:07.680 So it's not a poll, but it gives you a rough idea. All natural sciences like physics, biology,
00:28:15.600 chemistry all of them all they do is give increasingly accurate approximations of everything
00:28:23.360 so the mass of the electron we know to the 10th digits well what is the 12 digits we don't know
00:28:28.720 but we want to find out i'm a physicist so i'm taking this example so everything we do in in
00:28:34.320 the natural science is to have an approximation of the real world and then we refine it to have it
00:28:39.600 it so it can get better every generation now this is not as much a natural science because you're
00:28:46.240 dealing with something far more complicated than electrons or galaxies you're dealing with human
00:28:51.120 beings and human beings are complex sometimes they lie sometimes they confuse sometimes they
00:28:56.640 change their minds they're moody they're in a bad mood they're happy and so pollsters have to deal
00:29:01.920 with that so there's a bit of art add to the science but a pole is an approximation of the
00:29:10.000 landscape and so that's why if you take all these poles they're all little pieces of the big puzzle
00:29:15.920 that you try to assemble and for the people who want perfection you can't have perfection you
00:29:21.840 won't get it if you want the projections to be good enough that's what it is and so i hope i'm
00:29:29.120 I'm guiding better information to the landscape, to voters, even though some people will have very selective memories.
00:29:39.540 And if I may give you an example, Rachel, in the Alberta election, I remember in 2019, Jason Kenney, my projection was almost perfect.
00:29:49.780 It's good work and also got a little lucky.
00:29:52.380 That's how it works.
00:29:53.440 So I called the winner in 82 of 87 writings.
00:29:57.700 and many of them were right on so I was very happy about that but I cannot tell
00:30:03.920 you how many emails I got from people living in those five writings that I
00:30:08.140 missed saying you don't know anything you're a fraud you missed my writing
00:30:12.580 you said this party would win in theater and I keep telling you have to look at
00:30:16.180 the big picture if I'm right 94% of the time yeah okay so we don't report planes
00:30:22.600 that land correctly we just report on planes crashing so it's selective
00:30:26.700 memory. And it's OK, we have to live with that.
00:30:29.220 Yeah. And, you know, I think that leads well into the other thing I wanted to ask
00:30:33.580 you, which is what do you find to be the most useful takeaway from polls?
00:30:37.940 Where do you think that people seeing these conversations can get the most out of
00:30:43.380 them? Hmm.
00:30:46.540 Well, that's a that's a tough one.
00:30:50.780 I guess it kind of depends on the person, right?
00:30:52.900 It depends on the person.
00:30:55.220 Well, obviously, I focus on politics and I've been told by some observers that I
00:31:02.740 I deal too much with voting intentions, to which I reply.
00:31:07.860 If you go to a baker, you won't say, why is there so much bread around?
00:31:11.380 Yeah. And so this is what I do.
00:31:14.180 But voting intentions is only one indicator.
00:31:17.860 There's more.
00:31:19.380 Pollsters regularly ask, do you think the country is going into the right direction?
00:31:24.760 Do you approve the government?
00:31:26.140 Do you approve of this party leader or that party leader?
00:31:29.020 And so in these additions of numbers, we can find clues.
00:31:33.320 In the 2021 election, federal election, if memory serves, it started at mid-August.
00:31:41.240 And so the reason Trudeau and the Liberals went into an early election
00:31:45.720 is because the numbers were favoring them.
00:31:47.980 they were heading towards a majority. And in the first two weeks of the campaign, we saw not the
00:31:54.400 conservatives, but Erin O'Toole, the conservative leader back then, his personal numbers improved.
00:32:01.400 It's like the country was softening up to Erin O'Toole. And then days later, we saw the
00:32:06.560 conservatives surge back into a tie with the liberals. So the favorables of the leader was
00:32:12.760 an early indicator of a possible conservative comeback. And it happened. So this is what we're
00:32:18.540 watching. Voting intentions, of course, is top of mind. It's not the only indicator. And we try to
00:32:23.740 get clues from the mood of voters. It's not always easy. It's not always perfect, but it's good
00:32:30.040 enough. So basically, if you want to be a really big politics nerd, there's a lot more you can look
00:32:35.320 into than just voting intention. Absolutely. Oh, absolutely. You know, one thing that I always
00:32:41.420 wonder about as well is um and i think that it obviously is case by case but do you find that
00:32:47.200 you know it feels like there's a new scandal every few days in elections and stuff
00:32:52.080 do they really generally impact polls or is it just journalists looking for a story of the day
00:33:00.080 i i guess like does it usually just come down to a basic ballot box question or do you see a lot
00:33:05.140 of fluctuation based on scandals historically?
00:33:08.500 I'm glad you asked this question, Rachel.
00:33:11.060 I think we should not confuse a scandal
00:33:14.820 with the outrage industrial complex
00:33:18.460 that we see online every day.
00:33:20.060 I agree.
00:33:21.580 So so this is, I think, what my readers
00:33:26.180 and to a lesser extent, my students, I'm still
00:33:29.740 I'm still teaching in college physics in Montreal.
00:33:33.300 And, you know, many of my students are aware of what I'm doing.
00:33:35.940 And so I say to them, we follow the scientific method.
00:33:39.500 We are trying to be rigorous and we don't put our personal inputs in there.
00:33:43.980 So the daily media rounds of, as you said, scandals,
00:33:50.020 very little of it resonates with voters unless it becomes a pattern.
00:33:56.180 You may have seen or heard Mr.
00:33:58.540 Carney, you know, he misplaced the name of his candidate in Montreal, the survivor of the
00:34:05.300 Polytechnic shooting. Right. And so it was it was a mistake and it could have resonated more.
00:34:12.140 But he apologized and it was quickly forgotten because it didn't become a pattern. If he were
00:34:17.100 to do that every day, the comparison with old Joe Biden would come very fast. But so I've seen
00:34:24.420 people call this a scandal and it says well well so far the liberals are leading by 20 percent in
00:34:31.720 quebec so i guess they don't care so we have to listen to voters more than opinion makers i would
00:34:39.000 say and this is what polls do this is what scientific professional polling does is it goes
00:34:45.640 beyond the the headlines and it asks voters what they care about and right now the ballot box
00:34:53.040 question yes affordability is still high but dealing with the americans is just as high and
00:34:58.640 now you cannot divorce the cost of living with dealing with the americans because yeah if the
00:35:04.720 economy is hurt in canada if the auto industry suffers in ontario i live in quebec the aluminum
00:35:10.640 industry is big here if those industries go go down because of the tariffs and because of
00:35:16.640 tough relations with the us yeah cost of living is going to go high and it's going to be a problem
00:35:21.680 So, yeah, the polls give you the score of the game and give you a better sense of what's really happening than anything else you can see.
00:35:32.420 Definitely. And, you know, that always makes me wonder, too.
00:35:36.140 And I think that's one thing that's so interesting about what you do with 3-3 Canada is polls.
00:35:40.680 There are things that polls specifically on their own can't account for, right, that could influence the final outcome or at least struggle to maybe take into account.
00:35:51.680 Like you said, like you said, humans are complicated, more complicated than galaxies.
00:35:55.560 So I guess I did want to ask you, like,
00:35:57.720 what are the things that polls struggle to account for that can really change
00:36:03.360 that final electoral outcome?
00:36:08.200 Hmm.
00:36:09.760 I'm not sure what to say here.
00:36:12.360 So what do polls struggle?
00:36:13.880 I mean, there's one thing.
00:36:15.200 I mean, voter enthusiasm matters in the election.
00:36:19.200 And I think polls don't always capture it,
00:36:23.440 but also voter enthusiasm can be an illusion.
00:36:27.960 So the fact that we see, for instance,
00:36:31.040 the rallies that Pierre Poiliev is holding, these are big crowds.
00:36:34.520 It's not nothing.
00:36:35.600 It shows you how committed thousands of conservative voters are in these areas.
00:36:41.720 And so I am not discounting this properly at first hand.
00:36:45.800 Right. It matters.
00:36:47.240 It's just that then you look at the polling and you see, OK,
00:36:50.840 Pierre Poiliev draws big crowds, but it's those who go into a rainy night,
00:36:56.560 into a hangar to cheer their favorite leader.
00:37:00.320 This is not the median voter.
00:37:02.200 This is the very committed voters.
00:37:04.160 And we know from past elections and from polls that in general,
00:37:09.040 conservative voters are far more committed and faithful to their party than for the
00:37:13.960 NDP and the liberals and even the bloc.
00:37:15.840 So, again, I don't discount this.
00:37:19.720 It is impressive what Pea Poyev is doing in those rallies.
00:37:22.720 It's just that I'm not going to look at the polls, look at the rally and say,
00:37:26.120 well, clearly the polls are wrong.
00:37:29.480 The polls are telling the story of the average or median voter
00:37:33.920 and those rallies don't.
00:37:35.040 So we'll see.
00:37:36.840 But I receive a lot of criticism online about, you know, look at these rallies.
00:37:43.120 It's like, yeah, OK, great.
00:37:45.320 But that's not what I do. I don't compile rallies. I can compile the polls.
00:37:50.540 Yeah. And you also weave in the historic and demographic data and stuff, too, right?
00:37:56.000 Which is interesting.
00:37:58.120 Yeah. I mean, it's a fine-tuning thing. It doesn't change a whole lot.
00:38:03.880 The polls still drive most of the movement we see in the projections.
00:38:09.220 But there are case-to-case bases.
00:38:11.520 So sometimes a party, and I give the example on my website,
00:38:15.920 Quebec Soledad, the left-wing party in Quebec,
00:38:18.520 in the election in 2014, they had 7% of the vote.
00:38:23.420 And in one writing that they won, they had 50%.
00:38:25.960 Next election, they went to 15% of the vote.
00:38:30.100 If you go by a swing model, that means the Quebec Soledad
00:38:32.940 was going to get 100% of the vote in that writing.
00:38:35.380 Obviously, that was not going to happen.
00:38:37.060 And so it was helpful to look at the demographic of that writing
00:38:41.480 to see who is the QS voter, the age bracket, the gender and everything.
00:38:46.440 And so this helps.
00:38:47.440 But it is really fine tuning.
00:38:49.880 So I try not to to give too much weight to this fine tuning
00:38:57.440 because the polls usually tell the story.
00:38:59.440 And, you know, for those who don't believe me, I'm not always right.
00:39:03.040 But the record of my website is pretty good.
00:39:05.120 It's, you know, nine times out of ten, I have the right winner for writing.
00:39:09.040 So it's good. It's not perfect, but it's good.
00:39:12.060 The recipe works, you know, and maybe there's a bit of a flaw in the premise of my question there, right?
00:39:17.560 Maybe a properly worded poll can maybe account for most things that could impact a final outcome.
00:39:25.300 But it's also the fact that polls, we have provincial polls and national polls and very rarely we have local polls.
00:39:33.740 And I try to interpret this provincial data into what's happening at the writing level.
00:39:39.840 That's why when you look at a specific writing on 338, you see the uncertainty plus or minus 8% or plus or minus 7%.
00:39:48.100 It's a bell curve. It's a huge bell curve.
00:39:50.680 And it's not a bug. It's a feature.
00:39:54.440 If the results are uncertain, it would be taking voters for idiots to say, 1.00
00:40:02.020 Oh, no, I know, I know that writing in Edlington. 0.99
00:40:04.420 Yeah, the liberals are going to get 38.
00:40:05.780 No, I say they're going to get between, I don't know.
00:40:08.180 I'm making up numbers here, but 32 and 42, let's say.
00:40:11.100 Yeah, you have to give them a range of outcome.
00:40:13.580 And I know this is extremely nerdy, but this this is how, you know,
00:40:19.060 that's how we deal with with statistics and science.
00:40:21.980 It's there's much uncertainty in there.
00:40:25.220 A hundred percent, which is why I'm looking at the last question I had for you.
00:40:28.420 And I'm like, he's not going to answer this because he's not going to.
00:40:31.420 yeah try me try me well I wanted to ask um is there anything you feel comfortable predicting
00:40:39.660 and that's the word I feel like pollsters maybe hate is uh because right now you you deal with
00:40:44.960 the here and now based on the numbers right and that can you don't know what tomorrow is going
00:40:48.720 to bring but I am curious about whether there's anything that based on what you've seen historically
00:40:53.060 you feel even a little bit comfortable predicting about what might happen
00:40:57.300 I never want to give Philippe Grace if he gets this wrong, because I'm asking him to do something they don't normally do.
00:41:03.300 I'm going to tell you this, Rachel, the answer is I'm not comfortable at all.
00:41:09.300 Yeah, that makes sense. That's actually a good endorsement of your work.
00:41:12.300 Because if you look at the numbers, let's say in Quebec, okay, this morning, so the Liberals are in the low 40s, Bloc Québécois around 22 or 23%.
00:41:24.300 This is very close to the results in Quebec in the 2011 federal election.
00:41:32.540 Gilles Duceppe and the Bloc had 23% and won four seats.
00:41:37.340 They were wiped off the map by Jack Layton.
00:41:40.000 So we're seeing similar numbers right now, but substitute Layton for Carney and the Liberals.
00:41:45.980 I would not be comfortable saying, oh yeah, the Bloc is going to win a handful of seats.
00:41:50.120 But it's within the realm of possibilities.
00:41:53.020 So right now, I'm just going to tell you if unless there's a big outside event or a major stumble during the debates or during the campaign, all the indicators point to a liberal victory.
00:42:08.340 But this is not a prediction. I'm sorry. I'm very sorry.
00:42:11.900 That's OK. No, I knew as soon as I was looking at my wording of the question, I was like, he's not going to want to give a prediction.
00:42:17.780 That's not what these guys do. Well, I think that's basically all the questions I had for you.
00:42:22.700 But is there anything that you think people should know?
00:42:24.940 And I will also ask you to pump your socials in a second.
00:42:28.300 But just before we get to that,
00:42:30.260 is there anything that you think people need to know about polls
00:42:33.060 or about this election that I didn't ask you?
00:42:36.580 Don't forget that polls that do not say the same things,
00:42:41.780 polls that show variance or divergence during a campaign.
00:42:46.100 That's entirely normal.
00:42:48.260 It's normal to have a spectrum of results.
00:42:51.060 That's why I'm here.
00:42:52.100 And that's why I make a weighted average of it.
00:42:54.420 It's not that some polls are wrong and others are right.
00:42:57.920 It's that we throw darts at a moving target.
00:43:02.260 And so don't give pollsters a hard time because they're different.
00:43:06.620 If they all said the same thing, I would be far more worried
00:43:10.700 about the state of polling in the horse race right now.
00:43:15.180 So give pollsters a break.
00:43:17.200 What they're doing is fantastic.
00:43:18.760 I repeat, I'm not a pollster.
00:43:20.800 I'm a poll analyst.
00:43:21.720 I need these guys to do well. I cheer for them to do well, regardless of which party they put ahead.
00:43:29.720 It's not that hard for me to not be partisan. I try to, again, be analytical and go with the
00:43:35.480 numbers and it's worked so far. So that's what I'll say. So be friendly to your local neighborhood
00:43:40.920 pollster and your local neighborhood poll analyst. Philippe, thank you so much. Where can people keep
00:43:46.760 up with you and find your work uh on blue sky you can find me at 338 canada uh on um facebook
00:43:55.880 banned my page because apparently i'm news so you cannot access my facebook page yeah well i guess
00:44:02.600 uh you can go to my newsletter 338 canada.ca and the main website is 338 canada.com
00:44:10.360 and so uh go there and uh take screenshots and criticize me but do it politely please
00:44:16.760 Be nice to each other, okay?
00:44:19.540 All right, Philippe, thank you so much for your time.
00:44:21.580 This was a really great conversation
00:44:22.740 and I appreciate you hopping on Bubble Pop.
00:44:25.860 Thank you, Rachel, for having me.
00:44:27.340 Much appreciated.
00:44:30.480 Thank you for listening to this week's episode of Bubble Pop.
00:44:33.660 I'm Rachel Gilmore,
00:44:34.620 your least favorite person's least favorite journalist.
00:44:37.160 If you enjoyed this episode,
00:44:38.300 consider subscribing to Bubble Pop
00:44:39.740 on YouTube, Patreon, Substack,
00:44:41.580 and wherever you like to listen.
00:44:43.320 You can also find me on basically any socials.
00:44:45.880 I'm on Blue Sky, I'm on TikTok, I'm on Instagram.
00:44:48.960 Look around, I'm probably there.
00:44:51.160 I'd like to say a special thank you to Hans-Vivien Wenzel,
00:44:53.660 who is the man behind the incredibly groovy music for Bubble Pop.
00:44:57.180 You can find more of his stuff on Spotify under the artist name Hans,
00:45:00.880 lowercase h, with an exclamation mark.
00:45:03.920 Thanks again for hanging out with me.
00:45:06.680 This week, more than ever, I want to thank you,
00:45:09.420 because doing this is a dream.
00:45:12.240 And not only that, but I can defend myself and my work.
00:45:15.880 because of you. I couldn't do any of this without you so thank you and I'll see you next time.