In this episode, I talk about how polls are not as accurate as they seem to suggest, and why we should be worried about them. I also talk about the results of recent polls, and how they don't necessarily match up with reality.
00:00:00.000Hey guys, Wyatt Claypool here. I am a big believer in polling when it comes to analyzing politics, but I'm at the same time somebody who does counsel caution when taking one pollster's numbers as being concretely accurate.
00:00:17.360In this election cycle, I have had to discount a lot of pollsters that I just no longer trust, at least for this election. It doesn't mean that they're trying to make up numbers, but some of them might be. At the very least, it means that their samples aren't under control, which means that they're picking up a lot of overly excited activist liberals, and they're not picking up enough average Canadians when they're asking them who they're voting for.
00:00:41.440Let's put it this way. When Pierre Polyev, at the opening of this federal election, could have a rally of 2,500 people, and then two days later, he has another 3,000 people in Hamilton as we speak, and he can basically recreate at least a rally of 1,500 people in any given riding in this country, I don't believe Angus Reid when he says that the liberals are currently at 45% in his sample, and the conservatives are at 37 or 38.
00:01:09.080That's just not an election that's ever happened in Canada, where the conservatives used to be ahead by 25 points, and now they're behind by 8.
00:01:19.360And in fact, I didn't believe Angus Reid when he said the liberals were only at 16% in December. At the time, I said, that looks really bad for the liberals.
00:01:28.180At the same time, I don't actually think they're all the way down at 16%, five points behind the NDP, because I think it's just a bunch of people who are emotionally fed up with Trudeau, and they're no longer taking the polls, which means the liberal sample has slipped hard.
00:01:42.600This is what Nick Nanos is showing right now in Nanos' poll. I think this is far more realistic, so I'm adding them to at least the category that we can tentatively trust to at least be attempting to get it accurate and is trying to do sampling well.
00:01:58.380In his poll, he actually shows Trump and U.S. relations as the top issue for most Canadians at 31.2% of the population rating it as the top issue.
00:02:09.080It is distantly followed by the economy and jobs at 17%.
00:02:13.660And even in this poll, which is great news for the conservatives, the conservatives are ahead by three points and in fact are actually starting to widen their lead again.
00:02:22.640This is exactly what I've been predicting is going to happen, and I think that we are going to see this reflected also in what innovative research finds, as well as abacus data.
00:02:33.000But one point I want to talk about before I move on to some other stats I want to show you is that pollsters do get things wrong.
00:02:40.340Whether it's because they are attempting to get it wrong, or it's just because their sampling sucks, it happens a lot.
00:02:47.520It doesn't mean that we don't trust polls, it just means that we have to look at the poll and say, is this at all realistic to real life?
00:02:54.360Don't just say, well, none of my friends are voting liberals, so why are they even above 25%?
00:02:58.740Just look around on social media, look around in real life.
00:03:01.940What are people talking about at the coffee shop?
00:03:03.760The liberals are definitely increasing their vote share these days, but I don't think they're exactly leading the conservatives by five points.
00:03:11.820I don't think they're in fact leading at all.
00:03:13.860I think there's just a lot of artificial excitement.
00:03:16.400And a little data point I want to use to back this up is the Saskatchewan provincial election.
00:03:22.660I'm in fact borrowing a point that Frank Vaughn made on his YouTube channel, because it was a very good point, that look at these polls done by Insightrix, Mainstreet, Liaison, ResearchCo, Forum Research.
00:04:23.940They may have been even attempting to be accurate.
00:04:26.140And I think that Saskatchewan tends to actually have very, like, it's a harder population to poll.
00:04:32.900I guarantee if they were postal code matching their polls, they would have found out that they were mostly polling people in Saskatoon and Regina.
00:04:40.840And that's why it looked that close, in fact, with the NDP leading.
00:04:44.220Because what farmer, and there's a lot of farmers in Saskatchewan, what farmer riding his combine is going to pick up a phone and take a 15-minute polling call with you if you're doing in-depth, or even pick it up for 30 seconds?
00:05:37.320That's why I like innovative research.
00:05:39.660That's why I like at least nanos for the time being.
00:05:44.000I'd like to see them perform over the long run.
00:05:46.700And again, if they show the conservatives ahead by 10 points tomorrow, I'm not going to believe that.
00:05:51.060That's not realistic to this election.
00:05:53.280It's a tight election that I believe the conservatives are leading by a bit.
00:05:57.720Now, let's just jump over as well to some other numbers here that actually demonstrate the weakness of the liberals, even in these polls that are showing unrealistic numbers for the liberals.
00:06:10.080I've just mentioned Angus Reid before.
00:06:11.980They showed the liberals all the way at 16% in December, and now they're at 45% now.
00:06:17.420In fact, somehow, like, greatly ahead of the conservatives.
00:06:24.240It demonstrates that what we probably are just seeing is a lot of urban default liberals being polled that are pushing up the liberals' numbers.
00:06:31.100But even when you ask those people, how firmly are you voting liberal, it's not actually that firm.
00:08:08.840Maybe so he wins in Southeast, but the fact that the guy did not actually resign as the mayor of Edmonton proves he's only doing this as a, basically, a stunt to help the liberals.
00:08:19.900He's not going to try too hard, but he has the Edmonton mayor with them, and that might help hold on to Edmonton Centre.
00:08:25.500But anyways, moving on to some other stuff, when you actually look at them, even in, like, Ontario, the Angus Reid poll shows the liberals punching above, like, 50% in Toronto.
00:08:40.440At the end of the day, though, the NDP are always, like, a 25% party in Edmonton, and the conservatives in that poll were at, like, 34%.
00:08:48.780So the conservatives and NDP and Greens and even the PPC are all somehow underperforming in Toronto except the liberals.
00:08:56.840What this looks like is that Angus Reid is just polling liberals on a sugar high, and everyone else is taking the polls at a more normal rate.
00:09:05.860But because there are tons of liberals in these samples, that's what he assumes is going to happen.
00:09:17.180Well, you've got to assume because people don't answer polls at the same rates that they actually vote in.
00:09:21.820So if you have a sample where 59% of the people who vote in your poll are women, are you going to assume that 59% of people who are going to show up in election day are women?
00:09:31.620No, you reasonably will weight the women down to where they usually are at when it comes to turnout.
00:09:38.100And women actually do vote more than men, but just not, you know, a 59 to a 41 ratio.
00:09:42.960It's more like 48 to 52 or 51 to 49, a little bit leaning more, elections tend to lean a little bit more female than male.
00:09:52.400So you do have to make those assumptions.
00:09:53.940But in all of these polls, when I look at their regionals, it's people like Angus Reid and others, assuming that the election is going to be like 34% or 40% people over the age of 60.
00:10:07.280No, people over the age of 60 do disproportionately vote more than people under the age of 60, especially younger people under the age of maybe 28.
00:10:19.780That's an overperformance over every other election where they're assuming that it's tons and tons and tons of seniors showing up to vote liberal who are, in fact, a default liberal category.
00:10:29.680So that's how you have Frank Grays and Abacus, not Abacus, Abacus has actually been very clear.
00:10:35.600They also show the Conservatives currently with a plus three.
00:10:38.220Innovative research even shows the liberals leading by one.
00:10:40.960But I think innovative research does good work.
00:10:43.400That last poll they put out, I think that their demographics were still a little bit too urban, but still, it was a good attempt to be accurate.
00:10:51.940But when you see ECOS, when you see Leger, when you see even Main Street coming out and they are just assuming that the population that's voting this election is extremely college educated and much older, not really, not seeing it.
00:11:07.380In fact, this is the last thing I want to bring up here, is that in this Nanos poll, the Conservatives are in fact winning every single age bracket except older voters.
00:11:16.620And this is how other pollsters have been trying to show the liberals ahead.
00:11:21.960Even this, it seems like a massive overpoll that people 60 years and older are going to vote 50% liberal, yet maybe in certain regions, I don't think overall.
00:11:32.060But in the rest of the age categories, you have 40% people age and 29 voting conservative, 35% between 30 and 39, 38% between 40 and 49, and 43% between 50 and 59.
00:11:44.620So this 60-plus poll seems a bit odd to me that suddenly, as soon as people go over the age of 59, everything starts changing.
00:11:53.720And this is even in the Nanos poll that's accurate.
00:11:56.280I think even in these more accurate polls, they have to fight back against what is a massive response bias and what is a massively default liberal, older voting demographic who tends to be answering a lot of these polls and skewing them.
00:12:10.640But some pollsters, again, because they like the skewing, because maybe it gives the liberals artificial momentum, or maybe they're just genuinely bad at their jobs, are letting it seem like the liberal surge is never ending.
00:12:21.540Every surge ends. Every surge ends. Every election. As soon as the election is called, things narrow.
00:12:28.000So when pollsters are showing the elections start up, and it's even going, surging more, the news cycle doesn't back that up.
00:12:35.120It exasperates me so much. When the news cycle is actually pretty pro-conservative at the moment, Carney's messed up on a lot of fronts, he's ducking debates, he's been exposed for a lot of corruption, and a lot of sort of like compromises that he has, or a lot of exposures he has with the CCP and whatnot.
00:12:51.920This has not been an amazing news cycle for Carney, yet some of these pollsters, you'd think that he saved a small child from drowning and pure poly was seen kicking a puppy. Not happening.
00:13:03.520But anyways, that's it for me talking about these polls today. I'm going to be coming back as we get more polling data, and again, I'm not just trying to ignore polls because they show the liberals ahead.
00:13:12.760I'm ignoring polls that don't seem realistic. If I see a poll where the conservatives are at 48%, the NDP's at 21%, and the liberals are at 34%, you can guarantee I'm ignoring that one.
00:13:26.540If I see a poll where the Blocs have acquired somehow only at 4%, I'm ignoring that one. Blocs are going to get 8%. At least, they could even get 9%, especially with Carney ducking French debates.
00:13:36.620If I can see hallmarks of things that are off. ECOS shows sometimes, like the PPC at 10% in Saskatchewan.
00:13:45.020I know Saskatchewan's very conservative. Not going PPC, especially in the year 2025 when that party has zero momentum.
00:13:52.820Anyways, so that's it for me, guys. Make sure to like this video, subscribe to the channel, leave a comment, do all that great stuff if you've liked my election coverage so far, and I'll see you guys next time.