The National Telegraph - Wyatt Claypool - August 13, 2025


Don’t believe BAD polls - Liberals lead is small


Episode Stats


Length

18 minutes

Words per minute

165.33154

Word count

3,027

Sentence count

189

Harmful content

Misogyny

2

sentences flagged

Hate speech

1

sentences flagged


Summary

Summaries generated with gmurro/bart-large-finetuned-filtered-spotify-podcast-summ .

In this video, I explain why you should trust the polls and why you shouldn't. I cover three pollsters, Abacus Data, Liaison, and Nanos, who have released polls this week, and explain why they are not to be trusted.

Transcript

Transcript generated with Whisper (turbo).
Misogyny classifications generated with MilaNLProc/bert-base-uncased-ear-misogyny .
Hate speech classifications generated with facebook/roberta-hate-speech-dynabench-r4-target .
00:00:00.000 Hey guys, Wyatt Claypool here. I have the whiteboard out again for a video and I seem to be doing this
00:00:07.480 whenever something really interests me or something has gotten under my skin. And this is a video where
00:00:14.060 something has gotten under my skin. People on social media are way too credulous when it comes
00:00:20.620 to national polls and I want to go over some general principles on whether or not you should
00:00:26.440 believe a poll. I'm not saying that only polls where you like the results are real. I am saying
00:00:32.380 take the polls where the results make sense with how public opinion shifts and based on previous
00:00:39.140 election results or how polls used to be in different situations that were worse or better
00:00:46.120 than the one we are currently in. And so I want to talk to you guys about three different pollsters,
00:00:52.220 two of which I think are generally trustworthy and one of which, at least right now, I would not
00:00:58.920 trust. It's kind of a sliding scale of one I really like, one that's pretty good, and one that you
00:01:04.800 should not trust right now. But before I get into it, I just want to remind you, if you like the show,
00:01:10.220 make sure to leave a like on this video, subscribe if you are not yet a subscriber, and leave a comment
00:01:15.880 on what you think about this whole topic and situation. Again, most pollsters are trying to
00:01:22.440 get it right. I've said this in other videos, there are very few pollsters where they're just
00:01:26.360 making up numbers or doing an intentionally bad job in order to make their political side look good.
00:01:32.460 A situation where that has happened is recently, I believe, the Alberta Federation of Labour commissioned
00:01:38.040 a pollster to do a poll, and it showed the NDP, surprisingly, is above the UCP by four points.
00:01:44.620 The problem is, at the same time that poll came out, you had some of the most accurate pollsters in
00:01:50.820 Canada, like Janet Brown doing polls of Alberta, which found that the UCP was leading the NDP from
00:01:56.760 anywhere from eight points to 14 points. So no, that was just a bad poll, and probably an intentionally
00:02:05.000 bad poll. The Alberta Federation of Labour wanted what it wanted, and that poll gave it to them. They
00:02:10.680 probably did a real poll, but they just did a poll based on some fantasy scenario where, you know,
00:02:16.540 60% of people who head out to the polls are public sector employees, which is not an actual normal
00:02:22.360 turnout. That's just me assuming what they did, but now let's get into the three pollsters I want to
00:02:28.720 talk about today. I'm just going to add their names to the board, and then I will go over what their
00:02:32.820 national results have been. Okay, these are the three pollsters I want to talk about who have all
00:02:39.440 released polls this week. Abacus Data, Liaison, and Nanos. So the two pollsters I do like on this chart
00:02:48.740 are Abacus Data and Liaison. I don't even quite agree with the polling result that Liaison came to,
00:02:54.620 and I know there's no agreeing or disagreeing with their result. I mean more so they probably
00:02:59.180 should have been, you know, they probably should have done a couple of samples over again, but it's
00:03:03.780 not unrealistic. We're going to get down to Nanos later, and the poll result is just unbelievable,
00:03:11.780 and then I'm going to explain why in greater detail in a second. But let's start off at the top here
00:03:17.380 with what Abacus had. I'll be going over Liberal, Conservative, and NDP. So Abacus had, like we showed
00:03:25.240 in another video, 43 Liberal. It had 40% Conservative, and then it had 8% NDP. I would actually say this one
00:03:39.280 may even have the Liberals a little too low compared to the Conservatives. You know, I could see the
00:03:44.880 Liberals being at 43 or 44, or the Conservatives being at 39 or 38. I would say a 3% lead for the
00:03:52.940 Liberals right now during their honeymoon maybe is a little bit of a small lead. Maybe it's more like
00:03:58.720 a 4 or 5% lead. But I still think it's a good pollster. Their numbers tend to move slow. They
00:04:04.920 don't have wild swings in what they find after certain events, because most people don't change
00:04:10.640 their votes that fast. It takes like a cataclysmic event for people to really change how they're
00:04:16.160 going to vote. But now let's move on to Liaison. Liaison here has the Liberals at 44%. And then 0.99
00:04:26.260 Liaison has the Conservatives at 35%. I'm going to get over to why I think they have the Conservatives a
00:04:34.200 little too low a bit later. But then we have the NDP at 11. Maybe a bit of a high poll for the NDP
00:04:42.640 considering they only scored about 6.6% in the federal election. That's almost double what they
00:04:49.000 had gotten. So it's a bit of an overpoll for the NDP, a little bit of an underpoll for the CPC. But
00:04:54.200 I think 44% is very realistic for the Liberals right now. But now let's get over to Nanos.
00:05:01.720 They have the Liberals at 45%, which is not exactly too high in my opinion for the Liberals. The problem
00:05:10.960 is that they have the Conservatives at 32% and then the NDP at 12. The NDP, who are famously broke right
00:05:23.340 now, are somehow double what they were at in the 2025 federal election. And the Liberals, if you can
00:05:31.500 count, are leading by 13%. They're leading here in this poll with Liaison by 9% and they're leading
00:05:41.440 with Abacus by 3%. Now, I think the truth is somewhere in the middle between Abacus and Liaison.
00:05:47.660 And if you're ever analyzing polls for yourself, I would say find a few pollsters that you think are
00:05:53.380 good. Kolsovsky is a good pollster. I think Innovation is good. Liaison is good. Abacus is
00:05:59.240 good. Main Street can be quite good as well. Find a few of them and then kind of track the differences.
00:06:05.540 Find one that's the high watermark. Find one that's the low watermark. And it's probably
00:06:09.900 somewhere in between the two. I did this frequently during the federal election when telling you guys
00:06:15.660 where I thought things were. Some people still say, well, why? You said Polyoff was going to win.
00:06:20.000 I literally never said that. I always said it's a coin toss on who's going to win the minority 0.97
00:06:24.240 government. And I would tell people to not believe pollsters that were showing Liberal leads of 5,
00:06:29.480 6, 7, 8, 9%. Because if the Liberals were leading by that much, they would have a big majority right
00:06:36.100 now. And they don't. They only won the popular vote by like 2.2%. And that still only resulted in them
00:06:43.380 getting a minority government. Because in fact, the Conservative Party's voter efficiency
00:06:47.860 has gone up substantially. Before, it was for every, let's say, 100,000 votes, the Conservatives
00:06:55.140 would win a certain amount of seats. Like for every 100,000 votes, the Conservatives received an
00:07:00.360 election, they would be winning about three or four seats. But if the Liberals got 100,000 votes,
00:07:05.840 they're winning like five seats or so. So the Liberals didn't have to get as many votes on average
00:07:10.900 for their wins, where the Conservatives had to get a lot more. It's actually reversed right now,
00:07:15.480 where the Conservatives actually don't need as many votes per win that they get, because they are
00:07:19.940 concentrating their votes in the areas that they need to. But anyways, now I want to move on to
00:07:26.000 specifically Liaison's Atlantic sample. And then I want to move on to some crazy news about Nanos.
00:07:32.760 Because when you hear the kind of seat results that this Nanos poll would result in, you will get
00:07:39.600 what I mean when I say that certain polls should be put on the shelf of shame and ignored. It's not a
00:07:44.880 good poll if it is telling you something that does not even feel like it is from the reality that we
00:07:50.320 live in. But just give me a second to clean up the board, and we'll talk a little bit about Liaison.
00:07:56.140 Okay, so we are back with Liaison, and I want to show you some of the regional numbers,
00:08:01.580 so I can explain why I think the Liaison polls plus nine for the Liberals is not quite realistic.
00:08:09.580 Again, Abacus's plus three is probably a little too low. This one is probably a little bit too high.
00:08:15.940 Now, they have right now, in Ontario, the Liberals leading the Conservatives 47% to the Conservatives 36%.
00:08:27.140 Now, I don't think this is probably too off of reality. It's a nine-point lead in Ontario. If the
00:08:33.380 Liberals are doing well, they are naturally going to have a lot of the vote that they're doing best in
00:08:38.600 be in the province of Ontario. British Columbia tends to be a little bit closer, and that's what
00:08:43.600 Nano says. I could see the Conservatives a little bit higher than 36% here, and the Liberals may be
00:08:48.960 more like a 45%. Conservatives more like a 37%, 38%. But no doubt, in a honeymoon period, this isn't
00:08:55.320 too crazy to see on a regional sample, especially because regional samples are going to be smaller
00:09:00.740 naturally than the full national sample. Now, in BC, I think this result is very realistic,
00:09:07.060 very much a Conservative estimate on where the parties are at right now. We have 42% for
00:09:13.340 the Liberals, and we have 37% for the Conservatives. And I would even say that Ontario number, that
00:09:20.240 puts the Conservatives within punching distance of the Liberals. This would still make them
00:09:24.740 competitive. And then I threw Alberta on the list, because Alberta is usually a good ballpark 0.72
00:09:31.060 province to be using. This poll has the Conservatives leading in Alberta 59% to the Liberals' 27%.
00:09:44.060 Now, this is probably an over-poll of the Liberals. I believe they only end up getting like 23%,
00:09:49.300 and the same poll has the NDP only at 8% in Alberta, despite the fact that the NDP has a seat in Edmonton,
00:09:56.600 and the Liberals only have, I believe, two seats, one in Edmonton and one in Calgary. So this is
00:10:02.560 probably too much of the vote going towards the Liberals. And frankly, getting the Conservatives
00:10:07.520 under 60% in Alberta is quite difficult. Not too bad, though. But the Atlantic Canada polls is where
00:10:15.760 I start to basically be a little bit incredulous regarding the results. They put this collection
00:10:24.120 of provinces at 55% Liberal to 25% Conservative. Now, the problem with this is that if you were
00:10:33.840 watching the federal election and you were following the polls, especially in Atlantic Canada, you constantly
00:10:39.600 saw polls where it was saying 55-25, 60-20, or 60-22, or something like that. Ridiculously large leads
00:10:47.840 for the Liberals, where some were predicting the Liberals were going to win every single seat in Atlantic
00:10:53.320 Canada. And I believe the Liberals ended up basically at even. They didn't win any more seats. The seats
00:10:59.700 kind of shuffled around a bit. We ended up having blue found land, where most of Newfoundland became blue
00:11:06.600 in the middle. So the Conservatives actually gained in Newfoundland, but then they ended up losing seats
00:11:11.840 in Nova Scotia, and then they held on to what they had in New Brunswick. But again, with all those
00:11:17.780 people predicting the 60-20, you know, 65-20, 55-20 kind of results in Atlantic Canada that were going
00:11:26.020 to show a Liberal sweep, it didn't end up manifesting on Election Day. You will see on this chart, I'm just
00:11:31.500 showing on screen because it's easier, the Conservatives were able to get 40% New Brunswick,
00:11:36.020 35% in Nova Scotia. In PEI, they got 36%. Newfoundland, they got 39.7%. And for some reason,
00:11:43.580 had the Yukon highlighted there, but that doesn't matter. But in general, the Conservatives were in
00:11:48.560 the mid-30s, not mid-20s. And I don't think that has changed since Election Day. So when I see a plus
00:11:55.440 nine for liaison, mentally, when I see that their Atlantic result is like this, I naturally say,
00:12:02.500 probably more like, even if the Conservatives slip down a bit, put them at 31% overall. 31 to maybe
00:12:10.320 like, you know, a 52 seems more realistic. Again, you'll look at the Liberal numbers here.
00:12:16.680 In New Brunswick, Liberals, 53. Nova Scotia, 57. PEI, 57. Newfoundland, 54. Also in case, in part because
00:12:25.300 the Greens no longer are competitive in Atlantic Canada, which they used to have a seat in,
00:12:28.900 and the NDP are not competitive at all these days. So 52-31 feels like we actually are in a
00:12:36.240 realistic area. Maybe you could put the Liberals up to where they got last time at 57. But 25 is a
00:12:42.900 massive under-poll considering the results that we saw in 25, despite all the polls saying it was going
00:12:48.240 to be a Liberal wipeout. And the best Conservatives could do is just hope they don't lose everything
00:12:53.060 they had. But now, let's jump over to Nanos' regionals so that I can prove to you why certain
00:13:00.400 pollsters should not be trusted. And again, it's not because they're rigged, it's because I don't
00:13:04.940 think they're doing a good enough job to make sure that they're getting people from all over
00:13:08.940 the regions of the provinces that they're polling. If you poll Toronto, and that's your Ontario sample,
00:13:14.920 it's going to make all of Ontario look a lot like Toronto. And that's subtle foreshadowing
00:13:20.000 for what we're about to look at. Okay, now I have Nanos on the board, and I regret to inform you,
00:13:27.280 I've tricked you. I have regionals here right now for Ontario, BC, Alberta, and Atlantic Canada,
00:13:34.560 like with the liaison poll. But all this doesn't matter, because Nanos didn't release any regionals.
00:13:41.060 All we know is that the Liberals are leading by 13 points. That seems realistic.
00:13:51.380 And now I need to show you something I threw in my bookmarks here that shows right now the Liberals
00:13:59.260 are slated to win. Drumroll, please, while I actually get this on screen. According to the actual,
00:14:06.640 like the formula that Polling Canada uses to project the results of elections, they say that
00:14:13.320 on the latest Nanos poll, the Liberals would win 95% of seats in Ontario. So we don't actually know
00:14:21.780 what the Ontario regionals are for Nanos. But what we do know, based on the polling projections
00:14:28.160 or seats from Polling Canada, that they would win 95% of the seats in Ontario. Ontario seats would go 95% them.
00:14:43.120 That doesn't look, this, you can call me crazy, I don't consider it realistic for the Liberals to be
00:14:50.580 getting 95% of the seats in Ontario. Look, I need to show you what Sherea Teest has, because he was the
00:15:00.040 most accurate poll, seat projectionist in the last election. And he, when he projects this one, it's
00:15:06.920 absolutely ridiculous as well. I'll bring this up on screen for you guys. And it says,
00:15:12.820 according to Sherea Teest, who's actually only 17 years old, but again, he's just very good at
00:15:21.200 projecting elections. He has the Liberals winning, on his very good quality polling projection website,
00:15:28.700 224 seats compared to the Conservative 79. So we would be going 224 to just 79.
00:15:40.580 Now, do you remember how many seats the Conservatives won in the last federal election? Well, let's look it
00:15:50.480 up to see how much of a fall this would be. So, out of all of, out of the 20, in the 2025 election,
00:16:02.520 the Conservatives got, were able to get 41.31% of the vote compared to the Conservative Liberals,
00:16:08.120 43.7%, but they won 144 seats. The Liberals won 169. And apparently, since election day,
00:16:18.400 with the Liberals, with a very mediocre performance so far, would win, what is this, 31,
00:16:26.040 55 extra seats. So they would go up 55, and the Conservatives would almost not even win half of
00:16:35.200 what they had won before. I believe this is basically just a little bit over that. It's like,
00:16:40.640 I think, two seats over 77. If you doubled that, it'd be like 144. Yeah, I'm not great at math. But
00:16:46.640 they would lose this. No, this is not how polls work. 21, that would be like 65 seats or whatever
00:16:55.380 that they've lost. This is what I say when a poll comes out, and I don't believe it. This is what I
00:17:01.700 mean when I say that Nanos goes on the shelf of shame. In the last three months, a very mediocre
00:17:06.880 performance by Mark Carney and his Liberal government. Do you think this is how things
00:17:10.860 would go? I'm not saying the Liberals wouldn't be leading. They are probably somewhere between
00:17:15.140 Liaison and Abacus. They ain't leading by 13%, and they ain't taking 95% of the seats in Ontario.
00:17:21.800 What have they done that are going to make people rally around them harder after they failed to get a
00:17:26.100 trade deal done, backed off from the digital services tax, which was good for them to back
00:17:30.560 off on, but they look stupid trying to implement it in the middle of a trade negotiation. They're
00:17:34.920 backing off of actually being serious about building pipelines, it seems. They are not actually doing
00:17:39.440 anything to crack down on crime. They're on an extended vacation. They do not have a spring budget
00:17:43.920 put out, even though Parliament voted to get them to do that. They're putting one out in the fall.
00:17:48.260 They have a $92 billion plus deficit. They are not winning 95% of the seats in Ontario.
00:17:55.080 After a bunch of the guys in finance just found out that Carney doesn't actually have a plan.
00:18:00.400 But anyways, that should be it for this video guys. Hopefully you didn't mind this slight rant
00:18:06.320 I went on, but make sure if you did like it to like the video, subscribe to the channel, leave a comment, and
00:18:12.960 stay tuned for me to bring out the whiteboard and complain about something some other day. See you guys later.