The National Telegraph - Wyatt Claypool - August 26, 2024


Crazy bad polls in Canadian politics (how to spot them)

Topics
bad polls
Hosted by
Wyatt Claypool
Mentioned
Frank Graves Justin Trudeau Sam Harris Blaine Higgs Wyatt
New polls show BC Conservatives beating David Eby's NDP! David Eby's NDP campaign is failing - BC Conservatives set up to win!

Episode Stats


Length

5 minutes

Words per minute

189.93

Word count

998

Sentence count

57

Harmful content

Toxicity

1

sentences flagged


Summary

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

In this video, I talk about bad polls and how to spot them, and why you should be worried about them. I also talk about how to identify a bad poll, and what to do if you see them multiple times in a single day.

Transcript

Transcript generated with Whisper (turbo).
Toxicity classifications generated with s-nlp/roberta_toxicity_classifier .
Topics generated with Qwen2.5-3B-Instruct.
Hosts, guests, and mentioned names generated with spaCy (en_core_web_sm), reconciled against Wikidata.
00:00:00.720 Hey guys, it's Wyatt here with another video about polling, but unlike my last video where
00:00:06.640 I went over the political dynamics currently taking place in British Columbia, I just want
00:00:11.680 to talk about bad polls. And what does a bad poll look like? Because I see people on the left and
00:00:18.400 the right, but mostly the left, sharing bad polls and pretending that they are noticing or that
00:00:24.560 they are on to some real political trend going on. This has happened multiple times in British
00:00:29.360 Columbia. It's been happening in Saskatchewan. It's happening federally. It's happening in New
00:00:34.320 Brunswick. You will see a polling trend. Like, let's just take a second here and just say,
00:00:39.040 how do you identify a bad poll? I'm not talking about like a fake poll. You know,
00:00:42.800 fake polls happen where people put out some numbers that are obviously not backed up by a
00:00:47.680 real sample. It's a cooked sample to try and get a result. I'm not even talking about those. I'm
00:00:52.080 talking about like a poll where even if they were genuinely trying to make something accurate,
00:00:57.280 obviously they should have just pulled the plug and started over again because whatever they came
00:01:01.600 up with was not good. So in British Columbia, we just saw two polls come out that put the
00:01:06.320 Conservatives ahead by one point and three points. Now, if you released a poll saying the BC Conservatives
00:01:12.000 and NDP were tied, I'd believe that. If you released a poll saying that the NDP was ahead by one or two
00:01:18.000 points or three points even, I would believe that too. But there's a poll that just came out from
00:01:23.200 Council that literally had the BC NDP for some reason ahead by eight points. Yes, eight points.
00:01:34.240 As the trend has been going the other way, as the Conservatives keep gaining, or at least the NDP's
00:01:40.400 lead keeps sort of shrinking and becoming more sort of, I guess, tightened around the Vancouver and
00:01:47.760 Victoria areas that suddenly they're ahead by eight points. That's not a real poll. If you think that
00:01:54.240 this is a plus eight election in British Columbia, I don't feel like you are following the political
00:02:01.040 dynamics of that election very closely. Same thing in Saskatchewan. If you think that's a race where the
00:02:08.400 NDP is ever going to go ahead by like seven points, I don't feel like you're following what's going on very
00:02:14.800 well. Same thing with New Brunswick. If you think that the New Brunswick PC party, even if Blaine Higgs
00:02:22.160 is not popular, they're not going to be down by seven or eight points, which you see every once in
00:02:26.400 a while in a poll. Federally speaking, we saw ECOS from Frank Graves come out the other day. I believe
00:02:32.000 it's Frank Graves that runs ECOS. Frank Graves comes out with a poll that's showing that the Liberals are
00:02:37.600 catching a bit of momentum and maybe Justin Trudeau will be able to hold on. Because from his last poll,
00:02:42.960 he showed that the Liberals were behind the Conservatives by 20 points. And in this most
00:02:47.760 recent poll, they're only down by 14 points. No. Yes, 14 points is still a big lead. But no,
00:02:55.680 every other pollster is showing them around 20 points ahead, the Conservatives ahead of the Liberals,
00:03:00.720 18 points, 19 points, 21 points. If you come up with a 14 and there's no significant event that's
00:03:07.920 happened that would actually justify it going up or down like that, then you have to look at your
00:03:14.080 numbers and how you're getting them. It really is that if the poll doesn't match the mood, there's
00:03:21.280 something probably wrong. I don't want you to just then take this and say, oh, if I think that the
00:03:26.080 Conservatives in BC should be ahead by 10 points, well, any poll that doesn't say that is wrong. 0.80
00:03:30.800 Obviously, that's silly. It is a tight race. And generally speaking, that means the polls should be
00:03:35.680 tight. Maybe in two months, something big happens. And that justifies the NDP being ahead by eight,
00:03:40.880 or the Conservatives being ahead by eight, maybe. But out of nowhere, you don't just get the Liberals
00:03:46.640 federally gaining eight points in a poll or gaining six points in a poll, or you don't have the BC NDP
00:03:53.360 having like a 10 point swing back towards them in British Columbia. If you know enough on the ground
00:04:00.480 and you're kind of feeling things out, you can usually tell what should be realistic. I can
00:04:05.040 basically predict numbers ahead of time on what I think the mood of the province is. It's not because
00:04:09.280 I'm some marvel at politics. It's because I'm just, I just talk to people, I do door knocking,
00:04:15.040 I look online, and don't look fully online because online stuff can be super hyperbolic, but I know what
00:04:20.800 generally looks like a good realistic poll. Anyways, but that should be it for me today. As you can
00:04:28.960 see, I'm still in Japan walking around the streets because I couldn't find a good place to sit down
00:04:33.440 because I didn't want to annoy people and like, sit in front of their house or their apartment
00:04:37.040 building and then have people going by. And even here, it's kind of difficult in Osaka since it's a
00:04:42.080 pretty busy city and we're in the downtown core right now. But I'll see you guys hopefully in my home
00:04:47.360 studio within the next week or so, or at least finding better places to film while I'm out here.
00:04:53.040 But I just want to talk about that whole issue of bad polls. In general, races do not swing by 10
00:04:58.960 points. Kamala Harris didn't go from being extremely unpopular to extremely popular. There is a sampling
00:05:05.360 bias, there is a response bias, and this takes place in every election. Doesn't mean the pollsters
00:05:10.000 are lying, it means they're not doing their work properly. Anyways, see you guys.