Talk:Opinion polling for the 2025 Canadian federal election

Nothing to do about it unless Mainstreet issues a correction, but I'd be willing to bet errors were made in compiling the results of this poll. Specifically, looking at the crosstables, it looks suspicious to me that the NDP would be leading the CPC in the Prairies (SK + MB) 44% to 27% -- I suspect they got these mixed up. And even more suspicious is the LPC leading in Ontario with 39% to only 24% (and 3rd place!) for the CPC, right on the heels of Ford handily winning the provincial election and the LPO doing very poorly -- again, I suspect they got those mixed up. Note that I'm not making this comment from any sort of partisan slant; it's just that the overall numbers in that poll are so out of step with all the other current polls that I couldn't help but investigate, and found those particular regional numbers rather curious. Cheers, Undermedia (talk) 15:03, 29 June 2022 (UTC)

:That sounds like a reasonable theory. Is there any means to ask them to check it? - Ahunt (talk) 17:17, 29 June 2022 (UTC)

:Semi-related, but what do we do about their new poll. It seems to be a hypothetical. Do we just disregard it? Do we average it based on the 2 scenarios? Do we count both? MikkelJSmith (talk) 22:09, 22 July 2022 (UTC)

::That new poll seems rather apocryphal or at least a wild outlier compared to their last poll. Also worth noting that it is rife with obvious spelling mistakes, too: {{tq|Piette Poilievre as CPC Leader}}. Seems to suffer some odd QA issues. - Ahunt (talk) 22:16, 22 July 2022 (UTC)

:::Yeah, there's even a typo on the polling method according to Quito Maggi. It says online once on the document but it is an IVR poll. MikkelJSmith (talk) 22:53, 22 July 2022 (UTC)

::::It isn't giving me a lot of confidence. - Ahunt (talk) 23:09, 22 July 2022 (UTC)

::::{{u|Undermedia}}, I see you added it to the graph. Just in case you missed this discussion, should we also add the other versions? This poll had multiple polls in it depending on the leader. Or do we default to Bergen?

::::Also, I realized I typed 2 scenarios above when I meant 3. Oops.

::::MikkelJSmith (talk) 15:04, 25 July 2022 (UTC)

:::::Yeah, I think it's fair to put Mainstreet's July results with Bergen as leader, since she's the current leader and other pollsters are currently polling with Bergen listed as leader. Personally I don't put too much stock in any IVR polls anymore — too much wild volatility from one poll to the next, too many glaring outliers, too many badly botched election predictions. But this is an inclusive page for polling so we include 'em anyway. Cheers, Undermedia (talk) 16:07, 25 July 2022 (UTC)

:::::: I'm against including this poll in the general body, I think it would be inappropriate. It's a hypothetical and "I'd vote for a CPC lead by Candice Bergen" ≠ "I'm planning/leaning towards voting for the CPC in the next election". CASalt (talk) 01:54, 29 July 2022 (UTC)

:::::: Polling pages on US elections have a hypothetical section for hypothetical polls, thoughts on including on here {{user link|Undermedia}}? CASalt (talk) 02:00, 29 July 2022 (UTC)

:::::::I don't see how this is any different from all the other pollsters who are currently asking the ballot question with Bergen listed as the CPC leader (and have been for several months). And if you think about it, they're all hypothetical: everyone knows there's not going to be an election "today/tomorrow". Every time one of the major parties holds a leadership contest, polls like this come out, and that's never stopped us from including the results under the current/interim leader of the party. Cheers, Undermedia (talk) 11:52, 29 July 2022 (UTC)

Hi all. I was about to update the numbers in the latest EKOS poll to those in the now-available [https://www.ekospolitics.com/index.php/2022/09/conservatives-maintain-small-lead/ official release], but abandoned at the stage of updating the reference because there simply does not seem to be a consistent way of entering/formatting the reference data anymore. Looking through the list of polls in code form, it literally seems like no two are done the same way in terms of what information is included and/or what order it's entered in. It seems every time I go to enter a new poll there's something new or different about how the reference information for other recent polls has been entered. It's not even worth trying to list specific examples because it's altogether just a total mess, seemingly completely arbitrary. IMHO our 'old' system that served us well for many years whereby we'd simply enter a link directly into "Link" column and that was that was far better than this bewildering mishmash. Cheers, Undermedia (talk) 13:50, 26 September 2022 (UTC)

:I agree. How about changing it back? - Ahunt (talk) 14:15, 26 September 2022 (UTC)

::In theory that would be swell, but beyond it requiring a fair amount of work at this point to revert the whole page back to the old system, it seems someone recently went on a crusade to additionally retroactively convert the past several election polling pages to this new system, as well as at least some provincial election polling pages that I've noticed. So I'm not sure if we'd end up sparking some sort of editing war. Ultimately, I'm not opposed to the new system, but it definitely needs to be more consistent. Alternatively, we could perhaps use this discussion to settle on a sort of template for the standard pieces of information that should consistently be included in the reference for every poll added/listed. Cheers, Undermedia (talk) 18:20, 27 September 2022 (UTC)

:::Since you maintain the graph, would you like to propose a more consistent format? - Ahunt (talk) 18:24, 27 September 2022 (UTC)

::::Yeah, we could find a consistent format. Just like it's done for Canadian English, the date and the infobox pictures on the election pages. MikkelJSmith (talk) 23:05, 1 December 2022 (UTC)

:::::Yeah, sorry, I abandoned this discussion after no other regular editors aside from Ahunt initially showed any interest. I've since grown more indifferent about the issue, but would be happy to support a consistent format if anyone's up for suggesting one (I also admittedly wasn't terribly motivated to suggest one myself). Another problem is it's probably a safe bet that whatever consistent format gets proposed here will be overlooked/ignored by many editors adding polls to the table, and then we'll have to relentlessly spend time fixing them. Cheers, Undermedia (talk) 14:45, 2 December 2022 (UTC)

::::::Hey, @Undermedia, @MikkelJSmith2 and @Ahunt. If I am understanding correctly, by "reference system" you are discussing the change from an inline link (e.g., [1] or [Nanos]) to references using the ref template (e.g. template:cite web? If that is the case, I am the user that made the broad change across multiple provinces and elections to update from an inline link to the ref template, I did this maybe a year ago. My rationale for the change was looking through an old list and very few of the inline links actually worked. I tried to use the IABot (https://iabot.toolforge.org/) to update the inline link to a archived version, but the functionality was not there. Linkrot (WP:LR) and data sourcing is a problem for reliability of these pages, so I made a changes across a number of elections/provinces. Unfortunately, there does not appear to be a way around the inconsistent look of references for these polls. Some are presented by the companies themselves (sometimes with or without an author), other times they are published in newspapers/websites, and other times they are released on twitter. In my mind, it was important to ensure that a future user is capable of easily capable of tracking down the source through a proper refence that can be indefinitely maintained by tools such as the internet archive and IABot. As part of this process, I updated the most recent provincial poll lists to present information generally in the same order (poll, dates, source, parties, MoE, Sample, Method, lead), as this was not always consistent across these pages. If you have any suggestions to improve this let me know. Caddyshack01 (talk) 17:37, 2 December 2022 (UTC)

:::::::Hi @Caddyshack01. To be clear, fundamentally I have nothing against the change from the old referencing system to the new system, and I thank you for putting in all the effort to implement that change across multiple election pages. My beef is simply with the wildly inconsistent way in which editors are now entering the reference details for each poll in the table, i.e. the information & code entered in between the "" at the beginning and the "" at the end. As just one among numerous examples, I just noticed some of the recent ballot polls (including the most recent Nanos one) start with "" instead of just "", and I don't really know why, and I feel that at this point there's practically no way for the multiple regular & occasional editors of this page to keep track of how the referencing information should be entered for any given poll. Cheers, Undermedia (talk) 18:05, 3 December 2022 (UTC)

::::::::Some articles use a hidden comment to let people know information like that, of course with varying success. - Ahunt (talk) 18:12, 3 December 2022 (UTC)

Hi @Lilactree201: Are you aware of a publicly accessible link to the 9 Dec 2022 Nanos poll that provides all the same details as the 'subscribers-only' tweet by Nik Nanos that you entered? If not, I'm inclined to switch back to [https://twitter.com/CanadianPolling/status/1602740673228193792 Polling Canada's tweet]. That would entail removing the decimals as well as the 4-weeks-prior poll from 11 Nov, but since it's a rolling poll, we can 'make up' for the removal by giving full weight back to the next one from 25 Nov (by removing the "1/2" on the sample size); as long as the Nanos polls aren't more than 4 weeks apart, we're technically not missing any information. Cheers, Undermedia (talk) 12:11, 21 December 2022 (UTC)

:While the whole Tweet is not shareable, the photo with all the info that Nanos included is : https://pbs.twimg.com/media/Fj96VXZWIAAmxWf?format=png&name=large Lilactree201 (talk) 00:05, 22 December 2022 (UTC)

::Interesting—should we link to that instead of the 'paywalled' tweet? Cheers, Undermedia (talk) 13:51, 22 December 2022 (UTC)

:::That would be fine with me.Lilactree201 (talk) 23:07, 22 December 2022 (UTC)

::::Just noting that the tweet is now publicly available.--Lilactree201 (talk) 23:15, 5 September 2023 (UTC)

add seat projections? 2001:56A:FD10:9100:D18C:4A3C:9D64:B29C (talk) 03:07, 16 August 2023 (UTC)

:WP:CRYSTAL - Ahunt (talk) 11:34, 16 August 2023 (UTC)

::No. People can use those other dedicated websites for that. Words in the Wind(talk) 13:57, 16 August 2023 (UTC)

:::Generally, we should not be including or even mentioning projections due to WP:CRYSTAL. Readers who want review projections can look elsewhere. Sometimes, there can be a subtle distinction between posting projections themselves, and noting otherwise notable predictions. In theory, if following the election there was a really significant mismatch between polls/projections and results, and that "polling error" or "unexpected outcome" and the projections themselves were covered extensively in RS, it could conceivably be appropriate to mention them in prose about the results/aftermath. That is the only circumstance, I can see where we might go there. That said, that is a pretty narrow exception. I think it could be okay to include polling averages though if we think the site is influential enough, and a RS. This would essentially just be an average of most of the polls we are already listing. I have seen American election articles include RCP and 538 polling averages. That said, I am not sure that really adds much value to the reader.--Darryl Kerrigan (talk) 21:14, 16 August 2023 (UTC)

Some of the Nanos polls cite this JasonHRoy account on twitter (or...sigh...X). He seems to just be a guy tweeting out the numbers. Not saying his numbers are incorrect, just that I don't see how he meets standard as a reliabe source.

https://twitter.com/JasonFromNS/status/1696494879948394803Lilactree201 (talk) 23:49, 1 September 2023 (UTC)

:It doesn't. We should source these numbers from the polling agencies themselves. They usually post reports including their numbers and methodology online.--Darryl Kerrigan (talk) 23:58, 1 September 2023 (UTC)

::With Nanos behind a paywall most weeks the options are to cite Polling Canada as a source (iffy but at least have a reputation) or delete most of the recent Nanos numbers from the table.--Lilactree201 (talk) 00:10, 2 September 2023 (UTC)

:::What productive service does Wikipedia do by reproducing all of these numbers in articles where they can become stale and/or edited by editors with a political agenda? I think these opinion polling articles should only be created post-hoc when the data is solidified. 2605:B100:110A:5E60:259A:D16F:54CF:4B24 (talk) 01:25, 18 September 2023 (UTC)

:::: Ignoring the above comment... could we please get some input on dealing with the problem of random Twitter accounts being used as sources? Nanos polls citing Twitter users (JasonHRoy, 2close2call before he was suspended, now Purrlitics) keep getting posted. IMO we either need to delete them or replace them with Polling Canada. Lilactree201 (talk) 03:47, 3 October 2023 (UTC)

:::::I agree that citing JasonHRoy is really getting into a grey area, but I've been letting it slide without bringing it up for a while now, to the point where I think that deciding to disallow it at this point would result in having to delete a good number of Nanos polls from the table? I also agree that Polling Canada is at least a bit more legit, and am wondering if Polling Canada tweeted out all the same Nanos polls that Jason did, i.e. as a potential replacement source for all the polls that have already been added? These days tweets don't even seem to be listed in chronological/most recent order on accounts' profile pages anymore (nor does there seem to be a sorting option), so I personally wouldn't even know how to begin trying to dig through Polling Canada's tweets for past Nanos polls, what a mess! Cheers, Undermedia (talk) 14:06, 3 October 2023 (UTC)

:::::JasonHRoy has more detailed numbers than PollingCanada and it's the only account I can find other than Polling Canada because Polling Canada's rounded up numbers would affect the trend like "20.5" rounded as "21" is a huge difference. Mason54432 (talk) 01:08, 6 October 2023 (UTC)

:::::: The issue is not the detail of his numbers but his reliability as a source. User generated content from Twitter is generally unacceptable. See WP:USERGENERATED. Polling Canada, while iffy, at least has a long and verifiable public track record beyond just the Nanos weekly roll. Also as most Nanos polls have a weight of only 250 (1/4 of 1000) a change of 0.5 will not impact the trend line very much. Lilactree201 (talk) 01:32, 6 October 2023 (UTC)

:::::::Agreed that it's irksome on some level when 32.4-31.5 becomes 32-32, but Lilactree is correct, a more 'official' source is needed, each individual Nanos poll holds little weight in the graph, and I would also add that the rounding works both ways for every party, about half the time in their favour and half the time not; so overall it should balance itself out. Indeed, I just updated the graph with all the new rounded numbers and the trendlines are virtually indistinguishable from before. Cheers, Undermedia (talk) 13:15, 7 October 2023 (UTC)

I will work on swapping what can be swapped, I think most Nanos polls have also been posted by Polling Canada. Lilactree201 (talk) 22:38, 3 October 2023 (UTC)

:I've also been holding off as I hoped someone who engages more with Wikipedia rules/policy would step in. But this case is clear cut, so better late than never! Lilactree201 (talk) 23:22, 3 October 2023 (UTC)

::Hey there. Regarding the two completely removed Nanos polls (due to JasonHRoy sourcing). I found the two Polling Canada tweets for those polls (July 14 and July 28, 2023). Would it be possible for you to restore those entries, without decimal points, and sourcing Polling Canada? It may be beyond my Wiki editing skills.

::https://x.com/canadianpolling/status/1681347870656544775?s=61

::https://x.com/canadianpolling/status/1686415997417234432?s=61 Splashcat62 (talk) 06:26, 5 October 2023 (UTC)

:::Thank you very much for finding these, yes they will be added! Lilactree201 (talk) 23:51, 5 October 2023 (UTC)

Got reverted when I added them so I'll make sure people know this, Nanos releases the "others" value with their polls. Examples:

- On the Feb 23, 2024 Poll [https://nanos.co/wp-content/uploads/2024/02/Political-Package-2024-02-23-FR-Updated-v2-with-tabs.pdf], on page 23 "Other % 1.0"

- On the Feb 16, 2024 Poll [https://nanos.co/wp-content/uploads/2024/02/Political-Package-2024-02-16-FR-with-tabs.pdf], on page 23 "Other % 0.9"

- On the Feb 9, 2024 Poll [https://nanos.co/wp-content/uploads/2024/02/Political-Package-2024-02-09-FR-with-tabs.pdf], on page 23 "Other % 0.9" WanukeX (talk) 02:19, 1 March 2024 (UTC)

There has been some controversy regarding EKOS being used as a reputable pollster. The President of EKOS, Frank Graves, has repeatedly issued Tweets that clearly show his discontent with Poilievre and the Conservative Party, and I don't doubt he uses his own personal biases to influence the results of his "polls". [https://nationalpost.com/opinion/geoff-russ-ekos-boss-frank-graves-ill-advised-threat-to-keep-pierre-poilievre-from-winning This] is just one of many examples. Ak-eater06 (talk) 20:17, 21 August 2024 (UTC)

:I would argue even if we are to keep EKOS polls, the most recent from October 30th, should be removed. The pollster itself has said that due to a very small sample size in Saskatchewan and Manitoba, 40 out of 1340, that the results "should be interpreted with caution." Frank's bias aside, the firm is saying this result is likely inaccurate. Btprep (talk) 01:13, 4 November 2024 (UTC)

::I strongly agree that EKOS should be removed, there is ample evidence which indicates that it is not a reliable pollster and is skewing the aggregate polling data to better reflect Graves' own personal biases. Southasianhistorian8 (talk) 06:56, 14 November 2024 (UTC)

:::I am not sure it is up to us to opine which anomalous polls to include or not. In fact including these anomalous polls such as Ekos and Mainstreet probably is better at giving us proper averages over the long-term. Many people argue over whether one polling company or another has partisan biases but we have no reliable third party that rates pollsters in Canada such as is done by FiveThirtyEight in the USA. I would recommend to include before Wikipedia is accused of bias in pollsters. All large firm polls should be included in my opinion and the data can speak for itself. Words in the Wind(talk) 21:27, 14 November 2024 (UTC)

:::Also, I would be highly cautious at using partisan political op-eds as grounds to exclude pollsters from our tables. If there were better third party sources this would be more logical. It should be note that though Ekos has underpolled the CPC during the last two elections (like most pollsters), their final polls did correctly identify the level of support for the Liberal and NDP parties, or were within the MoE. As Ekos has over polled both the NDP and Greens as well as Liberals, this tells us something about their polling methods. As I previously mentioned, many pollsters underpoll CPC support and this probably has something to do with likely voter issues for those who actually turn up on election day. Words in the Wind(talk) 21:43, 14 November 2024 (UTC)

:Obviously I'm biased as I work for EKOS, but wow... c'mon guys. Our polls are not 'biased' regardless of what Frank's opinions on Poilievre are. There's no evidence of that. And why would you throw out an entire poll just because a particular sub-sample is too small? -- Earl Andrew - talk 16:39, 24 November 2024 (UTC)

::Once again we have an Ekos poll (Jan 8, 25) where not just on province, but an entire region of the country comes with a disclaimer of likely inaccuracy. From the poll "the Liberals have regained their lead in Atlantic Canada; however, this finding should be interpreted with caution given the small sample size in the region" This poll also omits Saskatchewan and Manitoba and cannot by any means truly be considered a comprehensive federal poll. I previously only had issues poll by poll with Ekos, but this is ridiculous. Frank's bias aside, the quality of the poll is very much questionable. 74.15.76.205 (talk) 12:07, 10 January 2025 (UTC)

::: This is an irrelevant point. Regional subsamples in national polls will have smaller sizes and a large margin of error. This article lists the poll for its national topline numbers, not its regional figures. CASalt (talk) 06:48, 11 January 2025 (UTC)

:::I don't think EKOS should be excluded. If we were going by that logic, than we should remove Angus-Reid who a lot of times if you look at the history they would always have a huge gap for the conservatives even years back going back +15 points when you can see all pollsters showed the liberals leading and Angus-Reid would come along with their "online only" polls based on their members... that's like doing a poll to the same focus group over and over again. Kflack (talk) 05:32, 28 January 2025 (UTC)

:Since this discussion has been reignited, let me chime in: EKOS absolutely should not be removed, and doing so would be a serious violation of Wikipedia polices on neutrality and comprehensiveness. First of all, it's standard practice on Wikipedia to include low-quality and high house-effect polls (as seen in US elections, with Rasmussen and what not), even internals are included. Secondly even if removing polls based on bias or low-quality was consistent with Wikipedia policies (it's not), an National Post opinion piece (and Wikipedia editors' personal apprehension of Graves' bias/misconduct) certainly does not meet the threshold for that finding. CASalt (talk) 06:44, 11 January 2025 (UTC)

:I agree, Since October 2024 they have been sabotaging the Polling average. For instance Their LPC numbers between October 30 and November 11 is up 4 points, But that's fine it's more than a week there might be some statistical noise though it was a horrible outlier as shown by other pollsters.

:With the two recent recorded polls by them is getting out of hand. Firstly, Their Toplines are between Jan 17-21 but their regionals showed on website are between 13-21 and it overlaps with his previous poll January 16. Then accordung to their filling dates it showed a 4 points movement for LPC within 5 days.

:I don't know if this is Data manipulation but it's really suspicious and genuinely confuse the Polling average.

:Until Other pollsters have shown similar I suggest removing EKOS from the current table or at least the two recent ones published by them, Or not record those numbers on the current graph. Mason54432 (talk) 07:24, 23 January 2025 (UTC)

THIS IS a CORRUPT POLL PAID BY CORRUPT LIBRRAL PARTY.

I believe EKOS should be removed from the polling average for the 2025 Canadian Election. Their data consistently shows a significant discrepancy compared to other major polling firms, which undermines its reliability. For instance, in a recent poll, EKOS reported the Liberals at 32% and the Conservatives at 39%, while other respected pollsters, like Nanos, have the Conservatives much higher at around 46% and the Liberals much lower, around 20%.

This sharp contrast raises serious concerns about the accuracy of EKOS’s methods and conclusions. Given that their results appear so out of step with the broader consensus, it’s clear that EKOS is providing an unrepresentative picture of public opinion. Until they can offer more consistent and reliable data that aligns with other major polling organizations, I believe it’s best to exclude them from the polling average to maintain an accurate portrayal of voter sentiment. 142.126.38.80 (talk) 04:49, 23 January 2025 (UTC)

:I agree, Since October 2024 they have been sabotaging the Polling average. For instance Their LPC numbers between October 30 and November 11 is up 4 points, But that's fine it's more than a week there might be some statistical noise though it was a horrible outlier as shown by other pollsters.

:With the two recent recorded polls by them is getting out of hand. Firstly, Their Toplines are between Jan 17-21 but their regionals showed on website are between 13-21 and it overlaps with his previous poll January 16. Then accordung to their filling dates it showed a 4 points movement for LPC within 5 days.

:I don't know if this is Data manipulation but it's really suspicious and genuinely confuse the Polling average.

:Until Other pollsters have shown similar I suggest removing EKOS from the current table or at least the two recent ones published by them, Or not record those numbers on the current graph. Mason54432 (talk) 07:12, 23 January 2025 (UTC)

::I disagree. The consensus for these federal polling pages going back over 15 years has been to be de-facto inclusive of all polls/pollsters as long as there is no hard, demonstrable evidence of falsification or other shenanigans; and certainly not to exclude polls based merely on subjective whims, or just because they appear like outliers (also always useful to remind oneself that statistical theory predicts that outliers will happen). At different times since I've been contributing to these polling pages, different pollsters have taken turns being 'persistent' outliers (at one point it was Forum polls, other times Leger, Abacus, etc.), and even Campaign Research which some years ago faced actual professional disciplinary action over dubious practices was welcomed back into the table once they cleaned up their act and re-established a sound reputation. Some rather dramatic language is used above such as "sabotaging the polling average" and "horrible outlier", yet similar characterizations were not levelled at e.g. several Mainstreet polls since last election showing much wider leads than all other pollsters at those times, so I suspect political bias may be playing a role here. For a level-headed analysis of why the recent EKOS polls may be telling a different story (for the time being) than most other pollsters, I recommend this [https://abacusdata.ca/making-sense-of-divergent-polls/ blog post] by David Coletto from Abacus Data. Besides, their Dec 19 poll was perfectly in step with other pollsters, so I see no firm basis to declare fraud just because they've now released a couple of polls that are out of step. It's true that right at this moment the recent EKOS polls are having an outsized impact on the graph's trendlines simply because they've released 2 out of the 3 most recent polls, but the more sensible thing to do rather than call for their banning is simply wait for other pollsters to release further polls, which they inevitably will sooner than later: either they will validate EKOS's surprising and currently out-of-step results, or they won't, in which case the trendlines will adjust accordingly. If you backtrack, one or two outlying polls like this (if that's what they turn out to be) end up having minimal to negligible impact on the trendlines once the usual roster of assorted pollsters continue to chime in; again, they're just momentarily having an outsized impact because of their recency and quick succession. So I advocate for simply waiting it out. Cheers, Undermedia (talk) 20:48, 23 January 2025 (UTC)

:::That seems like a sensible proposition to me. It may not be in line with Wikipedia policy to bar a notable pollster just because we're skeptical of their findings. Southasianhistorian8 (talk) 21:40, 23 January 2025 (UTC)

:::The issue with EKOS isn’t just that their results are outliers but that they consistently skew in a way that significantly diverges from other credible pollsters, such as Nanos and Abacus. Their recent polls showing the Liberals at 32% and Conservatives at 39% are not minor deviations but major outliers compared to Nanos, which reports Conservatives at 46% and Liberals at 20%. While outliers happen, statistical theory doesn’t justify persistent and extreme discrepancies. Past pollsters with questionable results, like Campaign Research, faced accountability and improved their methodology. EKOS’s consistent outlier status undermines the accuracy of the polling average, especially when their results disproportionately influence trends due to timing. Relying on the idea that the trends will “balance out” when other pollsters release more data doesn’t address the immediate issue of distorting public perception. Including EKOS without addressing these issues diminishes the reliability of the overall average and the integrity of the data presented. 142.126.38.80 (talk) 22:19, 23 January 2025 (UTC)

::::Respectfully, while I'm flattered that you seem to ascribe so much importance and impact to my graph, I think it's quite a stretch to claim that a Wikipedia page consulted by surely no more than a fraction of a percent of Canadians (let alone frequently enough to notice a possibly momentary anomaly like this) could be "distorting public perception". The other thing about the graph is that trendlines aside, it also shows all the individual datapoints, so that readers can assess for themselves at a glance that, for example, the LPC trendline is currently being dragged by just a few very recent polls that happen to show a big bounce for the party whereas many more polls just before those showed lower LPC support.

::::You've now twice contrasted the recent EKOS polls with Nanos, yet during a period from Aug to Sep 2023 Nanos released several consecutive polls showing a significantly tighter race than other pollsters, including a poll on Sep 1st that showed a tie when the previous poll by Leger showed the CPC up by 9 and the next poll by Abacus showed the CPC up by 14. These things happen in polling, and I'm no more in favour of banning EKOS polls because of it now than I was of banning Nanos polls because of it then.

::::FTR, I have my own reservations about EKOS's and other IVR pollsters' methods (I tend to agree with Coletto's article I referenced above), but recognize that my personal reservations don't constitute grounds to call for banning pollsters from this page. In any case, excluding outliers or even persistent outliers would defeat the purpose and spirit of aggregating polls, which is the aim of this page. Heck, in the recent U.S. election a similar case could have made to 'ban' AtlasIntel polls from aggregates because they singularly showed Trump consistently ahead nationally and in every swing state while almost all other pollsters were consistently showing Harris narrowly ahead nationally and mixed results across the swing states... Cheers, Undermedia (talk) 00:55, 24 January 2025 (UTC)

:::::LOL, now I agree these EKOS updates are getting a bit out of hand (not necessarily the numbers themselves, but the frenetic pace of successive releases). I'll therefore put further graph updates on hold for now until some other pollsters chime in, as I don't think I've ever encountered a situation before where 3 of the 4 most recent polls were by the same pollster (except possibly Nanos, but their weekly rolling updates are only given a quarter of the weight of a 1,000 sample size poll in the graph. Cheers, Undermedia (talk) 19:58, 24 January 2025 (UTC)

::::::Thanks for being rational. Mason54432 (talk) 06:51, 25 January 2025 (UTC)

::::::Have you seen their latest poll? Check it out. 142.126.38.80 (talk) 22:57, 29 January 2025 (UTC)

:::::::Where can I see it? Cremastra (talk) 23:12, 29 January 2025 (UTC)

::::::No pollsters publish 3-5 polls within a week, this is a straight up Push poll behavior I suggest if we can add some reminder/notes regarding EKOS recent polls. Mason54432 (talk) 08:10, 30 January 2025 (UTC)

:::::::[https://www.youtube.com/watch?v=G0ZZJXw4MTA And would you oppose the reintroduction of... national service?] Cremastra (talk) 13:29, 30 January 2025 (UTC)

:338Canada, which I regularly monitor, just released a "Bullseye" chart of pollsters, including EKOS, comparing how the differ from each other. You can look at it [https://338canada.com//bullseye.htm here]. Hover over the coral-coloured "EKO" box. Cremastra (talk) 16:00, 26 January 2025 (UTC)

::Thanks. They're all over the place. While their 2025 polls so far have been favourable to the LPC, this reminded me of their 24 Sep 2023 poll that was the first to show the LPC < 25% (22.9%) as well as the highest CPC support to date at the time (42.1%). Cheers, Undermedia (talk) 01:02, 27 January 2025 (UTC)

:Apparently Mainstreet is reporting an LPC lead in Ontario, which tracks with Ekos, despite being an online poll. [https://x.com/CanadianPolling/status/1883683431886029003] -- Patar knight - ^chat/_{contributions} 02:55, 27 January 2025 (UTC)

::We shouldn't be picking pollsters. Both [https://338canada.com/polls.htm 338 Canada] and the [https://newsinteractives.cbc.ca/elections/poll-tracker/canada/ CBC Polltracker] are using EKOS. I think we should follow what they are doing. Seems kind of WP:OR to do otherwise.--Darryl Kerrigan (talk) 06:40, 28 January 2025 (UTC)

:Strongly disagree. While EKOS is an outlier, that does not represent a good enough reason in and of itself to remove EKOS from the average or table. EKOS is treated by Canadian media outlets as a legitimate pollster, it would be a violation of WP:NPOV to start asserting which pollsters are and aren't legitimate, not on the basis of scientific methodology, but rather on the basis of the pollster just being an outlier. CaelemSG (talk) 07:40, 14 February 2025 (UTC)

Just a quick update to clarify that I've again put the graph on hold until other pollsters release further numbers, as it's problematic to have a single pollster exerting so much sway on the leading ends of the trendlines via this bout of unusually frequent releases. This is exclusively an issue with the graph, and in no way an argument to remove the EKOS polls from the table. Cheers, Undermedia (talk) 15:18, 30 January 2025 (UTC)

:Perhaps the graph should be changed to only include the most recent poll for each pollster. 2204happy (talk) 04:32, 31 January 2025 (UTC)

::I think that would likely prove fairly tricky coding-wise (the graph is generated by an R script that reads the data directly from the poll table as published on Wiki), but in my long experience contributing to these polling pages, the current situation is quite exceptional and presumably won't last, so I think it's OK for now to just put graph updates on hold as necessary. If EKOS persists to release polls at a frequency far outpacing other pollsters, I'll re-evaluate, but I doubt they will. Cheers, Undermedia (talk) 14:26, 31 January 2025 (UTC)

:::Understood, thanks for your response. 2204happy (talk) 00:17, 2 February 2025 (UTC)

• I'm surprised that anyone (especally an IP with no edit history) would suggest not using EKOS polls - or take this suggestion seriously. EKOS got a really surprising result - even to them. And so they went out and tested, and tested again. This is most unusual, but I don't see a concern, as it's one time. The biggest issues is that we have polls essentially asking two different questions. One is asking who is your preference, using something in the question like "The Liberals lead by Justin Trudeau". While others (such as Ekos and Mainstreet) adjusted the question to "The Liberals lead by a New Leader". No surprise that the later are showing the Liberals about 10% higher. If you look at the Leger poll from this week for Quebec (that's been added here), it puts Liberal support in Quebec at only 29%. But if you dig deeper, that's with Trudeau as leader. When they ask about Carney [https://x.com/CanadianPolling/status/1887387927854113192 they get 38%]. Meanwhile, if we doubt the accuracy of Mainstreet, it is then odd, that polling the very same people for the Ontario election gives results similar to the other pollsters. Ekos hasn't polled provincially since 2022, so hard to compare them. Nfitz (talk) 17:26, 6 February 2025 (UTC)
• :EKOS' Founder Frank Graves on Twitter: "Going to make sure you [Poilevre] are never going to lead my country. I don't make idle threats".
• :I think this is more than enough to question the accuracy of the polls, especially given how this steep decline has only been reported by them. 2204happy (talk) 05:59, 7 February 2025 (UTC)

::::It's a good thing the EKOS poll was included, as it picked up on a major trend a month early. This is why we shouldn't pick and choose which polls to exclude. --38.49.167.105 (talk) 14:41, 25 February 2025 (UTC)

I removed this poll because it is not even published on the [https://www.ekospolitics.com/ EKOS website], and the linked source is just a tweet from the founder of EKOS (and the tweet has no source in it).

It also seems suspicious that the numbers for the parties other than the CPC and LPC are nearly identical to the EKOS poll taken a few days before. CGP05 (talk) 02:28, 26 January 2025 (UTC)

:Multiple polls in the table link to tweets by pollsters, we've been allowing these as sources for polls for some years now.

:And it's indeed a 'rolling' poll that overlaps with the one before it: that's why it has "(1/2)" in the sample size cell, which reduces its weight in the graph by half. But I haven't updated the graph with this poll at all; see the discussion just above. Cheers, Undermedia (talk) 05:05, 26 January 2025 (UTC)

Mainstreet Ontario federal polls are mostly related to 2025 Ontario general election and will be rolling everyday until February 27, I don't think that's appropriate as other regional polls on here aren't daily tracking like Mainstreet where they publish regional polls monthly/regularly. Mason54432 (talk) 08:08, 4 February 2025 (UTC)

::I would tend to agree that listing the Mainstreet poll for Ontario everyday of their election may be overkill. We didn't include the federal numbers from the Mainstreet polls taken during the Alberta 2023 election. One poll per week from the Mainstreet polling might be an appropriate middle ground, or a first, middle and last poll.

:Lilactree201 (talk) 01:09, 5 February 2025 (UTC)

::First and last might be better Mason54432 (talk) 13:33, 5 February 2025 (UTC)

:::Is the first the Jan 26th poll or the Jan 30th one? Southasianhistorian8 (talk) 16:39, 5 February 2025 (UTC)

::::Jan 30 cuz it's after the Ontario election writ, but either one is fine Mason54432 (talk) 17:02, 5 February 2025 (UTC)

• I have no idea why you wouldn't include them. They are polls. They are real. If they are running a 3-day running total, then include every 3 days. This is especially significant because while their provincial polls show little change - and perhaps even the Conservatives gaining (within th margin of error), the federal portion is showing a statistically significant shift towards the Liberals (which I'd assume is related to recent actions by the United States). With February 4th results available, and very different from those from last month, why show zero polls from this month? Nfitz (talk) 23:52, 5 February 2025 (UTC)
• :Every 3 days comes out to 7 or 8 polls until the 27th, that still seems excessive for one pollster in a period of just one month. I think once a week might be more reasonable. Southasianhistorian8 (talk) 01:49, 6 February 2025 (UTC)
• :: I'd be fine with every 3 days or once a week. Lilactree201 (talk) 03:18, 6 February 2025 (UTC)
• :::Once a week or first and last are fine as well Mason54432 (talk) 15:09, 6 February 2025 (UTC)
• ::We add national polls almost daily at some times of the year. I don't know why we would be concerned about having so many Ontario-only polls - other than that they are using (I think) a 3-day total, with an overlap. If each daily poll was unique (and I don't think they are) I'd think we'd put them all in. Nfitz (talk) 03:57, 6 February 2025 (UTC)
• :::I agree that simply because a given pollster is releasing polls frequently is not a valid reason for excluding them from the tables on this page. The whole purpose of this page is to aggregate ALL available polls unless there's some hard, unequivocal indication they're dubious (e.g. professional censure of a pollster by the Canadian Research Insights Council), and it's certainly not up to us amateur Wiki editors to subjectively make that call. It's also beginning to concern me that there's been an emerging pattern here of late whereby calls by certain editors to remove/exclude certain polls/pollsters seem to correlate with said polls being more favourable towards a certain political party and less favourable towards a certain other party than other pollsters may be reporting. Folks, I've been contributing to these polling pages for the better part of 15 years, and the only way to manage them objectively/soberly and keep them from descending into an endless and inappropriate censorship battle is for all editors to check their personal politics at the door, full stop.
• :::Now, there seems to be some consternation about these particular polls being daily 3-day rolling polls, meaning two thirds of the data from each new poll overlap with the 2 previous polls. Valid concern. In the main national-level polling table, we deal with this by adding a fraction to the sample size indicating what proportion of the data is 'unique' and non-overlapping with the previous polls in the rolling series. The graph in turn multiplies the total sample size by this fraction, effectively reducing the weight of the rolling polls in the trendlines and essentially ensuring these partially non-independent datapoints aren't disproportionately influencing the trendlines. This is a fancy way to do it specifically with the graph in mind, so if folks for whatever reason don't want to do it this way for these Ontario-level polls with no associated graph, then the other logical way to do it would be to only include every 3rd poll, as has been suggested. Cheers, Undermedia (talk) 15:31, 6 February 2025 (UTC)
• ::::The claim that some editors are supporting certain candidates is a bit extraordinary. Obviously someone who is blatantly supporting a candidate should not be participating here, @Undermedia. Do you have any evidence of this? Nfitz (talk) 15:54, 6 February 2025 (UTC)
• :::::Nope, no damning proof, just a tendency I've been anecdotally detecting of late, but I could be wrong. Nevertheless, a page like this is inherently vulnerable to this sort of thing because the topic of political polling tends to be, by nature, politically charged! So just taking the opportunity to remind and invite all editors to approach this page through a strictly non-partisan lens. Cheers, Undermedia (talk) 16:02, 6 February 2025 (UTC)
• ::::::Addendum: I mean, I say no proof, but you do realize the page is currently protected from edits by non-confirmed users because of a flood of vandalism/censorship targeting specific recent polls? I trust the regular roster of confirmed editors here would not stoop to this level. Cheers, Undermedia (talk) 16:07, 6 February 2025 (UTC)
• :::::::Interesting. I only pop in every few years or so when there's something interesting going on. So I don't pay little attention. The problem with even confirmed editors, is that it's not like the problem ones are going to put a flag or something on their User Page saying "I support Trump" or MAGA or mysogyny or something. Though I just noticed the thread above about ignoring EKOS because people don't like the results!
• :::::::I'll restore the Mainstreet data using the 1/3 symbol like we do for Nanos. Gosh, I hadn't even realized that Nanos's numbers were a 4-week total! That won't show trend changes quick. Nfitz (talk) 17:13, 6 February 2025 (UTC)
• ::::::We don't have to cross a line into casting aspersions towards specific editors, but it's certainly no coincidence that an apparent LPC rebound is followed by two separate disputes over whether those polls saying so should be included or not (also, the bias doesn't have to be "blatant"). ―"Ghost of Dan Gurney" _(hihi) 22:00, 6 February 2025 (UTC)
• ::::::: Reading the details of the Mainstreet Ontario polls, they are prompting for the Liberal Party under a new leader. Should that perhaps be noted? Lilactree201 (talk) 22:58, 6 February 2025 (UTC)
• ::::::::Very interesting, and one would definitely expect that wording to have an impact on responses at this juncture. Out of curiosity, I checked the wording used by other pollsters, and interestingly EKOS does not allude to party leaders in its questionnaire. Apparently neither does Ipsos, Nanos or Abacus. But Leger's most recent poll still had "Justin Trudeau’s Liberal Party of Canada", along with "Pierre Poilievre’s Conservative Party of Canada", etc. Honestly I'm not sure what to make of this; we're in a pretty exceptional situation at the moment. It seems fair to specify "new leader" for the LPC since that's what's imminently going to happen (and there's zero chance an election will be held beforehand), yet it's also still technically accurate to say "Justin Trudeau’s Liberal Party of Canada". Cheers, Undermedia (talk) 00:53, 7 February 2025 (UTC)
• ::::::::As far as I know neither Ekos nor Mainstreet are mentioning Trudeau, and instead a new leader. At least one or two of the others are saying Trudeau. One would have to go through all the recent polls to check the question. Maybe. I suspect its what's driving the difference between those firms and the others. On the other hand, we are seeing a smaller but significant swing with (all?) pollsters. What have we done in the past few years with the resignations of various Conservative leaders. Nfitz (talk) 01:38, 7 February 2025 (UTC)

:kind of unrelated, but iirc there are other Ontario polls too that are not in the graph. All of the recent ones are from mainstreet. CGP05 (talk) 00:54, 27 February 2025 (UTC)

We've had this debate countless times over the years, and what we've observed over time is that if we allow editors to arbitrarily add various events that may or may not impact voting intentions to the table, it quickly snowballs out of control, and before we know it the table gets excessively cluttered with these non-poll rows, and we end up constantly having to debate whether this or that event merits inclusion in the table. This is why we settled some time ago on only allowing a very limited assortment of specific events such as party leadership changes, new governing arrangements among parties, and by-elections; though IMO it's debatable whether even these things consistently impact voting intentions in a significant way (they don't).

The reality is it's very hard to predict which events will end up having a significant impact: oftentimes certain events are anticipated to have an impact but turn out not to (in which case we end up of having to debate over their removal from the table after the fact); other times events that at the time seem minimally significant turn out to create a butterfly effect of sorts. For example, in the span of only ~6 months around the second half of 2023, average LPC support dropped a rather significant 6 points while CPC support rose by the same amount -- this was the most rapid and numerically significant shift in voting intentions since last election (certainly more than any of the listed by-elections, party leadership changes, or LPC-NDP governing arrangement) until the one we're now witnessing. It must have been caused by one or more events, though evidently not deemed significant enough at the time for anyone to propose adding to the table. Should we now dig back and recall what it/they was/were, and add it/them to the table retrospectively? As for the current shift in voting intentions -- what's really behind it? Was it Trudeau announcing his resignation? Trump's swearing-in? Trump's subsequent tariff showdown? Excitement over Mark Carney? A combination of all of the above? Should we just add them all to be safe?

FTR, my own view is that no events should be added to the table, but in the community spirit of Wiki I've always been willing to work towards a middle ground with other editors with diverging views; as long as the consensus we arrive at is clear and applicable in a consistent and objective fashion. So right now I'm just opening up the general discussion again, which seems to be necessary every so many years when the 'event-adding fever' starts heating up again. Cheers, Undermedia (talk) 18:27, 12 February 2025 (UTC)

:{{u|Undermedia}} could not have sumed it up better; I entirely agree with this.

:It's not only hard to predict which events will end up having a significant impact (sometimes you may have several simultaneous events, sometimes it's an event spanning not just one day, but a couple weeks, months or years), you also have the problem of identifying:

1. What does "event" mean? It's an executive order? An election? A whole campaign? A single protest? A chain of protests? A war? A pandemic? An announcement? A declaration? An interview? A speech? If it's a speech, it's a particular word or sentence? It has to be an event from the polled country or (as some people are attempting to introduce here) an event happening in another country?
2. What does "impact" mean? A party gaining (or losing) 1 point would count as such? It has to be 10 points? 5? 20?
3. For how much time does such "impact" have to last in order for it to be considered as "significant" for inclusion? Most opinion polling trends tend to not last forever. They usually last some weeks, then subside. And for trends lasting longer, you'd typically have multiple other events that could be contributing to such a trend. A government will typically wear out after many years in power and a chain of unrelated, insignificant (by themselves) events may end up triggering a new polling trend.

:Most of the time you would have to end up justifying that an event affects polling with circumstancial evidence, but that's it. And that is wholly within WP:SYNTH (if not outright WP:OR) territory. In order to avoid that, you would have to list every-single-event that "may" "possibly" affect polling, which would mean you would have more events than polls in the table; otherwise, you'd be just cherry-picking which events should be in and which shouldn't.

:We should remember that, in the end, this is a table of opinion polls, not a timeline of events. You have 2025 in Canada, 2024 in Canada, 2023 in Canada, etc. for that. That's why I also agree that, if anything, we should discuss whether the currently-shown events (leadership and by-elections) should be removed, not to add even more. Impru20^talk 20:30, 12 February 2025 (UTC)

::As a follow-up to this comment, just highlight that [https://en.wikipedia.org/w/index.php?title=Opinion_polling_for_the_2025_Canadian_federal_election&diff=1275514797&oldid=1275398183 this edit summary] by {{u|80.43.196.197}} is the perfect definition of WP:SYNTH, and self-evidences the main issue with attempting to single out any specific event in the table just because we may believe that it is affecting opinion polls. Once again: the table is on opinion polling, not a timeline of events nor a table on the impact of events in opinion polling. If you want to have a timeline of events, use 2025 in Canada. If you want to comment on the particular impact of any single event on opinion polling, use the 2025 Canadian federal election article and write it down in prose with proper sources (and even this will come mostly from media opinions or circumstancial evidence). Opinion polling tables are not meant for it, or even constitute the best way to convey such a claim due to space constrains. Impru20^talk 14:31, 13 February 2025 (UTC)

• Holy TLDR folks. Try and be succint. SYNTH doesn't apply - I quoted a source when I restored it. I agree that we should be careful what we add, but I don't think one can argue that acts of war and threats of occupation by our former best neighbour and ally shouldn't be mentioned, when we have ELEVEN entries for by-elections, and three for changes in leadership of the Green Party! There's no end of media reports that the election of the far-right and hostile government has changed the Canadian election (and the opinion pieces go further, suggesting that the Grande Orange may get the Liberals re-elected). Where is this previous discussion BTW - we certainly didn't used to post any by-elections, let alone every. That said, the date should probably be the inaugaration, not the umpteenth time out of umpteen-hundred that he threatened to annex Canada through economic force. Nfitz (talk) 17:44, 13 February 2025 (UTC)

:::Oppose adding additional events. Slippery slope. I also feel that we don't really need to add by-election dates or even resignations leaders in the polls and should keep it at adding when new leaders are appointed by parties. We have had these conversations for many countless notable national and global political events. Words in the Wind(talk) 17:50, 13 February 2025 (UTC)

::::The only previous discussion I see was about the NDP agreement, where everyone disagreed with you. What discussion was there elsewhere - particularly about by-elections? Nfitz (talk) 18:13, 13 February 2025 (UTC)

:::::This is merely the page for opinion polling after the 2021 election and leading up to the 2025 election. There have been pages like this in between previous elections going back quite some time, and I've personally been contributing to them since in between the 2008 and 2011 elections. It's across all these previous polling pages that numerous discussions on this topic have already taken place, and ideally there should be some consistency and continuity across these successive opinion polling pages. Cheers, Undermedia (talk) 20:11, 13 February 2025 (UTC)

::This further convinces me that no events at all should be added to the table: the by-elections should be removed, as should the party leadership events and governing arrangement events. Cheers, Undermedia (talk) 20:15, 13 February 2025 (UTC)

::Two comments here:

::1) 31 January was not Trump's only threat, why cherry-pick that? And exactly what do "acts of war and threats of occupation" have to do with a table which lists opinion polling? As said: if you have sources proving than that may have a relevant and/or notable impact in the election, then by all means go to 2025 Canadian federal election and add it there with the full context. If you want to have a list of all events, use 2025 in Canada. But what is the purpose of adding random, cherry-picked events to a table which lists opinion polls and which doesn't even let you enough space to fully explain their context? Yes, this is SYNTH if not outright OR.

::2) If what you propose is that we should remove all events if we cannot add others: I'd absolutely agree with you on that and would be in favour of removing all of them altogether (by-elections, leadership changes and government arrangements as well). Impru20^talk 08:32, 14 February 2025 (UTC)

:::FWIW, Opinion polling for the 2024 United Kingdom general election similar to us seems to list party leadership events, by-elections and "local elections" (the equivalent of provincial elections in Canada?—I don't think it would add any value to list those here), with the only other event beyond those being "Partygate scandal begins" listed on 7 December 2021. Though looking at the graph, it's not clear at all that this singular event had a particularly pronounced effect on voting intentions: Conservative support had already been steadily declining (and Labour steadily rising) since around June 2021, Labour had already overtaken the Conservatives before Dec 7, and these trends simply continued after that at roughly the same rate until they peaked around Feb 2022. So if anything I'd say that serves as an argument against selecting some singular event on a specific date that supposedly impacted voting intentions to an exceptional degree. Cheers, Undermedia (talk) 14:16, 14 February 2025 (UTC)

Frank Graves just posted a poll here [https://x.com/voiceoffranky/status/1899566662833336547], then within about 5-10 minutes deleted it, there are screenshots of the post around [https://x.com/RealAlbanianPat/status/1899570367653400667] but obviously a bit messy so probably needs a talk page discussion. WanukeX (talk) 21:27, 11 March 2025 (UTC)

:IMO we need to see something directly from EKOS. Carlp941 (talk) 22:09, 11 March 2025 (UTC)

:: I don't think we need to bother with it for now. If he posts it again we will know he is standing behind it. Lilactree201 (talk) 01:04, 12 March 2025 (UTC)

:::Agreed, a tweet deleted by its author is not an adequate source for poll on this page. Undermedia (talk) 02:01, 12 March 2025 (UTC)

Could someone ask Greg Lyle (president of IRG) what the field dates of his latest poll are? He has a twitter account. First page of the poll says 6-10 but survey methodology says 5-9. 96.20.147.13 (talk) 23:20, 11 March 2025 (UTC)

:Best to go with the detailed survey methodology on page 35, where March 5-9 is stated twice. The dates given on the title page (March 6-10) are likely an error, as it's furthermore unlikely the results could've been fully compiled, analyzed, and the whole report produced by the next day after the field work was completed. Cheers, Undermedia (talk) 00:29, 12 March 2025 (UTC)

::Agreed. There is no reason why we shouldn't go by the detailed methodology on page 35 (where the March 5-9 is stated twice) and go instead with the one on the first page, which is a likely error. Impru20^talk 09:55, 12 March 2025 (UTC)

:Is Innovative legit? Here's a quote from their [https://innovativeresearch.ca/wp-content/uploads/2025/03/Trump-Reaction-March-W1.pdf report] "Following the controversy in his role in moving Brookfield Asset Management’s HQ to the US, Mark Carney has dropped 10 points in net favourables." That sounds more like a personal attack than I've seen in such reports. This report is literally the first I've heard mention of "his role". This seems to be a controversy in their own minds. Nfitz (talk) 18:08, 12 March 2025 (UTC)

::There was some noise about the move awhile ago. ([https://www.cbc.ca/news/politics/mark-carney-brookfield-1.7469116 CBC], [https://www.theglobeandmail.com/canada/article-carney-says-he-should-have-been-more-precise-in-answer-on-brookfield/ Globe and Mail]) ARandomName123 (talk)^{Ping me!} 04:11, 13 March 2025 (UTC)

::Innovative Research Group is definitely legit. It's unfortunately not so uncommon for various pollsters to wander into this sort of questionable conjecture as they try (often too hard) to ascribe 'meaning' to their results, but IRG is otherwise very well established and the data itself is beyond suspicion. Cheers, Undermedia (talk) 13:17, 13 March 2025 (UTC)

Someone deleted it claiming it was "biased" and not accepted by aggregators. I checked it and this is not true, it is used by Fournier and Grenier(CBC). It was commisionned by Global News. I find it ridiculous to delete a poll by such a mainstream pollster. This is a deliberate act of trolling/sabotage imo. 96.20.147.13 (talk) 00:23, 14 March 2025 (UTC)

:Definitely it should not be removed. Looks like the editor who did that is an WP:SPA whose only edits so far have been to remove polls from this article. Impru20^talk 12:19, 14 March 2025 (UTC)

On mobile, it is impossible to scroll right/left on national poll results, someone please fix this. 201.95.56.29 (talk) 16:25, 14 March 2025 (UTC)

:Can it please be fixed impossible to scroll left or right is this intentionall 97.111.184.37 (talk) 21:37, 14 March 2025 (UTC)

:Yeah, this seems to be a recent development. But old revisions of the page also exhibit this behaviour, leading me to think this may be a Wikipedia problem, and not a result of recent edits to this page. Splashcat62 (talk) 22:23, 14 March 2025 (UTC)

:{{done}} I do some code modifications, it should work on moblie view now :) Haers6120 (talk) 00:18, 15 March 2025 (UTC)

Really? Ak-eater06 (talk) 22:20, 14 March 2025 (UTC)

:I'd suggest removing it from the data used in the graph, but keeping it in the table. pancake (talk) 23:19, 14 March 2025 (UTC)

::I agree Mason54432 (talk) 07:17, 15 March 2025 (UTC)

::I don't think excluding the poll is the right idea anymore, based on the points brought up by @Undermedia, @Carlp941, and @CASalt. pancake (talk) 16:46, 16 March 2025 (UTC)

::As of recently, the poll is becoming less of an outlier, so the arguments against inclusion are becoming less valid. pancake (talk) 16:02, 20 March 2025 (UTC)

:::It's valid even if it's an outlier. This article does not exist to predict the results of the election, it exists simply to catalogue the history of polling for this election. If you want election predictions there's plenty of other sites for that.

:::If there's outliers and there's nothing to suggest the poll wasn't conducted in good faith, it should continue to be included in the record of polling for this election. 103.8.18.128 (talk) 23:06, 20 March 2025 (UTC)

::::I completely agree, I was just pointing out the fact that the poll is becoming less of an outlier, because the whole reason this discussion began was because of that fact. pancake (talk) 15:58, 21 March 2025 (UTC)

:This suggestion shouldn't even be entertained. There is literally zero basis to exclude the poll, either from the table or the trendline. Not under Wikipedia rules, and also a general rule you should never exclude outliers from poll aggregations. CASalt (talk) 23:32, 14 March 2025 (UTC)

::Basis to exclude poll:scroll up a bit to March 11 EKOS Poll, same numbers, but released 2 days before the end of the field dates. Slightly suspicious? 170.203.207.175 (talk) 03:48, 15 March 2025 (UTC)

::: I don't believe those numbers were identical to the new numbers, but even assuming you're right, this is WP:OR. We can't even have this conversation until we have a reliable source saying EKOS makes up numbers or something. CASalt (talk) 06:33, 15 March 2025 (UTC)

::Idk man, the leader of EKOS literally makes public anti-conservative statements like stating that he would make sure Pierre never becomes prime minster. I think his public statements make it very real to suspect that EKOS is making blatant push polls designed to influence public opinion. In addition to their wildly off polling data, literally 10-20 points off any other. Nudrien (talk) 05:49, 15 March 2025 (UTC)

:::Do you have any reliable source that describes EKOS as engaging in push polling or making up numbers? Because WP:OR. Also the practice on Wikipedia is to include partisan pollsters, even internal polls are included, which are quite literally push polls. CASalt (talk) 06:30, 15 March 2025 (UTC)

:Yes the EKOS polls should be included, and yes they're already in the graph. I've just about run out of patience to even entertain this debate again for, like, the 20th time. Undermedia (talk) 23:41, 14 March 2025 (UTC)

::Their reliability rating has been dropped with other reputable pollsters such as calling them out directly for inaccuracy such as [https://x.com/DavidColetto/status/1900679363307127223 David coletto] from abacus data 67.208.31.164 (talk) 02:29, 15 March 2025 (UTC)

:Yes? Unless you have evidence of foul play on their end, you're just trying to exclude a result you do not like. Carlp941 (talk) 04:41, 15 March 2025 (UTC)

::I Certainly think LPC is currently leading in Alberta and CPC is only 5% in Manitoba. Totally accurate and Totally happening, just like when CPC was near 50%, Totally accurate numbers no doubt. LOL

::But for the sake of the policy here it should be included and let people look back how ridiculous EKOS is no matter who wins next election, whether you like this poll or not. Mason54432 (talk) 07:16, 15 March 2025 (UTC)

:::That's the spirit. Let them stand, and if they turn out to be "ridiculous", then readers can come back here for years to come to be reminded of how ridiculous they were. Better in a sense than potentially sparing them that exposure by wiping them from the record here and pretending they never happened, is it not? Undermedia (talk) 12:37, 15 March 2025 (UTC)

:::yep 👍 Carlp941 (talk) 16:56, 15 March 2025 (UTC)

:Definitely agree with {{u|CASalt}}, {{u|Undermedia}} and {{u|Carlp941}}: it should not be removed, much less on the basis of personal opinions.

:*Firstly: yes, it is an outlier, but we do not remove polls just because they are outliers, as that would breach WP:NPOV. Particularly, I am amused at how these debates are only sparked by pro-LPC outliers: EKOS +19 September 2023 poll, Mainstreet +30 November 2024 poll or Abacus +19/+20 February 2025 polls (I would even say that the Innovative +6 March 2025 poll can be regarded as somewhat of an outlier as of currently) did not seem to spark such debates despite these being clear outliers at the time; however, they were so in favour of the CPC.

:*Secondly: EKOS seems to have a tendency towards maximizing incoming polling trends, and not necessarily towards the LPC: in September 2023 they predicted a +19 CPC lead, the largest at the time, similarly to June 2024's +24; this mirrored what other pollsters started showing a few days/weeks later. Even in December 2024, EKOS predicted a +25 CPC lead (one of the highest at the time) with the LPC at barely 19%, and this was merely four months ago. Earlier discussions such as this one attempted to have EKOS removed back in January because it showed the CPC ahead by "only" 7 points, and even claims were made that EKOS was attempting to "sabotage" averages. Since then, we have seen that EKOS correctly predicted the polling trends that were then followed by other pollsters, and that the trend averages are not being fueled just by EKOS polls, as Ipsos, Mainstreet and Liaison all show a LPC victory now and even Abacus shows a tie.

:*Thirdly: Regional sub-samples in a national poll (particularly one with a 1,000–2,000 sample size) should not (and are not, actually) taken into account in this article due to the large margin of error that one would expect from the regional distribution, so: why is it even being used as an argument here?

:*Fourth: Weird shit happens some times. The PC went from 18–21% to 43–45% in opinion polls in 1993 upon Mulroney's resignation, and kept in the mid-30s following Campbell's swearing-in and until the election campaign began. The current LPC surge may or may not last, but it is not for us to decide whether it should be larger or smaller than predicted by some pollsters: we just report on what reliable sources tell us.

:*And last, but not least: I already hinted at this in the previous points, but we do not pre-select which opinion polls to add based on how their findings look to us. Our opinions on the findings are irrelevant. Whether we like the results or not, or find them compelling, is irrelevant. If the source is verifiable and reliable (and, thus far, there is nothing pointing out that EKOS is not) we report it. That's it, that's all. Impru20^talk 09:10, 15 March 2025 (UTC)

:Outliers happen, and the major aggregators include EKOS (their polls are also reported on by reliable sources). Are we going to have this discussion if a CPC+12 poll came out today? Dingers5Days (talk) 18:12, 15 March 2025 (UTC)

::Are we going to seriously add a poll that had the bloc with 1% in Alberta? Are you serious?

::https://ekospolitics.com/wp-content/uploads/20250314datatables_phone.pdf 142.115.13.121 (talk) 19:14, 15 March 2025 (UTC)

:::Not as serious as you opening a new thread on this very exact same issue (with this one still ongoing) because of no reason, it seems. Impru20^talk 20:28, 15 March 2025 (UTC)

On mobile it's impossible to scroll left/right on poll table. 189.55.7.169 (talk) 22:21, 14 March 2025 (UTC)

:Came here to also comment this. It's still working for the other tables, just the main national poll table where it is broken. Potentially someone with better coding ability can identify and fix it? 86.160.217.193 (talk) 23:47, 14 March 2025 (UTC)

:{{done}} I do some modifications to table code, it should work on moblie view now :) Haers6120 (talk) 00:16, 15 March 2025 (UTC)

::That helped on Android phone, but still broken on iPad Mini.

::Hmm, the 2019 and 2021 poll pages currently also have unscrollable tables. They used to work. This is probably a Wikipedia-wide issue (like last year’s text size issue). Splashcat62 (talk) 03:20, 15 March 2025 (UTC)

:::Hi, could you go to the Opinion polling for the 2025 German federal election article to check if you could scroll the table? Haers6120 (talk) 20:02, 15 March 2025 (UTC)

::::Sure thing.

::::No, that table will not scroll for me either, on iPad. Android works.

::::“List of prime ministers of Canada” doesn’t scroll, on either iPad or Android.

::::”List of premiers of British Columbia, doesn’t scroll on iPad, but does on Android.

::::I’ve seen a few other tables that no longer scroll for me.

::::(Google Chrome was the browser in all cases). Splashcat62 (talk) 19:46, 16 March 2025 (UTC)

:::::Got it! Yeah I think the issue is on wiki side :) Haers6120 (talk) 22:41, 17 March 2025 (UTC)

::Works on Apple mobile again. Thank you! 170.203.207.175 (talk) 03:50, 15 March 2025 (UTC)

Okay, I need some help here. I understand that it would be unfair to delete a poll simply because it’s an outlier, but in the latest EKOS polls data table, they reported the Bloc Québécois at 1% in Alberta. What reputable polling company would release such inaccurate numbers?

https://ekospolitics.com/wp-content/uploads/20250314datatables_phone.pdf 142.115.13.121 (talk) 19:17, 15 March 2025 (UTC)

:@142.115.13.121

:Hello!

:What are you exactly reproaching to EKOS here? The fact that the Bloc Québécois has "only" 1%, that the Bloc Québécois is mentionned even though they never even candidated in Alberta and won't begin today (more comprehensible complain for me than the first one), or something else?

:Have a nice day! Tacomniscient (talk) 19:49, 15 March 2025 (UTC)

::(Post-scriptum: I saw that you complained about that poll in another place. Please don't spam anymore, it's pretty annoying to be honest, particulary when you know how much that polemic about EKOS is tense here...) Tacomniscient (talk) 19:56, 15 March 2025 (UTC)

:::Spam? I replied to one person then added a topic - No ones’s taking this seriously at all. 142.115.13.121 (talk) 20:49, 15 March 2025 (UTC)

::::The persons polled could have been out-of-province students from Québec who still intend to vote for the BQ while currently living in Alberta. Could also be that the persons who were polled thought the BQ was a different separatist party and that's why it has a blip.142.161.104.234 (talk) 04:59, 16 March 2025 (UTC)

::::@142.115.13.121

::::Hello!

::::I understand, but the attacks against EKOS are so frequent that I personnaly don't really like to see people complaining several times (for this case, two, but it's still several) for the same topic.

::::Furthermore, I would have prefered to see you responding to my first message instead of just reacting to my post-scriptum, so I repeat: What are you complaining for exactly?

::::Have a nice day, Tacomniscient (talk) 07:32, 16 March 2025 (UTC)

I'd like to discuss the possibility of folding each of the tables by default like what was done in the UK opinion polls for 2024. This should help vertical mobility throughout the page. — Your local Sink Cat (The Sink). 06:02, 26 March 2025 (UTC)

What is the consensus about polling that is presented solely on podcasts/social media? I ask because the Herle Burley podcast https://www.youtube.com/watch?v=UuA5RwavSws&t=2062s has exclusive Pollara polling (March 19, L 40 C 36 N 12) and will be presenting daily tracking from that firm. That podcast has established operatives from the mainstream parties and is funded by mainstream companies like Telus and Enbridge, so it's not like it's a bunch of randos just spouting off whatever numbers. 2607:FA49:3E43:5900:2691:49BB:A441:5F58 (talk) 12:46, 26 March 2025 (UTC)

:If secondary sources like CBC/338 aren't incorporating it then no. Heartfox (talk) 14:46, 26 March 2025 (UTC)

::AFAIK, 338 hasn't included Pollara because they haven't released regionals with the data

::Pollara is up on https://canadianpolling.ca/Canada-2021 CFPoll (talk) 01:08, 30 March 2025 (UTC)

:::CBC isn't incorporating it (although Grenier has referenced the poll in his newsletter) https://newsinteractives.cbc.ca/elections/poll-tracker/canada/. Given the collaboration between the two I assume that if one of 338/CBC includes it, the other will, but that hasn't happened. Dingers5Days (talk) 18:32, 30 March 2025 (UTC)

::::I found the detailed poll results on [https://www.pollara.com/special_federal_election_2025/ Pollara's website] and added it to the table. Keep checking that same page for any further polls they may release. Cheers, Undermedia (talk) 23:45, 31 March 2025 (UTC)

:::::Oops, just realized the particular Pollara poll that was being discussed here isn't the same one I found. Nevertheless, I found one at that link, and others may eventually follow. Cheers, Undermedia (talk) 23:48, 31 March 2025 (UTC)

::::::338 is now including Pollara. But CBC isn't, yet... guess I was wrong. I don't have a problem with including it now though. Dingers5Days (talk) 01:04, 1 April 2025 (UTC)

For EKOS' last couple of polls, they have separately shown both the results of a full 5-day-long poll and the results of that poll from just the last 3 days. It seems like people are only adding the 5-day polls to this article. I agree that if we are only going to add one of the two, it should be the full 5-day samples, but what is the view on also including the 3-day polls? BCPoliGuy (talk) 01:09, 1 April 2025 (UTC)

:Tbh I do not think EKOS should be included at all, due to how wildly off of all other estimates their polls are. 65.130.1.23 (talk) 03:16, 1 April 2025 (UTC)

::Removing Ekos is not in the spirit of the article:

::{{Quote|This table provides a list of scientific, nationwide public opinion polls conducted from the 2021 Canadian federal election leading up to the 2025 Canadian federal election.}}

::This page isn't trying to forecast the election, it's just cataloguing the history of polling for this election cycle. Unless a poll has been evidenced to not be conducted in good faith or is not from a reputable organisation, it deserves to remain even if the results seem outlandish.

::Let's assume that when the election happens, the Ekos polls are found to have been unreliable. When people look back on this page 5 years from now, is it important that it documented Ekos' track record of poor polling? Or, if Ekos is actually correct, would it be important for people to see how unreliable the other polls were?

::Including Ekos in the list isn't going to affect the outcome of the election. 103.8.18.128 (talk) 03:57, 1 April 2025 (UTC)

::Why these discussions keep focusing on EKOS? I have not seen anyone discussing to remove Abacus or that new pollster Kolosowski despite being clear outliers right now, whereas Ekos' findings are now within the margin of error of others such as Pollara, MQO or Angus. Impru20^talk 07:18, 1 April 2025 (UTC)

:::Because it's a pro-LPC outlier, even though it's not really as extreme an outlier as it once was. The average Liberal lead is 5% in 338's aggregate, so Abacus showing a tie and EKOS showing the Libs up 12 are about equal in terms of deviation from the mean. There's no legitimate reason to exclude either of them. Dingers5Days (talk) 23:06, 2 April 2025 (UTC)

:I will address the actual point that was raised, not waste anymore time with the trolls who want censor and ban any information that displeases them. The problem with EKOS's 3-day roll-ups is that they fully overlap with the 5-day roll-ups, i.e. they're based on the same data, just truncated. They're not even like a rolling poll where there's a certain fraction of new data; they're fully redundant with the 5-day data. So including both the 5-day and 3-day roll-ups would be like adding the same data to the table twice, which would in turn get added to the graph twice and effectively give them undue weight in the trendlines. So having to choose between the two, it makes more sense to go with the 5-day roll-ups because they provide more total information, i.e. 5 days of polling data instead of 3. Cheers, Undermedia (talk) 11:16, 1 April 2025 (UTC)

::Seems fair! BCPoliGuy (talk) 15:23, 1 April 2025 (UTC)

Where is the Pollara poll from March 27th? It has liberals at 45 cons at 36 Decry9 (talk) 11:53, 1 April 2025 (UTC)

:If you're able to find a permanent publicly available source for it that includes all the major parties' numbers and the poll's basic methodological details (at minimum: field dates, sample size and polling method), then you can add it to the table. Otherwise, there's a 30 March Pollara poll that's been added, sourced from a proper report published on Pollara's website; as of now it seems to be only campaign-period Pollara poll that's published on their website. Cheers, Undermedia (talk) 14:10, 2 April 2025 (UTC)

Strange but the link is broken and the latest numbers aren't anywhere on the site, although they're up on Nik Nanos' personal substack. 2607:FA49:3E43:5900:2691:49BB:A441:5F58 (talk) 16:42, 3 April 2025 (UTC)

:It opens perfectly to me [https://nanos.co/wp-content/uploads/2025/04/2025-2783-ELXN-FED-2025-04-02-Field-Ended.pdf]. Impru20^talk 16:45, 3 April 2025 (UTC)

::Opens fine for me as well. isa.p (talk) 01:13, 4 April 2025 (UTC)

Why is the Kolosowski national poll from March 31 excluded? Bclark75 (talk) 03:15, 4 April 2025 (UTC)

:It's missing because no one has added it. :)

:Link the poll! isa.p (talk) 03:41, 4 April 2025 (UTC)

::There's a Kolosowski poll in the pre-campaign period table, with field dates 19–21 March. On Kolosowski's website, this same poll seems to have spawned 3 separate news releases on different topics, published on March 26, March 30 and March 31, respectively. Not sure if that's the poll you're referring to? Cheers, Undermedia (talk) 12:37, 4 April 2025 (UTC)

:I think Koslowski is being disingenuous claiming this isn't a partisan poll. He's an elected local official running on an "anti-woke" agenda. His website is suspect too. Doesn't pass the sniff test. Jawnbc (talk) 03:23, 5 April 2025 (UTC)

::LOL But liaison strategies run by former Liberal and NDP staff isn’t partisan, EKOS run by the man who calls Pierre a radical right wing racist and claims he would do anything to make sure Pierre doesn’t win, his poll isn’t partisan… give me a break 2605:B100:31B:9EB3:CC43:2103:5E80:1EB4 (talk) 15:47, 5 April 2025 (UTC)

:::Biased and partisan sources are allowed on WP. The mere fact that a source is partisan is not a good reason to exclude it. ComeAndHear (talk) 16:23, 5 April 2025 (UTC)

::::You seem confused. You're agreeing with the person you're replying to. The person you're replying to is claiming it's hypocritical to exclude a source because of partisan reasons but not other partisan sources 142.163.129.126 (talk) 17:58, 5 April 2025 (UTC)

:::::{{tqq|You're agreeing with the person you're replying to}}

:::::No, I don't agree with them, or at least with what they wrote. My point was only that biased sources are allowed on WP, contrary to the common misconception that they aren't. You can see this misconception being expressed in various forms in the discussions below, despite my comment. The person I replied to seemed as good of a person as any to reply to in order to clear up such misconceptions. ComeAndHear (talk) 23:18, 5 April 2025 (UTC)

:::Partisanship of pollster isn’t the issue. Angus Reid is a Tory, Nick Kouvalis (Campaign Research) is a conservative strategist, Frank Graves (Ekos) is openly anti-conservative, but all have a lot to lose by being wrong because they make a living by getting it right. An unknown “pollster” who also holds elected office and ran on an “anti-woke” platform may, by releasing polls that overstate conservative support, have much more to gain in his conservative political career than he has to lose in reputation in a fleeting foray into public research. Mmmuffins (talk) 22:23, 5 April 2025 (UTC)

:I really doubt the legitimacy of the Kolosowski poll, as its sources and methodology seem very suspicious. Also, neither the CBC Poll Tracker nor 338Canada considers it a valid poll. Therefore, I don't think including its results would add value to this page. Apsa 54 (talk) 05:08, 5 April 2025 (UTC)

::I do find it suspicious that the website is completely blank, there isn't any evidence of 'kolosowski strategies' polling for anyone, literally anyone at all...no work for any organization, business, media. But I don't think that's reason enough to exclude its polls, if you look at previous elections there have always been one or two firms that pop out of nowhere and their polls are included (for example a firm called 'delphi polling' in 2021 that doesn't even have a working website.) 2607:FA49:3E43:5900:2691:49BB:A441:5F58 (talk) 07:02, 5 April 2025 (UTC)

:::It is always odd when there is a new pollster, but precedent is we include polls regardless of the apparent or perceived partisan leanings of a pollster. Kiltarni (talk) 07:56, 5 April 2025 (UTC)

::::If you take a look at the numbers in the "Respondent Table" on the last page of the report, it's obviously an extremely skewed sample, with more than twice as many respondents (and 60% of the total sample!) who voted CPC in 2021 (870) as those who voted LPC (424); and only 28 respondents who voted NDP, versus 74 who voted PPC (NDP received >3.5x as many votes as PPC in 2021)! On the other hand, the fact that they actually share those numbers makes it seem like they're at least making an effort to be legitimate, if not particularly accurate... Cheers, Undermedia (talk) 11:48, 5 April 2025 (UTC)

:::::Another thing that's unusual is that nowhere in the report is there any breakdown of the results or the sample of respondents by gender or age. These are elemental demographic metrics that are found in absolutely every other poll. Undermedia (talk) 17:28, 5 April 2025 (UTC)

::::::Please take a look at my main comment on this page. I responded to one of your other comments. It's definitely fraud. Scryyyyyyyy (talk) 17:36, 5 April 2025 (UTC)

:::::The imbalance in who they polled shouldn't be an issue if they weighted it. Did they? If not, then there's clearly an issue with it (and the very weakness, is it's clear that it's not that Conservative support has collapsed since 2021; NDP support has collapsed and shifted Liberal - so significantly underpolling the NDP could be the difference here. But my comments are OR, of course. We need source, one way or another. Nfitz (talk) 18:42, 5 April 2025 (UTC)

::@Apsa 54 well said 2607:F2C0:F15D:FD60:8B50:F24B:E8D4:4FB0 (talk) 02:04, 6 April 2025 (UTC)

:Neither 338 or CBC has included it. If we’re going to be consistent on that, then I don’t think it’s worthy of inclusion. On a cursory glance I can’t find any other reliable sources (newspapers, media outlets etc) reporting on it either. Dingers5Days (talk) 14:15, 5 April 2025 (UTC)

::Tough call. On the other hand there's no 'smoking gun' I can see that conclusively demonstrates that it's not a legitimate poll (skewed samples are the norm, though to this degree I'm not sure), and the results aren't actually significantly different from the recent Innovative and Abacus polls. Maybe to keep the peace here, we consider that the Kolosowski polls balance out the EKOS polls and call it even? Cheers, Undermedia (talk) 15:06, 5 April 2025 (UTC)

:::https://www.whois.com/whois/kolosowski.ca

:::The domain was created less than a month ago.

:::His company website “reliahomes.ca” now redirects to his Polling site, but here is an archive of the old site before he tried to scrub it:

::: https://web.archive.org/web/20240119183307/https://www.reliahomes.ca/

::: https://ca.linkedin.com/company/reliahomes

::: https://web.archive.org/web/20250405040050/https://ca.linkedin.com/company/reliahomes

::: https://web.archive.org/web/20250405042446/https://ca.linkedin.com/company/robert-kolosowski

::: Dude literally still has the same phone number

::: https://www.kolosowski.ca/about

::: https://web.archive.org/web/20250405025115/https://www.kolosowski.ca/about

::: https://web.archive.org/web/20250405025117/https://www.kolosowski.ca/

::: > “Our founder: Robert Kolosowski. 10 years of experience in market research and statistics.

::: Previous work done in the banking, education, pest-control and software industries.

::: Experience governing budgets exceeding $1 billion per year.

::: Senior roles on 20+ election campaigns at the federal, provincial and municipal level.

::: Bachelor of Commerce in Economics and Management Science.”

::: The blatantly lying in his credentials. He graduated very recently. He has 10 years of experience in lying.

::: https://web.archive.org/web/20250405025117/https://www.kolosowski.ca/services

:::

::: https://www.robertfortrustee.ca/

::: https://tnc.news/2022/10/02/trustee-ontario1/

::: https://web.archive.org/web/20250405041540/https://tnc.news/2022/10/02/trustee-ontario1/

::: > “I believe our schools need to be a place where students learn to think critically and are able to discuss various perspectives on issues,” Kolosowski told True North. “Schools should not impose political views on students or teachers.”

::: > “I will focus on what unites us, not on what divides us. I will have zero tolerance (for) racism, identity politics and discrimination, and I will reject the divisive calls to remove O Canada from schools.”

::: > When asked about those wanting to reimpose mask mandates, Kolosowski said, “the public health science is clear: Covid-19 is not going to disappear, no matter how many times we close or restrict schools.”

:::I have way more evidence too, but it actually gets into details that shouldn't be on the internet. I've reported it to the CBC tips line. Google his name and look at what links you can find via Google images. It's insane. This poll needs to be deleted from this page. Scryyyyyyyy (talk) 17:09, 5 April 2025 (UTC)

::::If his credentials are being faked then that is a concern, that he has conservative or ant-woke or however you want to put it viewpoints is not disqualifying. On that point, it is the same argument made against Frank Graves and Ekos but in the opposite direction. Lilactree201 (talk) 20:54, 5 April 2025 (UTC)

:::::Look at the creation of the website. Less than a month ago. This is the first time this kid has released any polls. They have no track record. No proof of methodology. This is an early 20s red pilled kid faking polling data. There is zero credibility. Including this "poll" is an insult to the rest of the legitimate companies providing polls. Regardless of partisanship, this poll has zero validity and needs to be deleted. Do your own "whois" search or click on my first link. It's not a "concern". It is clearly, blatantly, indefensibly not credible. Scryyyyyyyy (talk) 21:44, 5 April 2025 (UTC)

::::::This is lilactree from an ip. I'm not denying you have a point on credibility. I'm only noting that we do not take partisanship or even explicit bias as reason for removing a poll, as was well established in the arguments over Ekos. 207.228.78.233 (talk) 23:16, 5 April 2025 (UTC)

:At this point no one has provided any substantive evidence that Kolosowski Strategies is a reliable source. Not being conclusively unreliable is not the standard for citation. Thus it should be removed per WP:INDISCRIMINATE "merely being true, or even verifiable, does not automatically make something suitable for inclusion in the encyclopedia", WP:V "Even if you are sure something is true, it must have been previously published in a reliable source before you can add it." No polling agregator includes it, the debate commission [https://www.debates-debats.ca/en/45/participation-criteria/party-leaders-that-meet-criteria/ did not incorporate it], it is [https://www.canadianresearchinsightscouncil.ca/member-directory/ not affiliated with the CRIC], etc. Heartfox (talk) 15:22, 5 April 2025 (UTC)

::Agreed. Wikipedia does not abide by the principle (cited by some) of "include everything to let people make the decision of reliability for themselves". This poll does not pass WP:RS so it should not be included, full stop. Dingers5Days (talk) 16:34, 5 April 2025 (UTC)

::Agreed. Biased sources are allowed on WP, but unreliable sources are not. How do we know that "Kolosowski Strategies" is a RS? What's preventing Joe Schmo from creating "Schmo Strategies" and releasing a "poll" showing CPC +5? How can we even verify that the poll was conducted, let alone that the respondents' responses were reported accurately?

::That said, I suspect the reason we're even having this conversation is that the poll in question showed an anomalous result of CPC +2. Had it shown LPC +3 no one would have batted an eye. Similar levels of caution should be taken with similarly obscure polling firms in order to avoid editorial bias. Just because sources are allowed to be biased doesn't mean that we are. ComeAndHear (talk) 17:02, 5 April 2025 (UTC)

::based on this and extensive discussion above, i think we should go ahead and remove the polls in question. isa.p (talk) 20:04, 5 April 2025 (UTC)

:I think it should be up to the readers to decide if it's credible, I say leave it in to give them the most information possible. It is in fact an opinion poll for the 2025 Canadian federal election. WatchfulRelic91 (talk) 16:18, 5 April 2025 (UTC)

:I think, Kolosowski must be deleted from the list, because another reputable aggregator doesn't include it on their list. Ekky1995 (talk) 16:42, 5 April 2025 (UTC)

:Clearly a fake poll from a fake pollster who seems to be nothing more than a Richmond Hill-based school board trustee (winning a by-election with 5% turnout) on a "parents' rights" claptrap platform. Please remove it. 209.128.12.6 (talk) 18:53, 5 April 2025 (UTC)

::Indeed, this 'poll' ought to be deleted. Dbrodbeck (talk) 19:39, 5 April 2025 (UTC)

:::Hilarious. All these proven fake polls supporting the Liberals are a okay, but ONE poll comes out in favour of the Conservatives and it's "dubious". Tell me there isn't a coordinated effort to shape public opinion in favour of the Liberals. You people think an EKOS poll showing Liberals at 50% run by a man quoted saying he will do everything in his power to stop Pierre Poullievre is legitimate; not to mention the bought and paid for Liberal polls coming from Mainstream Research and Liaison Strategies daily; are legit? This is a major attempt to undermine our democracy by the left. You people should be ashamed. 2607:FEA8:1E00:2CD0:3B38:3248:B461:F50E (talk) 20:39, 5 April 2025 (UTC)

::::Ekos has been polling longer than you've been alive, kiddo. As have many of the other firms (who are run by former Bloc or Tory operatives btw, funny you haven't mentioned that). Enough with your conspiracy theorism, you are contributing absolutely zero to this conversation. 2607:FA49:3E43:5900:2691:49BB:A441:5F58 (talk) 20:47, 5 April 2025 (UTC)

:::::it in no way matters what EKOS has done in the past, as I'm talking about what it's doing currently. There's no conspiracy theory here; not a single person believes the Liberals are leading at 50% for example. Look into the metric of that poll. Is what world is that not being flagged as "dubious" but this one is. Not to mention the arguments I'm reading here about this poll being thrown out because the pollster is "new". What kind of logic is that exactly? How can a pollster gain traction if it's polls are being thrown out simply because of that reason (which many are advocating for)? That's completely self-defeating - no new pollster would ever be admitted. The fact that you and people like you are keen to just ignore major questionable actions/metrics of Liberal leaning polls but immediately have this one removed from a wiki is dangerous. Please tell us all again that there isn't a bias in play. I bet you have a farm to sell me too. 2607:FEA8:1E00:2CD0:CCB7:F22F:A7:EE29 (talk) 02:15, 6 April 2025 (UTC)

::::::Keep babbling on about fake polls, kiddo. I'm sure it's doing a ton for your credibility. 2607:FA49:3E43:5900:DD5A:DB47:4CB:42BE (talk) 17:05, 6 April 2025 (UTC)

::::::Please read the entire discussion. The most credible arguments (i.e. ones that actually cite WP policies and guidelines) have to do with Kolosowski Strategies not being a reliable source, and neither is it cited by any reliable sources. As I wrote earlier, if a source is not reliable then we cannot even verify that the corresponding poll was conducted, let alone that the respondents' responses were reported accurately. If we allowed such polls, then there would be nothing preventing Joe Schmo of "Schmo Strategies" from adding a "poll" to this page that shows the Liberals leading at 60%. I'm sure you wouldn't like that, and neither would I.

::::::The other arguments you'll find in this discussion are specious at best. There are no WP policies or guidelines against "new" sources. In fact, all the sources on this page were at one point new. Should they therefore have been disqualified on that basis? Likewise, biased and partisan sources are allowed on WP, as are "outliers". The fact that a pollster has a Liberal, Conservative, NDP, or any other kind of bias, is not a good reason to exclude its polls. A pollster could come out tomorrow with a poll showing CPC +5, and we would not be justified in removing it simply on that basis.

::::::That said, if you find a poll on this page from a similarly obscure/unreliable polling firm (regardless of its bias, or lack thereof), feel free to mark it as dubious and bring it up for deletion in discussion. It's important to remain equally vigilant of all such sources in order to avoid editorial bias. As I wrote earlier, just because sources are allowed to be biased doesn't mean that we are. ComeAndHear (talk) 17:51, 6 April 2025 (UTC)

::::Wow, EKOS (and others like Mainstreet and Liaison) who have many many years reputation as pollster, suddenly called "fake" and "paid for Liberals polls" by NPC account 🤣 Ekky1995 (talk) 23:17, 5 April 2025 (UTC)

:I think it should definitely be removed. Not credible and doesn't meet criteria to be included. Isaacberman (talk) 21:19, 5 April 2025 (UTC)

::Support of cons at 41 is reasonable as it’s within their margin of error. If the poll was vastly weird I’d see the removal but its true we allow polls like EKOS even though no one seriously believes 50% of Canadians support the liberals (they haven’t polled like that in generations) 135.12.224.136 (talk) 22:20, 5 April 2025 (UTC)

:::If EKOS and other polls said "CPC will wins" in last 3 year, you still believe it.

:::Not vice versa 🤣🤣 Ekky1995 (talk) 23:12, 5 April 2025 (UTC)

:Goodness, what a mess. OK, let's try to straighten a few things out here:

:- A pollster being overtly partisan or having expressed political opinions is not a reason in itself to exclude their polls. If it was, I promise there would be very, very few polls remaining on this page.

:- A poll being an outlier—or even a pollster having a consistent bias in their results—relative to others are also not reasons in themselves for exclusion. Not to mention that I don't think either can really be said of Kolosowski, as their results aren't significantly different from those of recent polls by Innovative and Abacus.

:The above being said:

:- Kolosowski Strategies, as evidenced by a mere cursory look at their website, seemingly sprang up out of nowhere at the beginning of this election campaign, with the only releases listed on their website collectively generated from a total of 2 polls they seem to have ever conducted. This is moderately concerning, and not remotely the case for any other pollsters that appear on this page: all others have multi-year track records that are readily verifiable on their respective websites.

:- While Kolosowski deserve credit for sharing some data on the demographic composition of their respondents, they do reveal a very extreme skew towards past CPC voters (60% of the total unweighted sample of their latest poll) which—in conjunction with their ~2 weeks total experience as a pollster (see above)—objectively begin to make one wonder whether they know what they're doing when it comes to recruiting an adequate sample of respondents for a national poll.

:- Moreover, it's very unusual that no information is given on the gender and age breakdowns of their respondents. These are elemental demographic metrics that are found in absolutely every other poll on this page. What's more, Kolosowski's polls state that they are only "weighted by 2021 party support and region", not gender or age. Again, very unusual and definitely concerning.

:Hopefully this will help refocus the discussion on the relevant points. Cheers, Undermedia (talk) 23:48, 5 April 2025 (UTC)

::Still no one has put forth any argument as to why Kolowoski Strategies is a reliable source. It should be removed per WP:BURDEN: "The burden to demonstrate verifiability lies with the editor who adds or restores material, and it is satisfied by providing an inline citation to a reliable source." Heartfox (talk) 00:07, 6 April 2025 (UTC)

:::Yes, this seems to be a relevant point. What exactly would it take to demonstrate that Kolosowski is a reliable source, just so we're clear on this? Undermedia (talk) 00:18, 6 April 2025 (UTC)

::::Not op on this tread, but imo, same way 338 and every other aggregator does it. Evaluate the pollster's proven track record over time. It is the years of predictions compared with actual results that allows us to confirm. It's not just "does the data look right", it's, "how has this pollster performed over time?"

::::Kolosowski Strategies simply cannot be considered valid due to it's lack of track record and history. No one in their right mind says, "Oh I'm starting a polling company. Our first poll ever could be on market research for a product. We could slowly build up our credentials by showing a track record of getting it right. But nah, first poll is on the Federal election showing an outlier, babyyy."

::::Absurd to even consider a platforming a no name poll. I could make a website right now and send out emails/texts with a simple data stratification and weighting strategy like gets taught in a 3rd year uni course. That doesn't mean my results should go next to Angus Reid!!!!! Scryyyyyyyy (talk) 00:31, 6 April 2025 (UTC)

:::::I suggest we keep it concise and dispassionate otherwise this will turn right back into a theatrical mess of a debate. I think @Heartfox is on the right track by focusing strictly on Wikipedia rules, and I believe the crux of your argument similarly boils down to WP:RELIABILITY. So let's focus on that, and I reiterate my question: what exactly would it take to demonstrate that Kolosowski Strategies is a reliable source, as per Wikipedia rules? Personally I think inclusion in another poll aggregator like 338 or CBC may be an excessively high bar, but what about simply a reference to their polls by an established Canadian news/media outlet? Can anyone even provide an example of that? If not, then I think there's a pretty cut-and-dry case for excluding them from this page. Cheers, Undermedia (talk) 00:55, 6 April 2025 (UTC)

::::::I respect your position. I will agree to the referenced by an established Canadian news/media outlet. Although I will stipulate that it be a CTV/Global/Star/National Post/CBC or equivalent level. Plenty of small "outlets" will publish anything for clicks.

::::::If that is the position we take, I would also stipulate that the poll should be removed until a reference can be provided. Scryyyyyyyy (talk) 01:02, 6 April 2025 (UTC)

::::::I would agree it doesn't necessarily need to be included by 338 or CBC. We don't want to not include a poll just because the org is new, but [https://www.google.com/search?q=%22kolosowski+strategies%22 Google] shows no coverage by other sources and at this point it seems more unreliable than reliable. I am seeing a consensus here that it should be removed. Of course, if new information becomes available it can be added back but right now I think it is best to omit and remove from the tables and graphs. Heartfox (talk) 01:04, 6 April 2025 (UTC)

:::::::Agreed. I have not yet seen any proponents of the Kolosowski polls provide a single example of Kolosowski's work being referenced by an established/reputable media outlet. Unless and until they do—and given the other concerning aspects of this pollster and their polls that I summarized above—it seems reasonable to remove them from this page. Cheers, Undermedia (talk) 01:19, 6 April 2025 (UTC)

::::::::Kolosowski Strategies did some work for CBC News conducting regional level polling. They were cited on national television today. Here is the [https://www.cbc.ca/player/play/video/9.6742216 link] (1:30). Now that the election is over and the firm has done work for national mainstream media outlets, I don't see any reason to exclude their data. 172.97.164.92 (talk) 04:03, 30 April 2025 (UTC)

::Ultimately it's a decision of what this page is about. If anyone can create a website out of thin air within a month of an election we know is going to be pumped with misinformation and create polls that are going to be represented alongside established pollsters, we are a part of the problem. Taking no action undermines the trust in polls.

::It's simply unacceptable to platform an untrusted, untested, unverifiable "pollster" in 2025. This Wikipedia page impacts voters, their conversations, and voter turnout. If "Kolosowski strategies" is around next election and proves it's track record, then fine. But we cannot accept platforming a "company" made the week before the election was called with troubling data and not look at it for what it is: Intentional Misinformation. Take it down. Scryyyyyyyy (talk) 00:12, 6 April 2025 (UTC)

:::Like seriously. "Kolosowski strategies" isn't even an incorporated company. The domain is owned by his Pest Control LLC. Scroll down to the organization in "who is"

::: https://www.whois.com/whois/kolosowski.ca

:::"Reliahomes Canada Inc"

:::This is not a real pollster. Blatant misinformation attempt. It is terrifying to me that a case this obvious requires such litigation. Scryyyyyyyy (talk) 00:47, 6 April 2025 (UTC)

The only reason that is given for the poll being "dubious" is the political leanings of those that object to it.

If we include the likes of EKOS in the polling, then we have no choice but to accept this one as well.

It would be best for some of the objectors to seek professional help before they inject their own personal opinions into this page.

End of Discussion. 142.114.253.46 (talk) 21:22, 5 April 2025 (UTC)

:No need to start a new thread. And you've clearly not read all the objections. Seems like you're the one interjecting your personal opinion with the your haughty tone and advise to "seek professional help". 2607:FA49:3E43:5900:2691:49BB:A441:5F58 (talk) 21:32, 5 April 2025 (UTC)

:The reason that is given is that no reliable source includes it in their aggregate or reports on it. Abacus and Innovative show a close race/Tory leads, but they are not removed because reliable sources include them. Dingers5Days (talk) 22:40, 5 April 2025 (UTC)

:There's also prior precedence in the last few American presidential elections, where a lot of new polling firms were springing up and appearing in the applicable articles for those elections, leading to them being cut out of concern that they were trying to effectively use Wikipedia for free advertising. Some of them dropped off the face of the earth within a few months while others, for instance, didn't meet the criteria for being considered a reliable source in 2016, but did by the time of the 2020 election.--DaveJB (talk) 11:43, 6 April 2025 (UTC)

:There is many reasons listed above other than political leadings to object to the poll. I suggest you reread it. And look at the bits of the discussion about other pollsters that are included, despite well known pro-right and pro-left biases. Nfitz (talk) 19:25, 6 April 2025 (UTC)

We’ve recently removed all Kolosowski polling due to concerns including unverifiable methodology, lack of media presence, and overall reliability. Similar concerns apply to MQO. MQO is also not a known or previously established firm,it hasn’t been present in any other election and its methodology is difficult to verify. Notably, their sample weighting raises issues: they classify respondents into “Urban,” “Suburban,” and “Rural,” assigning only 17.8% to the rural category. While this may reflect Statistics Canada’s strict rural definition (<400/km², <1,000 population), it significantly underrepresents rural and small-town voters when contrasted with “suburban,” likely inflating Liberal support. Given these concerns, MQO should be treated the same as Kolosowski and excluded.

Let me know what you think Burns753 (talk) 20:57, 7 April 2025 (UTC)

:The secondary sources like CBC/338 are incorporating it, then it should be shown. 75.157.242.111 (talk) 21:07, 7 April 2025 (UTC)

::I think the issue with Kolosowski had more to do it was polling out of the back room of his pest control company than his polling methods. I would think that MKO (despite their possible biases) have a much more legitimate profile to be a reliable pollster. I think it is fine to include them, their removal would not be in the same category to me. Words in the Wind(talk) 21:09, 7 April 2025 (UTC)

:::338 Canada has included the MQO poll and classifies it at B- rank probably because the company is relatively new and does not yet have a track record for assessing the quality of their polling. The company is based in Halifax NS and conducted polling for the NS provincial election in 2024 and the Halifax Regional Municipality election in 2024. Their polls in these cases were informative and proved to be reliable. I think this MQO federal election poll meets the minimum standard for inclusion in the Wikipedia record. Methacrylate (talk) 21:25, 7 April 2025 (UTC)

:::I don’t think that because CBC is incorporating it means it’s valid. Right now they would have obvious reasons to be biased. Also, it’s not as if Kolonowski is a twitter account just writing out numbers - they have a full website and charts. Both companies aren’t recognized by any media, have suddenly appeared during the campaign, and have no possible reason of verifying their reliability. For me it seems that if we exclude Kolonowski but keep MQO, it’s for personal biases rather than being factual and consistent. In addition I have explained why the way that they weigh their polls help the liberals. Burns753 (talk) 21:26, 7 April 2025 (UTC)

::::I don't think it matters whether a poll supports the Liberals or the Conservatives; the only thing that matters is credibility. Even your reasoning for the Liberals’ polling advantage in the MQO poll isn't valid, as their approval rating is 44%, which is quite normal compared to other polls. 75.157.242.111 (talk) 21:38, 7 April 2025 (UTC)

::::See WP:BIASED here; sources with an ideological slant are allowed. As stated before, there's consistency in the polls currently included: they are all in the aggregates of and reported on by reliable sources, such as 338, CBC, and major newspapers (National Post, Toronto Star, etc). MQO is included in 338 and CBC's aggregates, is on the list for the debate commission's qualification criteria, and their past polls have been cited in sites considered WP:RS: [https://www.cbc.ca/news/canada/nova-scotia/public-engagement-apathy-municipal-election-halifax-1.7353328] [https://www.saltwire.com/prince-edward-island/support-for-three-main-pei-political-parties-too-close-to-call-mqo-231745] [https://halifax.citynews.ca/2024/11/22/new-poll-shows-houstons-pcs-maintaining-big-lead-over-ndp-and-liberals/] Meanwhile nobody has been able to find a reliable source mentioning Kolosowski Strategies; everything you see is from Twitter/X. I think we're going in circles on these questions of "why is [insert pollster here] allowed??" when the answer is not at all difficult to understand. Dingers5Days (talk) 15:45, 8 April 2025 (UTC)

:MQO is a real company that has been polling for 35 years per their website. They have actual employees and are a member of the Canadian Research Insights Council (CRIC), which most pollsters belong to. THey've polled previous municipal and numerous provincial elections. They fit all the criteria. 2607:FA49:3E43:5900:2691:49BB:A441:5F58 (talk) 21:45, 7 April 2025 (UTC)

• I'm not sure I agree with removing Kolosowski - and this one seems even more clear cut. I don't see the same issues. And the results are certainly similar to more than one poll in the last week (though even Kolosowski's were with the within the error limits). Nfitz (talk) 22:14, 7 April 2025 (UTC)

:On MQO's website, there are nearly 100 press releases stretching back to 2017—they have polled in numerous provincial and municipal elections. On Kolosowski's website, there are 5 releases stretching back two and a half weeks. One of them is a 'word cloud'. This discussion is a waste of time. Undermedia (talk) 03:37, 8 April 2025 (UTC)

:Hi all first off let me state a conflict in that where I work is owned by the same parent company as MQO. That said, the reason people don't know "MQO" is is was created by a merger of Media Quest and Omnifacts some time ago (you've probably heard of at least the latter), then rebranded as MQO. Just wanted to share that, I think omnifacts must have been 40+ years old? WayeMason (talk) 17:41, 9 April 2025 (UTC)

Many people have already mentioned the famous [https://338canada.com/federal.htm 338Canada], which serves as a sort of aggregator of all the polls included in the article. Should we include it in the article? Daminb 15:05, 12 April 2025 (UTC)

:To me it seems redundant because this page is also basically an aggregator of all the polls; and the graph's trendlines are basically the equivalent of polling averages, though I would personally argue they're more straightforward than 338's averages, which additionally factor in pollster 'ratings' that are likely to be controversial among editors here if the above debates are any indication. Also, there are other poll aggregators: if we add 338, would we also need to consider adding CBC's poll tracker as well as The Star's? There are probably even one or two others beyond those. Overall, I just don't see much if any added value, and at the same time it would likely generate yet more debates on the Talk page, as if we didn't have enough already. Cheers, Undermedia (talk) 18:18, 12 April 2025 (UTC)

Just as an FYI, a sample size is required to add a poll to the graph, otherwise it returns an error. So for example, there's nothing I can do at the moment with that new Abacus poll ending 10 April unless/until a more detailed source is available that indicates its sample size. Cheers, Undermedia (talk) 18:23, 12 April 2025 (UTC)

:Understood. The Toronto Star article that contained the info on the latest Abacus poll did not say anything about sample size (but did mention the MoE). Dingers5Days (talk) 19:22, 12 April 2025 (UTC)

::Chances are Abacus puts out a release on its website tomorrow, as it has every Sunday so far in the campaign. Undermedia (talk) 20:31, 12 April 2025 (UTC)

I think there is an error in what was listed. The attached file to the entry says the Conservatives have 39% and the Liberals 43%. This is in contrast to what is on the table where the gap is 1 point. 2001:56A:F202:FD00:8D05:F47D:CBD1:801 (talk) 05:56, 13 April 2025 (UTC)

:Hello! As you might have noticed, I was the one who added the poll (and I am quite new to this). It is pretty likely I made a mistake, but I would be glad if someone pointed out where it is, since the last graph shows the CPC at 43% and the LPC at 44%. Where should I look? Daminb 06:16, 13 April 2025 (UTC)

:UPDATE: I believe this might have been a misunderstanding. I added in the numbers and references as separate edits, so at one point the file did not match the numbers. Daminb 06:22, 13 April 2025 (UTC)

::From what I can tell it looks right. Scary id its only a 1 point lead but of course it is what it is. Welcome! Kflack (talk) 06:25, 13 April 2025 (UTC)

:The PDF also does not have any of the data broken down by subsections, it is blank. Not sure if that is an issue for us or not, but it does not look great for it to have that. Racsan24 (talk) 15:46, 13 April 2025 (UTC)

::Hey there. Looking back at [https://cdn.prod.website-files.com/66c8dfb086a015b3b519e988/67f9d77b69bb660b6e0149e3_2025-04-12_CAN_Daily_Tracker_Public.pdf previous] [https://cdn.prod.website-files.com/66c8dfb086a015b3b519e988/67f854633aec0c5da2bdd0d8_2025-04-11_CAN_Daily_Tracker_Public.pdf Mainstreet] [https://cdn.prod.website-files.com/66c8dfb086a015b3b519e988/67f70e463af1fb08a98de15c_2025-04-10_CAN_Daily_Tracker_Public.pdf polls], it just seems to be the way they are doing it. Despite this, polling websites still include them, I think. They still list the relevant categories (sex, age, region, race, education) for weighting, but those categories have not been broken down by party (I presume, to save time and effort). It is probably OK. Daminb 16:08, 13 April 2025 (UTC)

::The data from the blank sections of Maintreet's 'public' releases are exclusively available to paying subscribers. Not an issue as concerns listing their polls here. Cheers, Undermedia (talk) 16:23, 13 April 2025 (UTC)

Hello

I was about to remove this but wanted to ask here first

Canada338 doesnt list this nor CBC poll tracker. The Star article it links to I read the whole discussion and they dont mention a 4 point gap... anyone? Kflack (talk) 06:19, 13 April 2025 (UTC)

:Abacus has just uploaded it [https://abacusdata.ca/2025-federal-election-poll-liberals-lead-by-4/ here]. Daminb 10:03, 13 April 2025 (UTC)

:It was an unofficial release until yesterday 170.203.207.243 (talk) 13:28, 14 April 2025 (UTC)

{{Edit semi-protected|Opinion polling for the 2025 Canadian federal election|answered=yes}}

REQUEST: There is an inaccurate data point in the graph. The 44% Conservative Data point from the Mainstreet Poll is inaccurately plotted in the April 12th Date, and should be in the April 13th Date. Thank-you. Jryantha (talk) 17:59, 14 April 2025 (UTC)

:Actually, looking closer at this, it appears that the Conservative 43% from April 12 Mainstreet was likely plotted correctly, but that the 44% from the April 13 Mainstreet has not been plotted at all! Who can fix this? 97.109.168.36 (talk) 01:42, 15 April 2025 (UTC)

::It is plotted, but is covered by the red 44% from the Liaison poll. All the polls in the table are always plotted. Cheers, Undermedia (talk) 03:28, 15 April 2025 (UTC)

:::I appreciate the attempted explanation, however I'm not sure if this is correct. For example, if one looks at the Abacus poll from Apr 3 where both parties scored 39%, the dot is somewhat purple (red predominating, slightly transparent, but thin blue halo around it). In the case being discussed, the 44 dot on Apr 12 shows no evidence of blue whatsoever. Is there a better way to depict this if overlapping which is reliable and distinctive? 97.109.168.36 (talk) 12:35, 15 April 2025 (UTC)

::::I manage the graph so you're just going to have to take my word for it: it's there and is being duly factored into the trendline. Every time the graph is updated it automatically grabs all the polling data in the table; no polls are excluded. Sometimes different parties' dots overlap, making it difficult to discern the one underneath; it is what it is. Cheers, Undermedia (talk) 13:29, 15 April 2025 (UTC)

:File:Semi-protection-unlocked.svg Not done: According to the page's protection level you should be able to edit the page yourself. If you seem to be unable to, please reopen the request with further details. Shadow311 (talk) 17:51, 29 April 2025 (UTC)

Hey there, these questions are specifically for Undermedia. I just have a couple questions about your graph, basic stuff like what functions you use, and why you used them.

1. You mention that this is a 30-poll local regression, is there a reason why you went for a certain number instead of something more standard like using a nearest-neighbor of 0.25%?

2. Wouldn't using an exact amount of polls result in bias from old data, negating some of the precision of local regression?

3. Can you say what weights you are using for proximity in time. I'm assuming its tricube, although that's an assumption, so I just wanted to make sure.

4. Are you taking into consideration the Margin of Error within your weight? Using a Inverse Error Function?

Thanks for your time, and thanks for all the hard work you put into this wikipedia page. Been watching this page for a while now, just know it doesn't go unseen. Thought I would finally ask ;) Cheers, IonicFusion (talk) 19:52, 14 April 2025 (UTC)

:Hi. Thank you for your interest and I appreciate the kind words. I don't mean to be rude, but as a full-time worker + parent of 2 young children who barely has enough spare time to contribute to this page and manage the graph as a volunteer service, I'm afraid I simply don't have the time at the moment to provide detailed answers to your questions and potentially embark on an ensuing technical discussion. FYI, the graph is generated in the free version of [https://posit.co/products/open-source/rstudio/ R Studio] using the [https://ggplot2.tidyverse.org ggplot2] package—perhaps some of the answers to your questions can be found in the latter's reference docs. Otherwise, I can assure you it has been carefully/incrementally tweaked and refined over the course of nearly 10 years into its current configuration and presentation, indeed taking into account some of the points you alluded to. Sorry for what is likely a disappointing response, but I figure it's at least better than no response at all to your absolutely fair and courteous inquiry! Cheers, Undermedia (talk) 16:12, 15 April 2025 (UTC)

::I get it. I also have a full-time job and way too many side projects. I only ever opened the talk page when I saw EKOS (and now Mainstreet) clearly skewing their data.

::I'm currently working on a project [https://pollsteraudit.ca/ pollsteraudit.ca], with the goal of prevent this in the future.

::I'm asking about the graph out of curiosity, to see how others have implemented theirs compared to mine.

::With that in mind, if you would like me to automate the graph for you. I would happily do so, since I have already written a full API for it. IonicFusion (talk) 16:26, 15 April 2025 (UTC)

:This article is not trying to predict the election result, it is simply a historical record of opinion polling. As such, the graph shouldn't necessarily be trying to estimate the true levels of public support for a given party, but rather to show the trend the opinion polls followed, whether accurate or not. Nor should it speculate on political bias of a certain pollster outside of confirmed instances of manipulation.

:In other words, your site and this article have different goals, and so it is not necessarily appropriate to use the same graph. 103.8.18.128 (talk) 22:25, 23 April 2025 (UTC)

Juno News has been partnering with One Persuasion and doing polling weekly. The results are not outliers and the company is respected. Should these not be part of this article? 170.203.205.216 (talk) 14:23, 17 April 2025 (UTC)

:Are there adequately detailed individual reports/releases for each of these polls that could be linked to in the poll table? At a glance I wasn't able to find any on One Persuasion's website... Cheers, Undermedia (talk) 14:57, 17 April 2025 (UTC)

:Juno News should be deprecated. ―"Ghost of Dan Gurney" _(hihi) 15:07, 17 April 2025 (UTC)

:Not included on 338 or CBC, not reported on by reliable sources. Doesn't meet the standards. Dingers5Days (talk) 18:00, 17 April 2025 (UTC)

Are we going to be posting both the 3 and 5 day rollups going forward? In the past we have only taken one from each release. Racsan24 (talk) 23:13, 17 April 2025 (UTC)

:The new EKOS release contains two separate polls: (1) an IVR poll that is variously shown in the release as a 5-day roll-up (12–16 April) and a 3-day roll-up (14–16 April); (2) a telephone/online poll conducted from 14–16 April (i.e. same field dates as the IVR 3-day roll-up). The IVR 5-day roll-up and the separate telephone/online poll have been added to the table, which we have indeed done for some previous EKOS releases where they conducted two concurrent polls using the two different methodologies. Cheers, Undermedia (talk) 00:05, 18 April 2025 (UTC)

There's an editor insisting that this poll be labelled in the Polling Firm column as "YouGov (MRP)", even though "MRP" (which stands for "multilevel regression with poststratification") is not part of the name of the polling firm, but rather a statistical modelling method employed in the poll. I would like to point out that all pollsters employ some form of statistical modelling or weighting, so I think it needs to be better justified why this one in particular is so special that it merits a parenthesis next to the name of the polling firm; rather than simply invoking precedent on a few other countries' election polling pages. Moreover, it's not actually that clear in the linked report whether this is a 'conventional' poll or something different in nature, which perhaps shouldn't be listed among the other conventional polls on this page. YouGov is a well-established pollster, but strangely there are basic methodological details missing, such as the medium through which the poll was conducted (e.g. telephone, online); and regarding the sample size, whereas in one part it alludes to "modelled responses of 6,077 adults in Canada" (What does "modelled responses" even mean? At face value they don't really sound like 'real' responses...), in another part it vaguely states "based on over 20,000 interviews with Canadians up and down the country". So all in all, I'm not too sure what to make of this 'poll', and I note that as of now neither CBC Poll Tracker nor 338Canada has included it. Thoughts? Cheers, Undermedia (talk) 16:00, 23 April 2025 (UTC)

:Speaking as an unprofessional bystander: I think I agree with you on excluding "MRP" from the name of the firm, as it does not belong there. However, I think we should see how the poll situation develops and if it is going to be included in the aggregator's lists.

:Also, since I am pretty much illiterate in poll knowledge, are your concerns with the poll related to shady methods or just weird ones? I think it is important to know for the poll's inclusion. Daminb 17:01, 23 April 2025 (UTC)

:Just speaking of precedent here: The 2024 UK election polling Wiki page does include YouGov's MRPs under both the "national poll results" and "seat projections" tabs. I think it should be treated the same way, so long as the methodology for those MRPs were similar to this one. Dingers5Days (talk) 17:17, 23 April 2025 (UTC)

:yeah I wouldn't include the YouGov thing at all... clicked through and there's no MoE included in the article and those seat projections are absurd (it's not even the numbers being completely impossible it's just there's a bunch of seats that are given absolutely insane projections) I don't have much experience with YouGov as a pollster but that one poll looks like it was poorly run and it seems impossible to check their methodology to figure out how good it might be EndMaster0 (talk) 23:03, 23 April 2025 (UTC)

I just spent some minutes to read through YouGov source [https://yougov.co.uk/politics/articles/52054-first-yougov-mrp-2025-canadian-federal-election-shows-liberals-on-track-to-win-modest-majority-poll article], this sentence catch my interest, "In terms of vote share, we anticipate that the Liberals will win around 42% of the vote, the Conservatives 38%, the NDP 10%, the Bloc 6%, the Greens 2%, the Peoples Party 2%, and other parties 1%." I feel that it looks like those numbers are based on their model prediction, not an actual opinion poll number. In addtion, like Undermedia said, no MOE or interview method mentioned, the so called 'sample size' looks also a little bit unusual.

So I think we could keep the Seat projection section, but remove it from the Poll list.

what do you think? :) Haers6120 (talk) 22:53, 23 April 2025 (UTC)

:I wouldn't keep the seat projection section either... the numbers aren't too far off what TheWrit is predicting but if you actually look at the map many of the ridings are just completely different from the predictions 338 does or from the occasional riding level poll that exists. Mississauga-Lakeshore is probably the worst, YouGov predicts it goes conservative by 1%, 338 predicts it going Liberal by 9%, and riding polls predict Liberal again by 17-29%... on the other side of the country YouGov is predicting Cowichan-Malahat-Langford going to the NDP by 3%, while both 338 and riding polls predict it going to the conservatives (both by 7%) EndMaster0 (talk) 23:14, 23 April 2025 (UTC)

::Those are very small differences considering individual riding polls. A poll that is different by 20% could be suspicious. Neverthess, we do NOT exclude polls just because they are outliers. 70.70.2.105 (talk) 00:00, 24 April 2025 (UTC)

{{edit semi-protected|Opinion polling for the 2025 Canadian federal election|answered=yes}}

mainstreet poll for date 25-4-25

Liberals 43

Conservatives 41

NDP 7

Bloc 6

Green 1

PPC 1

Other 1 66.49.148.53 (talk) 20:52, 25 April 2025 (UTC)

:{{done}}: The poll has been added. Daminb 02:52, 26 April 2025 (UTC)

The first YouGov poll was ran from April 2nd to April 22nd the most recent one was ran from April 6th to April 25th. That's 16 days of overlap on polling durations of 20 and 19 days respectively. The two results are almost certainly sharing data potentially a majority of the data they use. Not sure what the precedent might be in this situation but at a minimum there should be something to let people know right away that those two polls are so strongly linked.

I have a similar issue with the most recent triplet of Innovative Research polls given they use data from the same day of polling in two different polls twice and there's no indication on their end on how the responses got sorted on those two days EndMaster0 (talk) 20:58, 25 April 2025 (UTC)

:I do agree with you. YouGov titles their most recent MRP as an "update" which indicates an overlap in data. Like you I'm not sure what the answer is. We could put a note indicating the overlap, or maybe just keep the final MRP? Dingers5Days (talk) 21:31, 25 April 2025 (UTC)

::I added the appropriate (non-)overlap fraction to the YouGov poll's sample size, which has in turn reduced its weight in the graph. However, the Innovative polls are quite a bit dicier to deal with in that way. We do see in their breakdown of interviews per day that the polls mostly don't overlap with one another though, so the issue isn't nearly as significant for weighting as two 6000-respondent polls 3 days apart that overlap by 85%. Cheers, Undermedia (talk) 13:15, 26 April 2025 (UTC)

:::Actually, the 2nd Innovative poll has ~70% overlap with the 1st, so I've added a fraction to reflect that; whereas overlap between the 3rd and 2nd is negligible (~3%), so no sense adding fraction to the sample size of the 3rd. Undermedia (talk) 14:34, 26 April 2025 (UTC)

::::If we're going to be consistent about this though, there are also a few other technically non-rolling polls throughout the campaign with overlapping field dates. For example, Abacus' 27 March and 10 April polls both overlap with their respective previous polls by 50%, and today's Leger poll overlaps with the previous one by 20%. As for EKOS, their overlapping IVR polls already have sample size adjustments, but then there's the question of their separate telephone/online polls that were conducted concurrently with a couple of their IVR polls... Undermedia (talk) 14:53, 26 April 2025 (UTC)

For example, FocalData ? Canadianpoliticaljunkie (talk) 16:06, 28 April 2025 (UTC)

:Link goes to a Google doc... hmmmm 🧐 Canadianpoliticaljunkie (talk) 16:07, 28 April 2025 (UTC)

::There's a more formal link here from their site: [https://www.focaldata.com/blog/trump-effect-may-cost-conservatives-victory-in-todays-canadian-election]. And looks like we posted new sections at about the same time, oops Dingers5Days (talk) 16:10, 28 April 2025 (UTC)

I know we're at election day and the polling period is over, but I want to make sure there is agreement on the inclusion of the Focaldata poll. It seems that POLITICO commissioned the poll: [https://www.politico.com/news/2025/04/26/canada-election-poll-favorability-00311632] USA Today also reported on it: [https://www.usatoday.com/story/news/world/2025/04/28/canada-election-trump-carney-poilievre-tariffs/83298428007/] It's a British firm that is included in the averages when they publish polls there. However, 338 and CBC do not include it in their averages. I think it should be included but would like to hear other arguments. Dingers5Days (talk) 16:07, 28 April 2025 (UTC)

:Fair enough, I'm open to it. I'd just never heard of it before. Canadianpoliticaljunkie (talk) 16:13, 28 April 2025 (UTC)

::No issue with its inclusion. OhanaUnited^{Talk page} 14:09, 29 April 2025 (UTC)

For future elections, do we want to set some policy around the inclusion of polls that only appear on the last day and have no track record, even if they are reputable pollsters in other places? Canadianpoliticaljunkie (talk) 14:17, 29 April 2025 (UTC)

:I see no compelling reason to exclude them, unless you have one to offer? Cheers, Undermedia (talk) 15:56, 29 April 2025 (UTC)

::Even pollsters who have been around for a while might have an interest in changing the narrative with a last minute poll (in either direction; I'm trying to not even be partisan about it. I've seen all sides try stuff like this.) Canadianpoliticaljunkie (talk) 16:03, 29 April 2025 (UTC)

:::Possible, though I see no evidence of such attempted last-minute narrative-changing this election. Are you talking about the Focaldata poll specifically? Their results are very similar to Abacus Data's and Innovative Research's final polls, and overall predicted a narrow LPC pop vote victory like everyone else. Same applies to the Forum poll, which was even more aligned with the 'averages'. Undermedia (talk) 18:21, 29 April 2025 (UTC)