Talk:Opinion polling for the 45th Canadian federal election

National polls table

Does this really need to be mapped out to 2025 at this time? We don't know when the election will be, it could be earlier than that. It also reduces the usefulness of the table for now, by making the recent information difficult to read.--Darryl Kerrigan (talk) 00:22, 27 January 2022 (UTC)

Do you mean the graph? - Ahunt (talk) 00:23, 27 January 2022 (UTC)

Yes.--Darryl Kerrigan (talk) 00:29, 27 January 2022 (UTC)

Ah, okay then we need to consult with @Undermedia:. - Ahunt (talk) 00:31, 27 January 2022 (UTC)

Yes, I tend to think this should be mapped out to the end of the year, and then expanded into 2023 next year, into 2024 that year, and so on until an election. I don't think we need all that blank space until an "election" date of 2025-10-20. Of course, an election could occur well before that date. Historically, minority governments tend to last about two years on average.--Darryl Kerrigan (talk) 00:37, 27 January 2022 (UTC)

That approach probably makes sense, but let's see if there are any technical considerations. - Ahunt (talk) 00:45, 27 January 2022 (UTC)

Hi folks. Thanks for the suggestion. I'll see what I can do. Cheers, Undermedia (talk) 18:11, 27 January 2022 (UTC)

Notable events

Further explaining this revert of mine, the NDP's agreement to give confidence and supply to the Liberals will have a major effect not only on Canada's future governance, but (likely) also voter intentions (e.g. No longer considering strategic voting). CentreLeftRight ✉ 21:56, 22 March 2022 (UTC)

Agreed. This is a significant event, which likely should be noted.--Darryl Kerrigan (talk) 21:57, 22 March 2022 (UTC)

Oppose I think we should not include this event based on a few reasons. One it is slippery slope as many events could influence voter intentions, The Freedom Convoy, various Covid-19 lockdowns (or not), and other events have all provided various shifts in opinion trends. It is not our job as Wiki editors to opine or provide opinions in the chart. I agree this is a significant event, but there are also arguably other events that should be included as well, which could lead to further debate. My understanding is both now and in the past we have only included changes in leadership as this actually changes the questions that are asked by the various polling companies, keeping the list simple and clean. I feel we do not need to clutter up the charts with other notable events, no matter how noteworthy they are. Happy with whatever the group chooses though. Words in the Wind(talk) 22:25, 22 March 2022 (UTC)

I hear your point about slippery slopes. There are many events that are notable (like the ones you mention) that would not be appropriate to add. The following are appropriate though: election dates, changes of government (like happened in BC between elections Clark to Horgan), leadership changes, formation of coalition governments, supply and confidence agreements, presumably a formal declaration of war. Other things that don't directly affect how Parliament functions generally should not be included, but I think this "supply and confidence agreement" fits the bill. I note that the signing of the BC "supply and confidence agreement was not included in the polling there, but I assume that was because it happened shortly after the election (which is noted) and since the confidence vote, Clarks resignation, Horgan's appointment, and then Clark resigning as party leader were all included in rapid succession, further mention might have been a bit much.--Darryl Kerrigan (talk) 23:13, 22 March 2022 (UTC)

I disagree that this is a "slippery slope" because the difference between the confidence and supply agreement and the Freedom Convoy (for example) is that the former directly affects how parliament functions and whether or not a new election will be triggered early, while the latter does not. CentreLeftRight ✉ 00:12, 23 March 2022 (UTC)

Agree The formation of a coalition or confidence and supply arrangement is more similar to a political party forming, or a change in leadership (both events that have been included in the past), than it is to other types of political events. CASalt (talk) 03:35, 23 March 2022 (UTC)

Mainstreet poll 16 June 2022

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: 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)

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 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)

New referencing system for polls so inconsistent that I'm about to give up

Hi all. I was about to update the numbers in the latest EKOS poll to those in the now-available 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 "<ref>" at the beginning and the "<ref>" 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 "<ref name="auto#">" instead of just "<ref>", 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)

9 Dec 2022 Nanos poll

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 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)

Seat projections

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)

JasonHRoy as a source

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)

"Others" Value on Nanos Polls

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 [1], on page 23 "Other % 1.0"

- On the Feb 16, 2024 Poll [2], on page 23 "Other % 0.9"

- On the Feb 9, 2024 Poll [3], on page 23 "Other % 0.9" WanukeX (talk) 02:19, 1 March 2024 (UTC)

Should EKOS be included?

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". 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)