Talk:Trafalgar Group

A dramatically large list of wrong polling calls for 2020

Trafalgar was wrong predicting the general POTUS results, along with being wrong on Wisconsin, Michigan, Arizona, Georgia, Pennsylvania. Biden has won by over 6 million votes and will likely get 306 electoral votes. It is my hope that these facts will be added to their "track record." Lee Nysted (talk) 20:35, 12 November 2020 (UTC)

@Lee Nysted: Thanks. Are there reliable sources for this? If so, we can probably add something to the article. —Granger (talk · contribs) 16:49, 13 November 2020 (UTC)

https://en.wikipedia.org/wiki/2020_United_States_presidential_election Mx. Granger ... Maybe this will help? Lee Nysted (talk) 17:26, 13 November 2020 (UTC)

Let me clarify: are there sources that analyze how accurate Trafalgar's polling was this year? It would be original research for us to do the analysis ourselves. —Granger (talk · contribs) 19:57, 13 November 2020 (UTC)

Biased article?

Why does this article focus so much on the presidential race of 2016 but not on the really low number of polls & general bias of the pollster + his average to low overall accuracy? Cherry picking data has no place in an encyclopedia. 2A02:908:1084:BFE0:E593:8D79:76CD:34FA (talk) 12:56, 19 October 2020 (UTC)

This pollster had a series of surprisingly true predictions, which are standing out compared to competitors. They were surprising because most of the others were uniformly and consistently wrong in their predictions at the same time.

Therefore it has been compelling to highlight these unexpected firm predictions that went against the "commong wisdom" that failed en masse.

So, such emphasis is completely justified IMHO. Filozofo (talk) 13:16, 23 October 2020 (UTC)

This isn't a good explanation. In fact, it's absolute garbage and reinforces the point: Cherrypicking data is entirely not worthy of an encyclopedia. It would be very easy to list the 25% of times were this pollster has been wrong as opposed to most other pollsters. It makes no sense to pick a single data point, unless you are trying to push a narrative. But again, this has no place on wikipedia. 2A02:908:1084:BFE0:144E:8F19:FC88:5AFF (talk) 23:52, 25 October 2020 (UTC)

The anonymous guy is completely wrong. Everything he mentioned should just be disregarded - it is emotional and likely political.

This wiki does need some work, but it is completely correct to highlight notoriety in an encyclopedia article.

It’s not cherry-picking data. We’re not here to persuade or dissuade people. And their notoriety has been 2016 and 2018 elections.

When we critique info ourselves, things are going to be problematic. For example, despite the initial comment of low to average accuracy, Trafalgar was the top 3 in accuracy in 2016 in PA, MI, GA. When you look at aggregate accuracy and percent error, it becomes even more pronounced even in FL (Trump prediction vs actual, Hillary prediction vs actual, etc.). Suparpantz (talk) 02:38, 26 October 2020 (UTC)

Trafalgar group was hilariously wrong in 2018 in determining the number of house seats, and missed the blue wave. For heavens sake, they called Kemp GA a +12 when it was only +1. I don't think it'd be a good characterization to claim that they were accurate in '18, as indicated by Harry Enten: https://www.cnn.com/2018/12/08/politics/poll-of-the-week-trumps-favorite-pollster/index.html — Preceding unsigned comment added by Wyxuan (talk • contribs) 00:06, 27 October 2020 (UTC)

Wyxuan - that should absolutely be added to the article, in my opinion, if you have a RS. But editors need to take the reigns and actually add it, it won't add itself. Chetsford (talk) 16:50, 27 October 2020 (UTC)

Being significantly below the average of most pollsters in all races and then focusing on a race where the accuracy was slightly above average is the definition of cherry-picking data. So please stop repeating the talking points and focus on the data, this is an encyclopedia, not a place where you need to push a narrative: That means stop ommiting data and at least pretend to try to stay apolitical. 2A02:908:1084:BFE0:117A:8C41:4289:8BF1 (talk) 09:23, 27 October 2020 (UTC)

No Source for Claim that FiveThirtyEight is Aligned with the Democratic Party

Hi @Filozofo:, I reverted your edit that FiveThirtyEight is aligned with the Democratic Party as it (1) did not have any source and (2) does not seem to be true. For example, I found an article criticizing FiveThirtyEight for adopting an "ideologically neutral" method of reporting. Other media bias websites have rated them as largely fact based and either a centrist publication, or only slightly to the left of center. No sources that I located identified them as being aligned with the Democratic Party.

Unless you can find a reliable source saying that FiveThirtyEight has aligned itself with the Democratic Party, I don't think your edit is appropriate. Thanks. DocFreeman24 (talk) 14:13, 24 October 2020 (UTC)

Reverted Edit Re Inclusion of 538 Rating

Hi @Homegr0wn1:, can you please explain why you keep deleting the reference to FiveThirtyEight's rating? I understand that you believe there are additional rating services out there and that you don't consider them authoritative, but I don't think that's a basis for removing their rating? If you'd like to add additional ratings, please of course feel free. But I don't understand the basis for removing this content. FiveThirtyEight is a fairly prominent media outlet in the polling world, so I don't really understand your objection and I think this content should be included (as both myself and another edit have reverted your edit).

PS. Sorry for reverting your last edit without leaving a comment. That was a misclick as I was writing my comment.

DocFreeman24 (talk) 00:21, 26 October 2020 (UTC)

The Accuracy and Performance should be based on measurable data (e.g., prior polling versus actual results). 528's letter grade of polling does nothing to help a user understand the accuracy or performance of prior polling. Instead, listing out polling versus results should be utilized. Users of Wikipedia should not be shown a letter grade to rate a pollster; they should be given the information directly. Also, 538's own page shows a 75% correct prediction by Trafalgar while other pollsters with much lower prediction percentages are given A letters. Accuracy and Performance should be based on correct predictions, not arbitrary letter grades. Also, 538 states "FiveThirtyEight’s pollster ratings are calculated by analyzing the historical accuracy of each firm’s polls along with its methodology." So, the rating also take into account methodology. Perhaps the 538 letter grade is better suited for a rating of their methodology rather than their accuracy and performance?

Homegr0wn1 (talk) 00:40, 26 October 2020 (UTC); edited after Homegr0wn1 (talk) 00:51, 26 October 2020 (UTC)

Hey, thanks for having a discussion on this and I think I understand your point. I don't think I agree that "528's letter grade of polling does nothing to help a user understand the accuracy or performance of prior polling" or that 538's ratings aren't based on prior polling results. In fact, 538's articles state that prior polling results are a significant part of their grading, along with other things like transparency of methodology, format of the type of polling used, etc. But our disagreement would seem to be a bit beside the point? If we moved the rating to a separate section entitled "Ratings of Methodology" would that resolve your concern? DocFreeman24 (talk) 01:32, 26 October 2020 (UTC)

Yeah, I think that's totally fair. Their methodology seems questionable, but they have been surprisingly accurate. So, I think that's a good resolution. Good call!

Homegr0wn1 (talk) 01:40, 26 October 2020 (UTC)

Great! I'll make that edit. Thanks for working through this with me! DocFreeman24 (talk) 02:04, 26 October 2020 (UTC)

I'm glad this has been resolved. For what it's worth I agree that FiveThirtyEight's rating is worth mentioning in the article. —Granger (talk · contribs) 19:07, 26 October 2020 (UTC)

References

Merger proposal

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.

To not merge, on the grounds of distinctly notable topics, and no structural advantage to a merge. Klbrain (talk) 12:05, 12 May 2021 (UTC)

---

Robert Cahaly should be merged with Trafalgar Group. This person is not significant aside from his involvement with the Trafalgar Group. Snooganssnoogans (talk) 14:25, 3 November 2020 (UTC)

Keep separate, each article is receiving thousands of daily hits separately. Cahaly has given many individual interviews on most of the major new sources (The New York Times, Daily Mail, Wall Street Journal, Politico, Washington Post, etc...) and is significant in his own right and it makes it much easier to find and read information about him when searching for his biographical information. He is also the individual named party in a U.S. Circuit Court of Appeals case.Swampyank (talk) 15:16, 3 November 2020 (UTC) 15:13, 3 November 2020 (UTC)

But his notability is not distinct from his involvement with the Trafalgar Group. If he were not speaking as a representative of the Trafalgar Group, he wouldn't be notable. Snooganssnoogans (talk) 15:17, 3 November 2020 (UTC)

He became notable for his polling work, but many articles only name him without mentioning Trafalgar Group. He has a long history in politics prior to starting polling and I also think his role in the robocalling case establishes additional notability beyond his polling work with Trafalgar. - 75.50.53.130 (talk) —Preceding undated comment added 16:08, 3 November 2020 (UTC)

• The pollster is definitely independently from the polling firm. So much so that if a merger were to take place it eould perhaps be better the other way around. FloridaArmy (talk) 23:55, 3 November 2020 (UTC)
• Agree that if the merge happens, then it should be the other way around, although it seems funny enough to leave up. Consider the fact that Conservapedia is a separate article from Andrew Schlafly, even though both are awful sources for information. Andrew has a footprint outside of that joke of a project, just like Robert does. Trafalgar is a side piece, but notable enough that conspiracy theorists, such as the 45th President, used it for public gish galloping attempts. Robert has the footprint, though, that Trafalgar does not. Seems they should be separate articles, but if it happens, the group should collapse into Robert's article. 24.217.244.49 (talk) 03:43, 2 December 2020 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Revert removal

Reverted recent text. Don't understand why this long-standing text was removed Andre🚐 19:16, 2 December 2022 (UTC)

New York Times bias

An editor has been removing text critical of Trafalgar Group and references to the NYT, with the claim (in edit summaries) that since NYT has their own poll, they are therefore biased against Trafalgar Group. This reasoning seems specious to me; it would imply, for example, that no newspaper can report reliably about another newspaper. I don't believe we can just assume that NYT must be biased against this (and presumably all other) polling organizations without a source that says that. CodeTalker (talk) 23:27, 23 May 2024 (UTC)

Accuracy of 2020

Hi, I changed section about election 2020. But it was reverted because of referencing errors. I don't know, if I will have a time for correcting. Can you repair it? I was writing about important of accuracy in %. I a polster predicted win/lose by 0,1 % and another by 5 % in close election, then the first polster was more accurate. Hrdlodus (talk) 07:58, 18 September 2024 (UTC)

@Hrdlodus

I added in accuracy from error, cited to secondary FiveThirtyEight statement about them as the second most accurate pollster. Anyone who knows anything about stats and polling knows items like error and bias are what matters, otherwise, as you said, someone who predicted Biden winning by 50 points is "better" than a tight race predictor who leaned Trump. I left in the prediction of Trump winning cited to secondary RS from NYT, removed the primary OR that the rest of the section was filled with, and replaced it with RS characterization of race calls. KiharaNoukan (talk) 15:00, 28 September 2024 (UTC)