Trafalgar was the worst pollster of the California recall vote. Now it's pretending VA Gov is a tie

[Sabuca]

Newsom won by 24% while Trafalgar had him up by only 8%.
That's an abysmal 16% error on the margin.
Yet, this Trumper pollster still has the nerve of pretending that things are tied in Virginia.

https://projects.fivethirtyeight.com/california-recall-polls/
https://www.thetrafalgargroup.org/news/va-gov-poll-1014/

[TheLegend]

1. And the biggest joke is they still have a A- from 538

[Sabuca]

2. Nate is in love with Trafalgar

He hasnt addressed their CA debacle.

[hlthe2b]

3. Well, I'd calmed down by the time the Newsome recall election began, but I'm nervous as hell

about the VA Governor election. So, I'd love to be convinced that it is not so close. But even beyond that pollster's record, what evidence is there that things are better than assumed?

[USAFRetired_Liberal]

5. The foxnews poll

That was released the same day

https://www.realclearpolitics.com/epolls/2021/governor/va/virginia_governor_youngkin_vs_mcauliffe-7373.html

[lagomorph777]

14. Average is still only McCauliffe +1.8

I'm nervous. I hope my ballots arrive today.

[FBaggins]

4. California does not provide useful information

Recall elections do not lend themselves to standard polling - particularly because we had no idea who would show up to vote.

Trafalgar leans red, but they aren’t a bad polling firm. They were better than most of alternatives in both 2016 and 2020.

There’s also plenty of reason to believe that VA is painfully close right now. It would definitely be a mistake to convince people that there’s nothing to worry about and McAuliffe is really up by 16.

Early voting number are well below where we want them to be and far too small a proportion are coming out of the NOVA suburbs for my taste.

[Sabuca]

6. There's no evidence recalls are harder to poll (And he had Trump beating Biden)

Than other polls. Do you have any data regarding that comparison?

Besides, I am not simply arguing that Trafalgar was bad. I'm saying they were THE WORST.
So even if we assume that you are correct (that recalls are oh so hard to poll), then why was it harder for Trafalgar to poll than everyone else?
Finally, Trafalgar had Trump ahead of Biden by 2% in the popular vote.

Link to tweet

?t=lPp9JHrAG1rQt94nwNnplw&s=19

[FBaggins]

7. There's a big difference between you not wanting to believe something and there not being evidence

Do you have any data regarding that comparison?

Perhaps the fact that not a single pollster was able to come within the statistical margin of error of the eventual results? Or even close?

As for Trafalgar - you're committing the cherry-picking fallacy. As a counter-example, we had a critical statewide election here in NC a few months ago. You may remember it? It would have made Manchin irrelevant if we had won.

In the 3-4 weeks preceding the election, not a single poll (except one) showed the Republican ahead... with some showing double-digit leads for the Democrat. That one poll showed the Republican up by two points. He won by two points.

Care to guess the pollster?

Hint - they were also the closest or second closest in the state on Trump actually winning here - when CNN and NBC each had Biden up by 6.


We could play that game forever. The key takeaway is that actual data on the ground says that we have to worry about VA. Posting a thread that claims essentially that we have nothing to worry about is counter-productive at best. People need to panic, not relax.

[Sabuca]

8. "not a single pollster was able to come within the statistical margin of error"

You act as if several pollsters miss a result by more than the margin of error, they are automatically just as bad...despite the fact that Trafalgar missed the victory margin by a whopping 16%.
No, being off by 8% and being off by 16% or not equally bad.

[FBaggins]

10. Moving the goalposts doesn't help you

The point was offered to demonstrate (at your insistence) that the recall election was unlike a normal election from a polling standpoint. It was. Not because any of the pollsters were incompetent.

Your cherry-picked claim of one firm performing worse than anyone else in a race that nobody polled accurately at all was easily rebutted by pointing to a far more relevant race that they polled better than anyone else. They were also the best or second best in 2020 presidential polling of FL and PA and WI and OH.

There's a reason that 538 grades them so highly.

Now we'll all just pretend that nobody noticed you dodging the main points of the argument yet again.

[Sabuca]

11. So your statistic is one election?

I asked for data about recalls vs. Non-recall elections, and ironically, the anti=cherry -picking crusader offers a sample of one (1) as "evidence".
Do you know what that's called?

[FBaggins]

12. Now you have trouble counting?

I cited five elections... and previously told you that we could do this all day.

Realclearpolitics rated them as the most accurate pollster of the 2016 cycle (scoring dozens of statewide races - including the only pollster to show Trump winning MI/PA).

I asked for data about recalls vs. Non-recall elections

There aren't many statewide recalls and California's recall process is recognized as unique... there was only one relevant data point. Particularly since demonstrating whether that specific recall's polling had any relevance to any other polls was the entire point of the debate.

Keep digging. It's entertaining.

[Sabuca]

16. You said recalls are hard to poll, based on 1 instance ( You hoped I forgot the Walker recall)

No responsible person would reach the hasty, evidencee-free conclusion that a certain kind of contest is hard to poll, based on one election.
And then came to understate man of all understatements: "There aren't many statewide recalls". Dude, you cited one. "Not many"? How insightful. But then, you said recalls at hard to poll. That was a horrible analysis.

The worst part is that you remember very well the 2013 Walker recall, but you've been silent about it, hoping that I didn't remember. Bad news for you, I do remember.
And so do you. You are familiar with RCP, so you must know that the poll average was 6.7%, and Walker one by 6.8%. Therefore the polling average was extremely accurate.

I want your help picking a pumpkin for Halloween. You only see the pretty ones.
https://www.realclearpolitics.com/epolls/2012/governor/wi/wisconsin_governor_recall_election_walker_vs_barrett-3056.html

[Sabuca]

9. "not a single pollster was able to come within the statistical margin of error"

You conveniently pretend that if several pollsters miss a result by more than the margin of error, they are automatically all just as bad...despite the fact that Trafalgar missed the victory margin by a whopping 16%.
No, being off by 8% and being off by 16% are not equally bad.
I think if you admit that Trafalgar was super awful in the recall forecast you will be in better standing.

[Polybius]

13. To be fair, we overperformed in Cali

It wasn't supposed to be 24 points, and many Democrats were expected to pass. In the end, nearly everyone voted.

[former9thward]

15. Monmouth has them tied at 46.

They are the most recent poll and I guess they are "pretending" too.

https://www.monmouth.edu/polling-institute/reports/monmouthpoll_VA_102021/

[FBaggins]

17. You were saying?

It looks like Trafalgar picked up the shift correctly and their final poll is in the running to be the most accurate of the bunch.

[FBaggins]

18. And while we're on the subject...

... guess which pollster came by far the closest in NJ?