Can you remind us how those polls performed in 2018? what about 2014, 2012, 2010, or 2008?

I'm sure you would never focus all your analysis on just 2016, like the poker player who plays any hand with a 6 because 6s are hot

So if Trump wins again all you guys trolling this thread will be back "explaining" how it was another anomaly?

If I flip heads twice in a row again, you gonna be back shouting it's 50/50 it was an anomaly?! Like do you even know what the probability of Trump's 2016 "win" was, given the last polling numbers and the same for a 2020 win?

Lastly, newbie accounts are exiled until after the election.

If Trafalgar is wrong in 2020, which of these explanations seems the best/worst?

Trafalgar is better than standard polling and 2020 is a fluke
Trafalgar is worse than standard polling and 2016 was a fluke
Polling is a crapshoot and you can't trust any numbers

He was making an incompleteness point but I'd bet that if a counter-example is found then there is a mistake in the proof.

Dr. Malmuth,

Do you think the endorsement of Donald Trump by the popular rapper 50 Cent might produce a pro-Trump surge among young black voters that square polling firms might miss out on? Also, where do you stand on the Nicki Minaj vs. Meek Mill beef?

Thank you for your time,

-TMcT

I didn't cherry pick anything. I posted what I did because these two polls, according to their representatives, have an approach that's different relative to how they collect information from their questions.

Your other stuff is just nonsense. In general, when the economy is doing well, there should be more disposable income for gambling. When there is an economic shutdown and the economy contracts, this won't be the case.

MM

I have no idea how they did in other years. What I'm saying is that based on the interview I heard the following:

1. Their approach in how they get their information through questions is different.

2. If it turns out that they are more accurate then the other polls, expect other polling organizations in the future to adopt their methods.

3. If their results are not accurate, they'll probably disappear in the future or at least change their methodology.

In some of my early posts in this thread, I was speculating that there might be problems in the questions themselves and these two polling organizations seem to agree. That's why I mentioned it.

And one other thing. Polling is much more sophisticated than playing a hand with a 6.

MM

But this is not the first year they've been on the scene. In 2018, their results were not accurate, but yet they didn't disappear, nor does it appear they've changed their methodology.

Hi BD:

This is actually a good question. However, polling is not just one data point as your question seems to indicate. There are lots of different states, lots of other races besides the presidency, and methods that work well in some spots may also work poorly in others. All of this will need to be reviewed.

Best wishes,
Mason

First, I doubt if 50 Cent's endorsement will have much effect on how young black voters will vote. However, if enough endorsements like this happen, then the weight of them in total might have an effect.

As for your other question, whatever you think is fine.

Mason

I suggest that you contact the Trafalgar group directly and see what they say. Whether a specific polling organization is good or bad, was not the purpose of this thread. I just found it interesting, based on the interview I heard, that their approach was different.

Mason

My point isn't even that Trafalgar is bad. My point is that your standard for judging whether a polling outfit is good or bad is nonsensical. We know for a fact that certain polling outfits keep giving results that strongly favor one side, and yet they hang around even after elections when they were very wrong and nevertheless still consistently give results that strongly favor the same side. To think that a polling outfit would fade into obscurity based on a bad result doesn't line up with reality at all. In fact, conservative-leaning polling outfits seem to have a business model where they hang their hat on being one of only a handful that get things more right than the consensus when things break conservatives' way, even if that only happens rarely, rather than being in the middle of a large pack with consistently average accuracy. Trafalgar is not even unique in this regard.

Actually it would be a terrible bet for me. If young black men are truly individually 20% to vote for Biden then the chances that 44% of a few million will actually vote for him is WAY smaller than a counter example existing.

Are you interested in a wager on the % of black americans supporting trump? I know the Trafalgar poll had it at 44% so we could start there. I'd want to take the under on that number.

We could adjudicate using the CNN exit polls, or another mutually acceptable medium.

What if Kanye, 50 Cent, BlocBoy JB, Fivio Foreign, and the legendary Lil Pump all endorsed Trump? Do you think that might generate a strong pro-MAGA force among the young black voter bloc? Maybe there is a huge shift that even Trafalgar doesn't know about. 44% might be an underestimate if you ask me.

https://www.xxlmag.com/rappers-suppo...nt-trump-2020/

But I didn't judge whether a polling outfit is good or bad. All I was saying was that Trafalgar and the other group, according to what their representatives said, had a different approach from most polling organizations to their questioning. If you look at my early posts in this thread, I was talking about how if questions are not carefully worded they can produce erroneous results -- response variance, response bias, and conditioning.

As for whether a particular business stays in business often has something to do with the quality of their product. I know this from personal experience as I've watched a number of publishers in our field of poker/gambling disappear.

But a good product can also have something to do with the need that is out there. If, for instance, as you point out, that polls which consistently lean certain ways have customers because these customers might want polls that lean certain ways, then in a sense they're putting out a good product for that customer.

Mason

If Trafalgar said 44 percent and you think that's wrong, and it seems to me that 44 percent would be wrong, and you feel strongly about it, then I suggest you contact Trafalgar and don't challenge me.

Again, the purpose of this thread and the reason I started it was to discuss possible issues in the polling. Since years ago I did work in this area, I thought I might have something worthwhile to contribute.

Mason

I'm not an expert on politics or what might attract certain groups of voters to specific candidates, but my opinion is probably not.

Well, i know they went on Hannity and all but If their methodology produced 44% for trump from black Americans, it’s very likely their methodology is **** - similar to the poker book arguing you should always raise a hand with a six.

Not sure why you’re implicitly defending a methodology that produces such an asinine result.

If you’re not defending them, and you also think that result reveals their polling ideas as horseshit, than we can abandon this line of discussion since I agree.

Also, I’m not going to call them. Even though I long ago stopped playing poker for a living, I can’t turn off my instinct to exploit the fish. And someone who wanted to stand behind that trafalgar result of 44% black Americans supporting Trump.. would be a fish, to say the least.

I felt their appearance on Hannity was important to this thread since they explained to some degree how their methodology dealing with how they ask questions was different. I don't think I ever said that their different way of asking questions, more indirect than what most polls do, was superior to the standard way of asking questions. But I did explain in some of my posts above that poorly constructed questions can produce poor data, and here were two organizations who were addressing this area.

As to whether they got it right or not, I think we might know when the election results are in.

Also, I think you're looking at one data point when you keep talking about the 44 percent. Their results should have lots of data points and they all need to be looked at, after the election, to get an idea as to how well they have done and whether this different approach of asking more indirect questions is better.

MM

I suggest you look into the statistical concept known as events of signature significance.

A reputable polling methodology doesn’t produce 44% Republican support in this environment. Not even once.

Hi Everyone:

There have been a number of statements about the Trafalgar Group doing a poor job of predicting in 2018. So, I found this:

https://www.realclearpolitics.com/ar...in_138621.html

and it looks like Trafalgar did a good job of predicting in 2018.

I've also tried to find where Trafalgar says that 44 percent of the black vote in Michigan is for Trump but have not been able to. Perhaps someone can link me to it.

Best wishes,
Mason

http://www.basedonactualmath.com/201...-iowa-was.html

Re events of signature significance.

Also I had been relying on the thread re 44% the number. If they did t publish it I retract the relevant criticisms.

I just found this:

https://twitter.com/NateSilver538/st...215875/photo/1

and it looks like Trafalgar is saying that Trump will win 27.8 percent of the Black vote in Michigan. That's a lot different from 44 percent.

mason