What The H*ll Is Happening With These Alabama Polls?

[sun rise]

This to me is a very good article because it lays out the advantages and problems with polling methods which has resulted in wildly different predictions for tomorrow's election.

The analysis of poll numbers had this "fun" note:

But perhaps the most interesting poll of all is from the online firm SurveyMonkey. It released 10 different versions (!) of its recent survey, showing everything from a 9-point Jones lead to a 10-point Moore lead, depending on various assumptions — all with the same underlying data.

Although releasing 10 different versions of the same poll may be overkill, it illustrates the extent to which polling can be an assumption-driven exercise, especially in an unusual race such as Alabama’s Senate contest....

 

[Mindmaster]

This to me is a very good article because it lays out the advantages and problems with polling methods which has resulted in wildly different predictions for tomorrow's election.

The analysis of poll numbers had this "fun" note:

But perhaps the most interesting poll of all is from the online firm SurveyMonkey. It released 10 different versions (!) of its recent survey, showing everything from a 9-point Jones lead to a 10-point Moore lead, depending on various assumptions — all with the same underlying data.

Although releasing 10 different versions of the same poll may be overkill, it illustrates the extent to which polling can be an assumption-driven exercise, especially in an unusual race such as Alabama’s Senate contest....

They did what they always do - polled only people in Democratic heavy urban areas. As a rule:

If they say:

tie - the Republican is leading by 5%
losing by 5% - is dead even
losing by 10% - is winning by 5% or more

 

[Revoltingest]

I predict that someone will claim to have predicted which ever result ensues.

 

[Subduction Zone]

I predict that someone will claim to have predicted which ever result ensues.

With these polls all over the place one of them is bound to be fairly accurate. Sadly it is much like the year of end predictions made by "psychics". With enough predictions some of them are bound to be true. All one has to do is to cherry pick one's predictions after the fact and TaDA: Amazing accuracy!

 

[Revoltingest]

With these polls all over the place one of them is bound to be fairly accurate. Sadly it is much like the year of end predictions made by "psychics". With enough predictions some of them are bound to be true. All one has to do is to cherry pick one's predictions after the fact and TaDA: Amazing accuracy!

On a related note....
Do you know the Texas Sharp Shooter fallacy?
The Texan takes a shot at a distant barn.
Then he walks over to it, finds his bullet hole, & paints a bulls eye around it.

 

[Subduction Zone]

On a related note....
Do you know the Texas Sharp Shooter fallacy?
The Texan takes a shot at a distant barn.
Then he walks over to it, finds his bullet hole, & paints a bulls eye around it.

Yep, this could very easily end up in a classic version of it. If a person is wrong 90% of the time and gets lucky once that is never a good reason to follow him. Sadly that is not the way that it works quite often. If one person is right he will be able to profit from his luck.

Though there may actually be a valid pollster or two working this case. All we can do is to wait until the actual results are turned in. Hopefully we find out tomorrow evening.

 

[Revoltingest]

Yep, this could very easily end up in a classic version of it. If a person is wrong 90% of the time and gets lucky once that is never a good reason to follow him. Sadly that is not the way that it works quite often. If one person is right he will be able to profit from his luck.

Though there may actually be a valid pollster or two working this case. All we can do is to wait until the actual results are turned in. Hopefully we find out tomorrow evening.

It's easy for someone to have a vague inkling that something will happen.
And if eventually something which fits into this broad set of scenarios occurs,
then lo! They predicted precisely what just happened.

 

[sun rise]

They did what they always do - polled only people in Democratic heavy urban areas. As a rule:

If they say:

tie - the Republican is leading by 5%
losing by 5% - is dead even
losing by 10% - is winning by 5% or more

You totally ignored the article including that Fox news poll was one of them. But I guess to you Fox news is a Democratic promoting news outlet.

 

[sun rise]

It's easy for someone to have a vague inkling that something will happen.
And if eventually something which fits into this broad set of scenarios occurs,
then lo! They predicted precisely what just happened.

SurveyMonkey has that covered already since they can use one of their 10 scenarios as being successful.

 

[Revoltingest]

SurveyMonkey has that covered already since they can use one of their 10 scenarios as being successful.

Is SurveyMonkey anything like Surrender Monkey (but with less cheese consumed)?

 

[Subduction Zone]

You totally ignored the article including that Fox news poll was one of them. But I guess to you Fox news is a Democratic promoting news outlet.

You mean this one:

"FOX News 12/7 - 12/10 1127 LV 3.0 40 50 Jones +10"

Fake news!! Fake news!!

 

[sun rise]

You mean this one:

"FOX News 12/7 - 12/10 1127 LV 3.0 40 50 Jones +10"

Fake news!! Fake news!!

That's the one. Those far-left Republican-hating fake news types at Fox news have done it again. They're the worst possible people - Hillary supporters. They refuse to say anything bad about her or anything good about God-Emperor Trump. Lock them up.

 

[pearl]

I stopped paying any attention to polls after Trump.

 

[Stevicus]

This to me is a very good article because it lays out the advantages and problems with polling methods which has resulted in wildly different predictions for tomorrow's election.

The analysis of poll numbers had this "fun" note:

But perhaps the most interesting poll of all is from the online firm SurveyMonkey. It released 10 different versions (!) of its recent survey, showing everything from a 9-point Jones lead to a 10-point Moore lead, depending on various assumptions — all with the same underlying data.

Although releasing 10 different versions of the same poll may be overkill, it illustrates the extent to which polling can be an assumption-driven exercise, especially in an unusual race such as Alabama’s Senate contest....

Polls are kind of useless. Every time I think of polls, I think of Linus running for school president and Lucy conducting polls.

 

[metis]

I stopped paying any attention to polls after Trump.

I pay attention but take them with a grain of salt about the size of the Rock of Gibraltar.

 

[Terrywoodenpic]

On a related note....
Do you know the Texas Sharp Shooter fallacy?
The Texan takes a shot at a distant barn.
Then he walks over to it, finds his bullet hole, & paints a bulls eye around it.

And how did he know it was even his shot...
everyone shoots at barns.....

 

[Revoltingest]

And how did he know it was even his shot...
everyone shoots at barns.....

Take credit where one can.
Like inventing the internet!

 

[It Aint Necessarily So]

SurveyMonkey has that covered already since they can use one of their 10 scenarios as being successful.

One of the requirements of high quality prophecy or prediction is that it not be accompanied by failed predictions.

Speaking of which, have you heard of the scam where a prisoner claims to have psychic ability in a series of letters or emails sent to random marks, who are also told that he will split the profits from sports bets that he cannot make himself from within in exchange for the tips if they will place the bets with their money. Of course, he adds, he doesn't expect his mark to take his word for it, so he will prove himself with five correct predictions first - "Just wait and see."

Do you know where this is going?

Game 1: 1000 people get team A1 as the predicted winner, and 1000 get B1

After the outcome of the game is known, only the batch that got the correct prediction gets a follow up communication for Game 2, 500 being told team A2, and 500 told B2.

Continue three more times until you have a have a group of 500 /(2x2x2) = about 60 people who have seen five correct predictions, and you might get about half to place a bet on your 6th prediction, half of whom will win and probably send money for the next tip.

It's all the unseen failed predictions that would have revealed the scam.

 

[It Aint Necessarily So]

I stopped paying any attention to polls after Trump.

Good call. Polls don't take voter obstruction or electronic voting machine hacking into account.

Voters Are Left in the Dark By Alabama’s Election System - WhoWhatWhy

 

[Nous]

This to me is a very good article because it lays out the advantages and problems with polling methods which has resulted in wildly different predictions for tomorrow's election.

The analysis of poll numbers had this "fun" note:

But perhaps the most interesting poll of all is from the online firm SurveyMonkey. It released 10 different versions (!) of its recent survey, showing everything from a 9-point Jones lead to a 10-point Moore lead, depending on various assumptions — all with the same underlying data.

Although releasing 10 different versions of the same poll may be overkill, it illustrates the extent to which polling can be an assumption-driven exercise, especially in an unusual race such as Alabama’s Senate contest....

I think the first explanation given by article makes sense--the polls that are limited to landlines get respondents that are older and whiter. I'm thinking Jones has a good chance of winning. If he does, it's definitely noteworthy in showing how that sexual assault allegations against Moore apparently had an impact on Alabamans willingness to vote for him.