Vinay Prasad a Progressive Explains Reality (DeSantis)…

So you want me to demonstrate the veracity of the source of an opposing viewpoint?

yeah, I’m going to say nahhhh. I’ll pass.

 

The inference, in this case, is the sentence, not paragraph, that I quoted. In terms of endogenous selection, you are right and I should have been more exacting in my terminology. It would be endogenous response rather than selection in a survey setting (it isn't really different mathematically, just differs theoretically). And there is no evidence that the people who responded differ in any way from those that didn't respond. So, unless there is some reason to believe that, then there is no issue with a small response rate as long as the data has enough power to show the relationships. To be honest, the bias would need to be extraordinarily high in one direction to even come close to mattering here.

 

I go back to 4th of July example. I don’t think you will find a soul alive who would give any credibility or credence to a national survey where less than 4% responded on such a small sample size. I stick with my opinion that this survey is useless. I gave you a very viable and reasonable hypothesis on why 96% of those asked refused to answer. I don’t care jack s$it about excuses that the industry invents to justify or question the validity of this voodoo science endeavor we call polls/surveys. Time and time again these surveys/polls are shown to be grossly inaccurate. 2016 election! 2022 Florida election. 2022 Fed election. Etc etc etc

 

The 2016 election polls was only off by about 1% from final results. 2022 federal elections were quite close as well. Regardless, this has nothing to do with election polls, especially late stage election polls which require a likely voter model.

 

What would you consider to be a valid sample size?

 

I think polls are rather stupid, but a sample size Od 4% is meaningless. If you put 100 people in a room and randomly grab four people and ask them whether they like their steak: rare, medium rare, medium or well done, you have a very small mathematical possibilities of choices. There is very little chance that those four people will reflect the opinions of the other 96 people. 4% is meaningless. To prove this point, you will notice that the networks will not call an election until at least 40% of the vote is in, even when you consider exit polls. The rare exception is if the exit polls come back with such a disproportionate number. I cannot fathom that any reasonable person would suggest that 4% of 5000 people polled demonstrate a consensus on any issue. Further, there are over 1 million practicing physicians in the United States according to what I found on the Internet. Are you really going to suggest that a poll of 200 of them provide a basis to draw a conclusion? Zero chance of that. That is .02%, which is virtually nothing.

 

Statistics though would say if you put 10,000 people in a room, and pull 400 people out of that group, you will likely get a pretty reliable estimate of the population's distribution as long as the process to pull those people out is random.

 

Maybe, maybe not. The point is: the end result is speculation. The 2022 Florida Governors poll is a good example. As I recall, a month or 2 before the election many of the dozen or so polls were a statistical tie. Some showed a Desantis 10 point lead. Point is, they were all over the place and I dont think any poll was close to the end result. 2022 congressional and senate polls were even worse. I have been voting since 1982. Granted, I am only one person, but I have never met a single person who has changed their vote for any candidate, for any emotion. Maybe there are a few people out there, but I have never met them. Also, during every election cycle, I talk to my friends about the upcoming election. I have friends on the left and the right and in the middle. I have never met an undecided voter for a general election. Primaries yes, but not close to the voting date. The point is, I think the number of people who walk into a polling booth who don’t know who they’re going to vote for is about the same number of people who have stepped foot on the moon. Now, I don’t claim to have any expertise in poll taking or statistics. But….I know malarkey when I see it. The end result of any poll, at best, is an educated guess, on average speculation, at worst a blind dart throw.

 

Expected...don't need pesky facts getting in the way of your opinion.

 

No, polling is not speculation. It is evidence. Asking people what they think to answer the question of what they think is one of the more established methodologies in the world

BTW, the 2022 polls were historically accurate.

The Polls Were Historically Accurate In 2022

And yes, there are undecided voters in the world, as well as those that are undecided whether they will vote. They are largely less educated and relatively apolitical. People you don't talk politics with, generally.

 

I don’t think you understand how Internet message boards work.

 

If 2022 is what you consider accurate then I would hate to see what qualifies as inaccurate in your eyes.

 

It is a quantitative question. 4-5% random error with less than 1% bias.

 

I don’t think we are talking about the same polls. LOL

 

Read the article. It is the totality of polls.

 

I watched every network’s predictions before the 2022 election. 100% predicted a red wave. Some even predicted a tsunami. That didn’t happen. Pubs significantly under performed….except Desantis, who over performed

 

Now you are conflating qualitative predictions with polls. They aren't the same thing. Senate polls were biased by 0.3% to Republicans while House polls were biased 0.2% to Democrats.

 

No my friend, I’m just talking about polls that the average man and woman are confronted with daily, and more basically the fallacy of asking 5000 people a question, and drawing a global conclusion based upon the responses from the 200 people who cared to respond.

 

Again, the numbers I am providing were from all publicly available polls. And, once again, unless you have some evidence that those respondents differ in some systematic way from non-respondents, basic statistical theory will tell you that none of this is an issue.

 

You have a very warped way of looking at things. I don’t have to prove anything to you, and it is not my obligation to prove that your hypothesis is correct. If you walk into a room of 100 people and
ask four of them what their favorite color is, and then come to me and tell me that everybody in the room favors blue because you spoke to four people, I am literally going to laugh at you. You can try to throw out all of the pseudoscience crap you want about statistics, and I’m still going to laugh at you….as well as the majority of rational thinking people. This is not an attack on you, it’s an attack on the process that you are cheerleading.