The NBC News poll was conducted April 12-16 of 1,000 registered voters nationwide — 891 contacted via cell phone. — and the poll has an overall margin of error of plus-minus 3.1 percentage points.
Did you have a point? That’s a standard sample size for polls.
1000 people with 89% contacted via cell phone isn’t going to be representative of the US voting population. How many people do you know under 45 that answer random numbers?
1,000 registered voters nationwide — 891 contacted via cell phone.
We’re used to that as a “low number” because it’s easy to get.
But you know what?
That’s a fucking giant sample size, it’s more than enough for American voters, and while you can poll more, it quickly starts to dilute the worth.
Like, they’re calling random people, it ain’t like they’re walking down the street asking everyone and taking the first 1,000 to respond, which explained why it wasn’t 1,000 respondents…
But this?
How many people do you know under 45 that answer random numbers?
https://www.documentcloud.org/documents/24564257-240126-nbc-april-2024-poll-4-21-2024-release
Bruh, it’s a legitimate poll, you don’t have to “just ask questions” when it takes to clicks from the webpage you were already on…
But the other guy is right. There’s a problem with polling now because many people don’t answer the phone. It doesn’t matter how many people you have in the sample if it’s biased.
In this case it’s clearly biased against people who don’t answer random numbers. The “not answering” cohort may be correlated with other population groups like people with higher education and higher earnings. The survey may be systematically missing this chunk of the population, making the results biased too.
Higher educated democrats not surveyed -> the survey misses their opinions -> the survey is wrong when the results come in at election time.
Mate, it used to be cold calling landlines…
Shit is getting more accurate, not less.
Like, do you think more and better communication makes it harder to get reliable polling? You think doctors, lawyers, and rocket scientists were answering every phone call during dinner when it was landlines?
And the switch to cell phones was like, a decade ago?
Why is it now a problem?
Like, what is your version of what’s happening that polls aren’t reliable now?
Are you saying polls have been broke? Because polls were right in 2008, 2012, 2016, and 2020. And not a lot has changed…
So do you think this year “the media” is somehow doing a psyop campaign and rhats why people aren’t hyped about Biden? Because, newsflash, we weren’t last election either.
So I’m really struggling to understand why out of the goddamn blue everyone and their brother who has never even walked down a hallway in college where statistical analysis was studied in the last decade suddenly became experts.
This is the exact same bullshit the Republicans started as. Now they’re running trump.
Like, what is your version of what’s happening that polls aren’t reliable now?
Didn’t they explain?
In this case it’s clearly biased against people who don’t answer random numbers. The “not answering” cohort may be correlated with other population groups like people with higher education and higher earnings. The survey may be systematically missing this chunk of the population, making the results biased too.
Higher educated democrats not surveyed -> the survey misses their opinions -> the survey is wrong when the results come in at election time.
I admit I’m a bit surprised that we’re having a debate over whether or not selection bias is a thing (it most certainly is).
But that’s always been the case…
And polls are normally right
People think they’re bad now, because Hillary and the media said if she was projected by less then the margin of error, then she won those states and we can count on them.
Which is stupid, and was only done because they thought she was more popular and resulted in just enough people staying home that trump won…
How does that mean now when the polls are even worse, that we should ignore the polls and carry on with false confidence?
It doesn’t make any any logical sense.
