Political Discussion

Ghost said:
Man where are these polls lmfao. I’ve heard about them for eons but have never been polled in my life, are they EVEN REAL HAS ANYONE SEEN THESE POLLS
Click to expand...
I'm part of a national poll. For the big ones they randomly select from a nation wide sample. But you got to think that we have 330M people and a large statistically significant nation wide poll only needs a couple thousand respondents. So most will never get contacted for one.
 
ayayrawn said:
Litchman says Harris is pretty well on track to win unless things go catastrophically wrong for dems. Not his official prediction yet, but he basically says fuck all polls, they're not what matter.

www.newsweek.com

Polling 'Nostradamus' issues Kamala Harris update

"A lot would have to go wrong" for Harris to lose in November, according to historian Allan Lichtman.
www.newsweek.com www.newsweek.com
Click to expand...
Very scientific for a historian, lol.
 
ayayrawn said:
He's gotten 9 out of the last 10 elections right (and the one he was Wrong - 2000 - he technically WAS right). I think he is more of a strategist really.
Click to expand...
Historians aren't scientists anyway and I was just being a smart ass. I think he has some strong points, but it's worth noting that the average weighted polls have been more right than wrong over that time period too. And the 2022 election cycle was seen as the most valid and reliable set of polling data to date.

I'm not one who thinks stats tell the whole story and they can often be used to mask the reality. But they're a pretty good indicator of general trends, if not predictions. Adding more to get a full picture is needed since polls are static snapshots and rely on how the questions are worded. You can't really factor in something like the effect that foreign interference will have on an election, for instance. But someone's feelings are never a good thing to base any of this on. There are more kinds of data than just stats. Historians usually draw from that larger barrel, which is a good thing, even if I think using it to predict is foolhardy.
 
Hemotep said:
The scientific polls do not use landlines! Only media polls still do that. Many use snail mail, same as the census because it's the most assured way to get a true random sample and a high response rate.
Click to expand...
They also weight responses because they don’t get enough contact with certain demographics, because let’s be honest, how many people are answer their phone when an random out of state number pops up on their caller ID.

This is where the data turns sour. Pollsters tweak things based on previous polls being off in an attempt to make it more accurate but past results don’t predict future performance. The 2022 midterms were pretty rough for pollsters. I think too much has changed since 2020 to give them an accurate read.
 
TenderLovingKiller® said:
They also weight responses because they don’t get enough contact with certain demographics, because let’s be honest, how many people are answer their phone when an random out of state number pops up on their caller ID.

This is where the data turns sour. Pollsters tweak things based on previous polls being off in an attempt to make it more accurate but past results don’t predict future performance. The 2022 midterms were pretty rough for pollsters. I think too much has changed since 2020 to give them an accurate read.
Click to expand...
They don't use phones.
 
TenderLovingKiller® said:
They also weight responses because they don’t get enough contact with certain demographics, because let’s be honest, how many people are answer their phone when an random out of state number pops up on their caller ID.

This is where the data turns sour. Pollsters tweak things based on previous polls being off in an attempt to make it more accurate but past results don’t predict future performance. The 2022 midterms were pretty rough for pollsters. I think too much has changed since 2020 to give them an accurate read.
Click to expand...
The 2022 midterms were the most accurate data polling ever according to 538.
 
TenderLovingKiller® said:
They also weight responses because they don’t get enough contact with certain demographics, because let’s be honest, how many people are answer their phone when an random out of state number pops up on their caller ID.

This is where the data turns sour. Pollsters tweak things based on previous polls being off in an attempt to make it more accurate but past results don’t predict future performance. The 2022 midterms were pretty rough for pollsters. I think too much has changed since 2020 to give them an accurate read.
Click to expand...
And the weighting I'm talking about isn't based on different groups being unrepresented. That is bad statistics. I mean the meta analyses that weight each poll based on total respondents so that it counts for the appropriate amount when combining with other national polls. They aren't adding anything to the data in those instances. They are just counting some as more and some as less based on the total size of the survey so that they can compare the data across surveys.
 
Hemotep said:
Historians aren't scientists anyway and I was just being a smart ass. I think he has some strong points, but it's worth noting that the average weighted polls have been more right than wrong over that time period too. And the 2022 election cycle was seen as the most valid and reliable set of polling data to date.

I'm not one who thinks stats tell the whole story and they can often be used to mask the reality. But they're a pretty good indicator of general trends, if not predictions. Adding more to get a full picture is needed since polls are static snapshots and rely on how the questions are worded. You can't really factor in something like the effect that foreign interference will have on an election, for instance. But someone's feelings are never a good thing to base any of this on. There are more kinds of data than just stats. Historians usually draw from that larger barrel, which is a good thing, even if I think using it to predict is foolhardy.
Click to expand...
I don't think he even really looks much at stats. He moreso contextualizes elections based on his own set of Criteria (the 13 keys), and has just found that it's close to foolproof. I think he and you would both agree that stats aren't everything.
 
Hemotep said:
And the 2022 election cycle was seen as the most valid and reliable set of polling data to date.
Click to expand...
This is not what I have seen at all…

The ‘Red Wave’ Washout: How Skewed Polls Fed a False Election Narrative via the New York Times​

Pollsters got it wrong in 2018, 2020, and 2022. Here’s why political polling is no more than statistical sophistry via Fortune​


The only people claiming that polling was accurate are the pollsters that were objectively wrong on their polling. They will say things like the results were just a few points outside the margin of error which is true but if the margin is 7% and then they are off another 3% that is a 10% swing which is huge when swing state margins are especially tight.

I am a big data analyst guy. Love statical analysis in sport and polling but in elections they have proven unreliable as of late. I hope pollster figure it out and maybe 2024 will be different but I haven’t seen anything to suggest that it’s become any more reliable going into this election season.
 
You must log in or register to reply here.
Back
Top