Advertisement
by Valrifell » Fri Oct 02, 2020 10:15 am
Imperialisium wrote:Washington Resistance Army wrote:
I don't think you understand what margin of error is in polling. Which isn't a bad thing, you just appear to be speaking from ignorance.
I’m not speaking of ignorance. I know very well how margin of errors work given my background. The problem with NSG is the precedence of taking a generic outcome pertaining to the outcome of something and then extrapolating it to create an entire persona, narrative, and even argument for the person you’re disagreeing with.
“Polls were wrong because of outcome,” isn’t technically incorrect. They (which can be any number of specific polls) predicted the wrong outcome of who won. No more or less from that comment should and could be gleaned. But as you see no one wants to really discuss how that happened or why they disagree; or, could it feasibly occur in predicting the outcome of the 2020 election. So they instead go for the horrible fallacies of just invalidating the person by going “you’re ignorant, you don’t know math, you are X here.”
by Imperialisium » Fri Oct 02, 2020 10:16 am
Australia-New Zealand-Taiwan wrote:Biden is a pedophile
by Washington Resistance Army » Fri Oct 02, 2020 10:17 am
Imperialisium wrote:Washington Resistance Army wrote:
I don't think you understand what margin of error is in polling. Which isn't a bad thing, you just appear to be speaking from ignorance.
I’m not speaking of ignorance. I know very well how margin of errors work given my background. The problem with NSG is the precedence of taking a generic outcome pertaining to the outcome of something and then extrapolating it to create an entire persona, narrative, and even argument for the person you’re disagreeing with.
“Polls were wrong because of outcome,” isn’t technically incorrect. They (which can be any number of specific polls) predicted the wrong outcome of who won. No more or less from that comment should and could be gleaned. But as you see no one wants to really discuss how that happened or why they disagree; or, could it feasibly occur in predicting the outcome of the 2020 election. So they instead go for the horrible fallacies of just invalidating the person by going “you’re ignorant, you don’t know math, you are X here.”
by Valrifell » Fri Oct 02, 2020 10:18 am
Washington Resistance Army wrote:Imperialisium wrote:
I’m not speaking of ignorance. I know very well how margin of errors work given my background. The problem with NSG is the precedence of taking a generic outcome pertaining to the outcome of something and then extrapolating it to create an entire persona, narrative, and even argument for the person you’re disagreeing with.
“Polls were wrong because of outcome,” isn’t technically incorrect. They (which can be any number of specific polls) predicted the wrong outcome of who won. No more or less from that comment should and could be gleaned. But as you see no one wants to really discuss how that happened or why they disagree; or, could it feasibly occur in predicting the outcome of the 2020 election. So they instead go for the horrible fallacies of just invalidating the person by going “you’re ignorant, you don’t know math, you are X here.”
But that's the thing, apart from Wisconsin the polls weren't really wrong. Pretty much everywhere else, including how the national vote numbers went, the polling was perfectly within margin of error. You can't have much better polling than that.
Admittedly yeah a bunch of the election models should be mocked (like the one that gave Clinton a 99% chance of winning) because a lot of them ignored very glaring flaws in the Clinton campaign and the background in the Rust Belt that led people like me to say Trump was going to get a narrow victory there, but again the polling was mostly solid.
by Cordel One » Fri Oct 02, 2020 10:20 am
Australia-New Zealand-Taiwan wrote:Biden is a pedophile
by Imperialisium » Fri Oct 02, 2020 10:22 am
Valrifell wrote:Imperialisium wrote:
I’m not speaking of ignorance. I know very well how margin of errors work given my background. The problem with NSG is the precedence of taking a generic outcome pertaining to the outcome of something and then extrapolating it to create an entire persona, narrative, and even argument for the person you’re disagreeing with.
“Polls were wrong because of outcome,” isn’t technically incorrect. They (which can be any number of specific polls) predicted the wrong outcome of who won. No more or less from that comment should and could be gleaned. But as you see no one wants to really discuss how that happened or why they disagree; or, could it feasibly occur in predicting the outcome of the 2020 election. So they instead go for the horrible fallacies of just invalidating the person by going “you’re ignorant, you don’t know math, you are X here.”
Polls are not predictive, it's quite that simple. They are snapshots of support at any given point in time, and we saw that as we got closer to election day Trump and Clinton closed in on each other in a few key states to the point it was within the MoE, which coupled with the high percentage of undecideds that we caught, makes it (in hindsight) relatively unsurprising that the election shaped out as it did.
The polls right before election day were pretty much right! They captured the correct level of support between the candidates within error! That is the mark of a successful measurement!
by Imperialisium » Fri Oct 02, 2020 10:25 am
Washington Resistance Army wrote:Imperialisium wrote:
I’m not speaking of ignorance. I know very well how margin of errors work given my background. The problem with NSG is the precedence of taking a generic outcome pertaining to the outcome of something and then extrapolating it to create an entire persona, narrative, and even argument for the person you’re disagreeing with.
“Polls were wrong because of outcome,” isn’t technically incorrect. They (which can be any number of specific polls) predicted the wrong outcome of who won. No more or less from that comment should and could be gleaned. But as you see no one wants to really discuss how that happened or why they disagree; or, could it feasibly occur in predicting the outcome of the 2020 election. So they instead go for the horrible fallacies of just invalidating the person by going “you’re ignorant, you don’t know math, you are X here.”
But that's the thing, apart from Wisconsin the polls weren't really wrong. Pretty much everywhere else, including how the national vote numbers went, the polling was perfectly within margin of error. You can't have much better polling than that.
Admittedly yeah a bunch of the election models should be mocked (like the one that gave Clinton a 99% chance of winning) because a lot of them ignored very glaring flaws in the Clinton campaign and the background in the Rust Belt that led people like me to say Trump was going to get a narrow victory there, but again the polling was mostly solid.
by Fartsniffage » Fri Oct 02, 2020 10:27 am
US media are reporting that Covid-positive President Trump has cancelled his only scheduled event on Friday as he and the first lady isolate in the White House.
The event was set to be a phone call on Covid-19 support to vulnerable older people.
by Thermodolia » Fri Oct 02, 2020 10:27 am
Imperialisium wrote:Washington Resistance Army wrote:
But that's the thing, apart from Wisconsin the polls weren't really wrong. Pretty much everywhere else, including how the national vote numbers went, the polling was perfectly within margin of error. You can't have much better polling than that.
Admittedly yeah a bunch of the election models should be mocked (like the one that gave Clinton a 99% chance of winning) because a lot of them ignored very glaring flaws in the Clinton campaign and the background in the Rust Belt that led people like me to say Trump was going to get a narrow victory there, but again the polling was mostly solid.
Hmmm I see. Wouldn’t have any spitballing of MoEs and how pollsters and/or aggregates got those figures lying around would you?
by Thermodolia » Fri Oct 02, 2020 10:27 am
Fartsniffage wrote:US media are reporting that Covid-positive President Trump has cancelled his only scheduled event on Friday as he and the first lady isolate in the White House.
The event was set to be a phone call on Covid-19 support to vulnerable older people.
Perhaps his symptoms aren't so minor?
by Imperialisium » Fri Oct 02, 2020 10:31 am
by Valrifell » Fri Oct 02, 2020 10:33 am
Imperialisium wrote:Valrifell wrote:
Polls are not predictive, it's quite that simple. They are snapshots of support at any given point in time, and we saw that as we got closer to election day Trump and Clinton closed in on each other in a few key states to the point it was within the MoE, which coupled with the high percentage of undecideds that we caught, makes it (in hindsight) relatively unsurprising that the election shaped out as it did.
The polls right before election day were pretty much right! They captured the correct level of support between the candidates within error! That is the mark of a successful measurement!
My issue isn’t so much with the math as their methodology must have been flawed. Which in turn makes their math give a less accurate outcome. I wouldn’t be surprised if the MoE was actually higher than the 4.5% someone else posted here (if that percentage is accurate). But that may be predicated on what they had to work with in terms of numbers, accountable variables, and whatever formula they use.
From a forensic background stand point 4.5% MoE is extremely high. At least uncomfortably so and is pretty close to the 5% margin to be considered accurate at all.
by Valrifell » Fri Oct 02, 2020 10:34 am
Fartsniffage wrote:US media are reporting that Covid-positive President Trump has cancelled his only scheduled event on Friday as he and the first lady isolate in the White House.
The event was set to be a phone call on Covid-19 support to vulnerable older people.
Perhaps his symptoms aren't so minor?
by Kexholm Karelia » Fri Oct 02, 2020 10:34 am
Fartsniffage wrote:US media are reporting that Covid-positive President Trump has cancelled his only scheduled event on Friday as he and the first lady isolate in the White House.
The event was set to be a phone call on Covid-19 support to vulnerable older people.
Perhaps his symptoms aren't so minor?
by Cordel One » Fri Oct 02, 2020 10:37 am
by Imperialisium » Fri Oct 02, 2020 10:38 am
Valrifell wrote:Imperialisium wrote:
My issue isn’t so much with the math as their methodology must have been flawed. Which in turn makes their math give a less accurate outcome. I wouldn’t be surprised if the MoE was actually higher than the 4.5% someone else posted here (if that percentage is accurate). But that may be predicated on what they had to work with in terms of numbers, accountable variables, and whatever formula they use.
From a forensic background stand point 4.5% MoE is extremely high. At least uncomfortably so and is pretty close to the 5% margin to be considered accurate at all.
And I've already mentioned that their methodology was off in WI, due to undersampling.
Your background as a forensic scientist is irrelevant which talking about more pure forms of statistics and mathematical modeling and when what we're measuring is something entirely unrelated to forensics. The MoE in 2016 was average for political science, it's just how that works, and has been like that since the advent of modern polling science.
by Zurkerx » Fri Oct 02, 2020 10:38 am
Fartsniffage wrote:US media are reporting that Covid-positive President Trump has cancelled his only scheduled event on Friday as he and the first lady isolate in the White House.
The event was set to be a phone call on Covid-19 support to vulnerable older people.
Perhaps his symptoms aren't so minor?
by Imperialisium » Fri Oct 02, 2020 10:40 am
by Shrillland » Fri Oct 02, 2020 10:40 am
by Fartsniffage » Fri Oct 02, 2020 10:41 am
by Northwest Slobovia » Fri Oct 02, 2020 10:43 am
Imperialisium wrote:Hmmm I see. Wouldn’t have any spitballing of MoEs and how pollsters and/or aggregates got those figures lying around would you?
by Imperialisium » Fri Oct 02, 2020 10:45 am
Northwest Slobovia wrote:Imperialisium wrote:Hmmm I see. Wouldn’t have any spitballing of MoEs and how pollsters and/or aggregates got those figures lying around would you?
Better: here's a long list of current polls. If you click on any of the polls linked in the first column, it'll take you to all the polls for that state. For example, Florida, where we see that most MoEs are 3.5-4.5, but St. Pete Polls, whoever they are, likes big samples, and has it down to 1.8.
They also have 2016 polling still available, if you really wanted the data from last time. (WI listed in "leans Clinton" for example).
by Valrifell » Fri Oct 02, 2020 11:00 am
Imperialisium wrote:Valrifell wrote:
And I've already mentioned that their methodology was off in WI, due to undersampling.
Your background as a forensic scientist is irrelevant which talking about more pure forms of statistics and mathematical modeling and when what we're measuring is something entirely unrelated to forensics. The MoE in 2016 was average for political science, it's just how that works, and has been like that since the advent of modern polling science.
Yeah and I’m also saying their methodology was flawed.
Nor is how math works somehow different from field to field.
Advertisement
Users browsing this forum: Aadhiris, Duvniask, Elejamie, GMS Greater Miami Shores 1, Ifreann, Kannap, Kaumudeen, Kaztropol, Kerwa, Kreushia, Socalist Republic Of Mercenaries, The Jamesian Republic, Uiiop, Uvolla, Valrifall, Zurkerx
Advertisement