|
Typo posted:Bernie is standing up to the think tank noeliberal number fuckerstains lanyards the working class hate, he is the most popular politicians in america. The workers both republican and democrats are behind him, just have a little faith that he's got this and don't let the capitalist propaganda polls fool you This but unironically (USER WAS PUT ON PROBATION FOR THIS POST)
|
# ? Sep 22, 2019 18:56 |
|
|
# ? Jun 8, 2024 07:00 |
|
crazy cloud posted:what gets me is how there is literally nobody willing to even attempt articulating how *any* of the "credible" professional polling companies' methodology is in any way rigorous. Here's the methodology: make 15 completely arbitrary guesses that are based on nothing but your gut, and then conduct a poll with the guesses you made taken as an immutable given, fiddle with your weightings according to the likely voter screen you made up, which si another completely arbitrary guess, and then hit send I mean, the only possible methodology they can really use is to assume that the same people will vote who voted in the past. It's inherently incapable of predicting anything that would result from a change in voting patterns.
|
# ? Sep 22, 2019 18:59 |
|
crazy cloud posted:i have a bachelors degree in mathematics, friend, I understand how stats work. This is the thing where you are taking like, classroom econ101 and trying to apply it to late stage capitalism Yeah yeah, I have clas degree too, and took a while bunch of stats classes to go with it.
|
# ? Sep 22, 2019 18:59 |
|
mediaphage posted:Goonpoll 2020, lets start collating the questions now It's already at the top of the page op
|
# ? Sep 22, 2019 19:00 |
|
Ytlaya posted:I mean, the only possible methodology they can really use is to assume that the same people will vote who voted in the past. It's inherently incapable of predicting anything that would result from a change in voting patterns. yeah it's insane that there's no way to predict who is going to vote in 2020 but we still look at polls as real
|
# ? Sep 22, 2019 19:01 |
|
HootTheOwl posted:It's already at the top of the page op Well that’s what I get for only checking the most recent posts in my bookmarks
|
# ? Sep 22, 2019 19:02 |
crazy cloud posted:what gets me is how there is literally nobody willing to even attempt articulating how *any* of the "credible" professional polling companies' methodology is in any way rigorous. Here's the methodology: make 15 completely arbitrary guesses that are based on nothing but your gut, and then conduct a poll with the guesses you made taken as an immutable given, fiddle with your weightings according to the likely voter screen you made up, which si another completely arbitrary guess, and then hit send The credible outlets typically provide the screener, survey questions, detailed information about the sample, and collection methodology. Most don't need to weight the results based on sample. You can fire up excel and back into all their aggregate result using the provided crosstabs. You can re-weight the sample using the crosstab data for what-if scenarios if you so choose. It's not a vast conspiracy of numbers fuckersteins that think Bernie will take their SPSS.
|
|
# ? Sep 22, 2019 19:05 |
|
Bri Gray continues to be better than any of us deserve: (note her current handle) https://twitter.com/briebriejoy/status/1175826956320948224
|
# ? Sep 22, 2019 19:08 |
|
KingNastidon posted:The credible outlets typically provide the screener, survey questions, detailed information about the sample, and collection methodology. Most don't need to weight the results based on sample. You can fire up excel and back into all their aggregate result using the provided crosstabs. You can re-weight the sample using the crosstab data for what-if scenarios if you so choose. It's not a vast conspiracy of numbers fuckersteins that think Bernie will take their SPSS. oh okay then thank you
|
# ? Sep 22, 2019 19:11 |
|
liberal logic: the most popular candidate in the country can't win:
|
# ? Sep 22, 2019 19:12 |
|
KingNastidon posted:You can re-weight the sample using the crosstab data for what-if scenarios if you so choose. It's not a vast conspiracy of numbers fuckersteins that think Bernie will take their SPSS.
|
# ? Sep 22, 2019 19:14 |
|
Majorian posted:Bri Gray continues to be better than any of us deserve: (note her current handle) This “Bernie Sanders Bro are all racist” lie is going to be way way way less effective since team bernie actually has way more minorities supporting him then he did at the start of 2016.
|
# ? Sep 22, 2019 19:25 |
|
spunkshui posted:This “Bernie Sanders Bro are all racist” lie is going to be way way way less effective since team bernie actually has way more minorities supporting him then he did at the start of 2016. Bernie lost a lot of his white supporters from 2016 to the out-and-out white supremacist candidate, Joe Biden, in 2019. This is why Bernie is a lot of Biden voter's 2nd choice, it's not just name recognition.
|
# ? Sep 22, 2019 19:29 |
|
Typo posted:you don't get it, if polls were accurate, hillary would have clowned trump. stop trusting that number fuckerstain poo poo it's melting your brain like... the polling was pretty accurate, tho, why does this keep coming up?
|
# ? Sep 22, 2019 19:31 |
|
crazy cloud posted:Bernie lost a lot of his white supporters from 2016 to the out-and-out white supremacist candidate, Joe Biden, in 2019. This is why Bernie is a lot of Biden voter's 2nd choice, it's not just name recognition. he's lost a bunch of 2016 liberals to warren but they probably would have vote flipped in the general anyway if he's the nominee. It's better to focus on mobilizing disengaged voters who finally have a good and cool leftist candidate to vote for instead.
|
# ? Sep 22, 2019 19:31 |
|
Ytlaya posted:I mean, the only possible methodology they can really use is to assume that the same people will vote who voted in the past. It's inherently incapable of predicting anything that would result from a change in voting patterns. Reminder that this assumption was exactly why the pollsters shat their pants over and over in 2016
|
# ? Sep 22, 2019 19:35 |
|
Chilichimp posted:like... the polling was pretty accurate, tho, why does this keep coming up? National polls were, state-level polls weren't. That's even more the case when it comes to primaries.
|
# ? Sep 22, 2019 19:35 |
|
Majorian posted:National polls were, state-level polls weren't. That's even more the case when it comes to primaries. Voter suppression and EC gerrymandering is a bitch, ain't it like... polls weren't super turbo wrong, but pollsters were with their prognostications. The "there's not way Trump can win" brain rot was loving strong. Besides, polls are just one data point, and folks shouldn't hang on them, but we're all guilty of it.
|
# ? Sep 22, 2019 19:41 |
|
Chilichimp posted:Voter suppression and EC gerrymandering is a bitch, ain't it They certainly are, but there weren't any new voter suppression laws in Michigan, Florida, or Pennsylvania in 2016. The polls were still dramatically off and Hillary famously lost in all three of those high-EV states.
|
# ? Sep 22, 2019 19:46 |
|
Nonsense posted:https://twitter.com/kath_krueger/status/1175825037686968320 NYT readers so boug that property development is more relatable for them than, you know, human health, the thing actually being discussed. lol
|
# ? Sep 22, 2019 19:51 |
twodot posted:While you can do this in the sense that multiplication and division continue to exist, it is not actually a credible way to fix a sampling error. It's a future looking poll. You can't possibly know if there's a sampling error. If the belief is that some segment of the sample will turn out at slightly higher or lower rates than the sampled population then re-weighting is pretty valid. The only risk is creating higher error than what you would have had with larger n-size in that segment. I don't think every polling outfit is intentionally creating a bad sample to slight Bernard Sanders.
|
|
# ? Sep 22, 2019 19:57 |
|
I wonder how long it will take for the Democrats to just completely drop any and all attempts to appeal to POCs etc? Will they bother after we've been completely disenfranchised, or after we're all sitting in concentration camps because the cop who arrested me at a protest thought I looked "Mexican?"
|
# ? Sep 22, 2019 19:58 |
|
KingNastidon posted:It's a future looking poll. You can't possibly know if there's a sampling error. If the belief is that some segment of the sample will turn out at slightly higher or lower rates than the sampled population then re-weighting is pretty valid.
|
# ? Sep 22, 2019 19:59 |
|
The "White Terror" health care system.
|
# ? Sep 22, 2019 20:00 |
|
"it's also too small for everoyne to live in" "no one would rather be homeless than live in a house" whoever wrote this deserves to have violence perpetrated upon their person also, stop engaging typo, he's loving with you, this is his new gimmick, must've gotten bored of the "communist china true believer" one
|
# ? Sep 22, 2019 20:06 |
|
Doctor Jeep posted:
so I'm not allowed to change my opinions after reading the posts in this thread?
|
# ? Sep 22, 2019 20:14 |
|
Typo posted:so I'm not allowed to change my opinions after reading the posts in this thread? It has less to do with your opinions and more to do with the fact that you're an unfunny gimmick.
|
# ? Sep 22, 2019 20:15 |
|
Typo posted:so I'm not allowed to change my opinions after reading the posts in this thread? shut up, idiot
|
# ? Sep 22, 2019 20:17 |
|
mediaphage posted:Goonpoll 2020, lets start collating the questions now 1. Does your house have stairs in it?
|
# ? Sep 22, 2019 20:18 |
|
i made horse boy mad. https://twitter.com/HoarseWisperer/status/1175852394451742722 (USER WAS PUT ON PROBATION FOR THIS POST)
|
# ? Sep 22, 2019 20:21 |
|
Twitter really needs to probate people with velocity like Hoarse or Seth Abramson, nobody wants to read your stream of diharrea and likes to block out anybody actually replying to you.
|
# ? Sep 22, 2019 20:25 |
|
Nonsense posted:Twitter really needs to probate people with velocity like Hoarse or Seth Abramson, nobody wants to read your stream of diharrea and likes to block out anybody actually replying to you. he immediately blocked then got so mad he unblocked so he could quote tweet me to his followers lmao
|
# ? Sep 22, 2019 20:27 |
twodot posted:It is not. If you think a demographic is under polled, multiplying the under polled result does not give you better information. Can you explain the math behind why you believe this? Let's imagine a poll of 1000 people where 300 are under the age of 40, or 30%. You believe they'll actually make up 35% of voters. If all other factors (education, race, etc.) are not significantly different among under 40 and above 40 populations then the only harm in re-weighting is the error from projecting off a lower sample. But at a certain point the difference in MOE between 300 or 350 is negligible. And re-weighting doesn't affect the result in the older segment -- you oversampled that segment such that its segment level MOE is better than had you chosen 35%/65% split initially.
|
|
# ? Sep 22, 2019 20:29 |
|
KingNastidon posted:If all other factors (education, race, etc.) are not significantly different among under 40 and above 40 populations
|
# ? Sep 22, 2019 20:34 |
twodot posted:The reasons demographics are different is because they have different qualities. If you have under sampled a demographic, you have bad data. Multiplying bad data does not create good data. Not really. Let's take an extreme case. You run two separate polls of 1000 people each -- one under 40 and one above 40. The other sample demographics beyond age are reflective of that age group. You get results from those two polls. The polling data for each age segment has low MOE because of high sample size. Now you need to decide what split between under 40 and above 40. Take the split and weight the results accordingly. Why is this somehow less accurate than if you had polled 350 people under 40 and 650 people over 40 in a single 1000 person poll? Assuming the non-age demographics are consistent across both polls, which there's no reason they can't/shouldn't be.
|
|
# ? Sep 22, 2019 20:49 |
|
KingNastidon posted:Not really. Let's take an extreme case. You run two separate polls of 1000 people each -- one under 40 and one above 40. The other sample demographics beyond age are reflective of that age group. You get results from those two polls. The polling data for each age segment has low MOE because of high sample size. edit: it's also true that polling is useless for non-campaign managers, I'm just also saying that you can't fix bad polling by multiplying numbers together twodot fucked around with this message at 21:01 on Sep 22, 2019 |
# ? Sep 22, 2019 20:56 |
|
^^^ I think that you could take a poll with bad sampling and turn it into the "equivalent" of a poll with better sampling and a lower N, though I could be missing something. Like if you had a poll with 100 under 50 and 900 over 50 (for an extreme example) you could "transform" it by taking a random sample of the over 50 set better proportioned to the under 50 one. Though this is assuming that there's no other issue related to how each of those groups are sampled (that could bias it in favor of a certain type of person, etc). More important than the accuracy of polling is the purpose it serves. The recent Citations Needed episode does a good job of articulating why the reporting of polls serves no actual purpose to society and is actually harmful in the sense that it creates a feedback loop where people are constantly told that it's not realistic to support things that don't already have majority support (which has the effect of depressing support for any departure from the status quo, particularly in light of the fact that many liberals are convinced that left-wing politics are mutually exclusive with defeating Republicans). I like the term they use for the way things like polls factor into these discussions; they call it the "normative-descriptive shuffle" and use it to describe the rhetorical trick where people take arguments that are about morality and try to redirect them towards merely describing the way things currently are (with the implication that it's somehow bad or unrealistic to go against that). Ytlaya fucked around with this message at 21:01 on Sep 22, 2019 |
# ? Sep 22, 2019 20:58 |
|
Ytlaya posted:^^^ I think that you could take a poll with bad sampling and turn it into the "equivalent" of a poll with better sampling and a lower N, though I could be missing something. Like if you had a poll with 100 under 50 and 900 over 50 (for an extreme example) you could "transform" it by taking a random sample of the over 50 set better proportioned to the under 50 one. Though this is assuming that there's no other issue related to how each of those groups are sampled (that could bias it in favor of a certain type of person, etc).
|
# ? Sep 22, 2019 21:04 |
|
Does anybody know a succinct explanation of the democratic nomination process and the difference between a primary and a caucus? Preferably from a source that won’t make a non-super-politically-involved person flinch? Basically been talking with some folks and need a good link to send them via text message.
|
# ? Sep 22, 2019 21:11 |
|
|
# ? Jun 8, 2024 07:00 |
|
JIZZ DENOUEMENT posted:Does anybody know a succinct explanation of the democratic nomination process and the difference between a primary and a caucus edit: Also, in a primary, electioneering is prohibited. When you go to the polls, you are expected to have chosen your candidate and it's illegal for anyone to tell you or suggest to you who to vote for once you're there. In a caucus, candidates' supporters are allowed to discuss their candidate with other people and try to rally them into their caucus group until the head count is called. This Is the Zodiac fucked around with this message at 21:32 on Sep 22, 2019 |
# ? Sep 22, 2019 21:24 |