https://x.com/IAPolls2022/status/1752775226612474145
🇺🇲 2024 GE: @QuinnipiacPoll (15)
— InteractivePolls (@IAPolls2022) January 31, 2024
Nov. 2023
(R) Trump: 48% (+2)
(D) Biden: 46%
.
Dec. 2023
(D) Biden: 47% (+1)
(R) Trump: 46%
.
Jan. 2024
(D) Biden: 50% (+6)
(R) Trump: 44% https://t.co/o5quaNGa46 pic.twitter.com/6jP0KUpUp8
!chuds in shambles
Jump in the discussion.
No email address required.
Like normal, this poll comes down to party lines.
Rep: 91% Trump, 7% Biden
Dem: 2% Trump, 96% Biden
Which means the poll depends on their weights for Dem/Rep.
See, they assume there is some population party split like 25% Rep, 40% Dem, and 35% Ind.
Then they do the polling and might gather from the calls something like 35% Rep, 35% Dem, 30% Ind.
Then they say "oopsie, better match up our sample to the population". And weight the Rep answers less and the Dem more to bring it inline.
So like usual, these polls really all depend on what numbers they are using for the party splits at the population level. And there were some really weird ones back in 2016.
Source: Statistician, you can make data say whatever you want. We call it "interrogating the data".
Edit: I would guess they changed their weights with the new year, which prompted the shift.
Jump in the discussion.
No email address required.
Brilliant observation, Mr. Silver.
Jump in the discussion.
No email address required.
My pleasure. That'll be $538 pl0x
Jump in the discussion.
No email address required.
More options
Context
More options
Context
Hillary Clinton has an 85% chance to win. Trust The Science.
Jump in the discussion.
No email address required.
haha holy shit
Jump in the discussion.
No email address required.
More options
Context
It was still like 90% an hour or 2 before they called it
People hate on 538 but they communicated it better than anyone
Jump in the discussion.
No email address required.
More options
Context
More options
Context
loooool cope
Source: A better statistician than you. I bet you got your PhD in bongland or some shithole like that
Jump in the discussion.
No email address required.
Never said I was a good one. I'm sure the data shows you are better than I.
Jump in the discussion.
No email address required.
I did a snowball sample survey, everyone says I'm the best statistician in my high school
Jump in the discussion.
No email address required.
How long did it take to give everyone a snowball?
Jump in the discussion.
No email address required.
More options
Context
More options
Context
darn straight
Jump in the discussion.
No email address required.
More options
Context
More options
Context
You sounded smarter as a fake nurse than a fake statistician tbh
Jump in the discussion.
No email address required.
I never claimed to be a nurse, but do continue coping
Jump in the discussion.
No email address required.
Coping about what?
Jump in the discussion.
No email address required.
Jump in the discussion.
No email address required.
Idk u said continue to cope?
Jump in the discussion.
No email address required.
More options
Context
More options
Context
More options
Context
More options
Context
More options
Context
More options
Context
Last 2 polls on 538 https://projects.fivethirtyeight.com/polls/president-general
"not sure" is the nooticers best stat
Also black people are the only demographic left who still sees Biden as doing a better job than not
Yougov https://d3nkl3psvxxpe9.cloudfront.net/documents/econTabReport_f5kOruS.pdf
Jump in the discussion.
No email address required.
More options
Context
They don't really need to assume, pretty sure party registration is public or you can pay for the lists.
Jump in the discussion.
No email address required.
More options
Context
More options
Context