ukusagolf
Senior Poster
Posts: 206
Joined: Jan 2019
Reputation: 4
|
RE: US Election & Presidency
(03-11-2024 21:59 )Goodfella3041 Wrote: That Iowa poll has really caused a stir because the pollster has been one of the most accurate out there. In every major election since 2008 — presidential, midterm, gubernatorial, senate, etc. — she has only had one significant miss.
All the same, the polling average in Iowa is still R+6.
Tighter than it was in 2016 or 2020, but still pretty safely Red.
Polling guru Nate Silver had an interesting interview recently
He said “If a pollster never publishes any numbers that surprises you, then it has no value"
So are polls meant for entertainment value? Click Bait?
It was about how the polls are too similar. Is it that close, or are most pollsters lazy and all come to the same conclusion, or afraid to stick out their neck to upset their audience?
He also said "his gut feeling" is a Trump win. I thought the whole point of using data was to take emotion out of the equation
Give them less, they will pay more
Bring on the TARIFFS
(This post was last modified: 04-11-2024 01:13 by ukusagolf.)
|
|
04-11-2024 01:11 |
|
Goodfella3041
Posting Machine
Posts: 1,691
Joined: Feb 2011
Reputation: 61
|
RE: US Election & Presidency
I think Silver’s model prediction stands at 51% for Trump to 49% for Harris.
To be clear, that’s not a poll — that’s what his model predicts based (mostly) on polling data that is aggregated and weighted for things like sample size, recency and historic accuracy of the pollster. I think some economic and demographic fundamentals are baked into the model too.
If after all that, the model predicts 51-49, then it’s effectively a coin flip and you might as well go with your gut feeling.
I think what he means by that comment about polls that “surprise you” is that polls which are TOO consistent are inherently suspicious. Think of it this way… If you flip a coin ten times, you’d expect to get 5 heads and 5 tails. But if you repeated that process ten times and you got 5 heads and 5 tails EVERY time, that would be pretty strange. If the outcome were truly random, then you’d be bound to get a cycle or two that come out 6-4 or 3-7, etc.
Similarly, 3000 polls that all show a more or less 50/50 result suggest that the sampling isn’t as random as it should be, or that there are results that pollsters aren’t releasing because they look “abnormal”.
(This post was last modified: 04-11-2024 06:14 by Goodfella3041.)
|
|
04-11-2024 06:02 |
|