Monday, September 09, 2024

Polls Are 💩

That poll said: The NYT/Siena poll everybody was talking about over the weekend, said: Except: Yeah, I know, Nate Silver 'models' the polls and comes closer to the truth ("close" only counts in horseshoes and hand grenades), but the idea that one poll tells us what the nation thinks, and another poll later tells us the nation has changed its mind, rather than the simpler explanation that if you ask less than 1000 people a question, and ask another group of people the same (or nearly the same) question, you might get different answers, is never considered.

We all know what if you put four people on four corners of one intersection and ask them what happened when there's a collision, you'll get four different answers.  And a few days later you'll get different answers, still. I took a psych class in college (I had a LOT of electives in my degree plan), and during the lecture in a lecture hall (v. a standard class room), a man burst in yelling, ran at the professor screaming at him, and eventually ran out of the room.  Eventually we found out that was planned, to show how much "eyewitness" testimony depended on what you saw, what you heard, and what you remember.  Almost everyone in that classroom had a different version of what happened.  So if you ask different groups of people who should be POTUS, you should expect different answers.  But how do you say those answers indicate anything except the responses of different groups of people on different days? 

Especially since polls never really come close to what the actual vote count is after the election.  "Close," yeah; but close only counts in horseshoes and hand grenades.

You might as well diving the future from chicken entrails.  And check several chickens; divination is tricky that way.
Posted by at

No comments:

Post a Comment

Subscribe to: Post Comments (Atom)