The Demise of Polls

[subgenius]

No matter who wins the President election of 2020, I do not see how the pollster industry survives. They once again missed the target in dramatic fashion. Is there any redemption for what are amounting to nothing more than political horoscopes?

[canpakes]

Agreed; this was no shining moment for 538 or so many others.

[Xenophon]

Agreed; this was no shining moment for 538 or so many others.

I'm curious what your particular issue with 538 was. Assuming Nevada and Michigan finish out the way they are trending their model is going to be nearly entirely spot on. Results look like they are going to fall right into the distribution models for both 538 and the Economist. The only true anomaly was they expected Florida to be a tighter race but that seems to be due to bad pollster considerations around Cuban voters. I agree that individual polls were fairly weak this time around but the actual modeling tools that take those polls, run them through thousands of simulations and provide probability outcomes were fairly on point this year.

ETA: Their modeling data below. A significant segment of their results falls right in line with where the results are going to be regardless of who wins. Yes they pollsters messed up again in over stating Biden's position but it seems most of the models out there are going to have done a great job accounting for that.


https://projects.fivethirtyeight.com/20 ... -forecast/

[Icarus]

The RCP average had Trump + 0.2 in North Carolina
Biden +0.9 in Florida
Trump +1.0 in Georgia
Biden +0.9 in Arizona

All within the margin of error I think.

Of course, Wisconsin and Michigan were obviously way out of whack.

[canpakes]

Xeno -

I’m thinking along the lines of margin of error. 538 seemed overly optimistic for my jaded senses when Nate was on an interview with Rick Wilson and Molly Jong-Fast on the New Abnormal podcast last week. This looks like much more of a squeaker than predicted and runs to the margins of ‘margin of error’. : ) But I haven’t taken a deep dive into their analysis; I’m just going off of his conversation. It may be that 538 ends up looking better than that sounded.

ETA: sorry, Xeno - it was an Ezra Klein podcast. Not enough sleep these busy days, lol.

[Themis]

The RCP average had Trump + 0.2 in North Carolina
Biden +0.9 in Florida
Trump +1.0 in Georgia
Biden +0.9 in Arizona

All within the margin of error I think.

Of course, Wisconsin and Michigan were obviously way out of whack.

I'm wondering if voter suppression is taken into account in polling.

[Res Ipsa]

Agreed; this was no shining moment for 538 or so many others.

I’m perfectly happy to wait and see the final vote tallies before condemning 538 to the dustbin. Too many people interrupted his model’s forecast of 90% chances of a Biden win as a prediction of a Biden landslide. That’s not correct. That 90% included lots of very close outcomes, including 270-268. If I recall correctly, the model predicted only about a 30% chance of a Biden landslide.

There’s no question polling has gotten harder, especially if supporters of one of the presidential candidates have become convinced that pollsters are part of the enemies of the people. There will be plenty of time for the polling postmortem once the votes have been counted.

[Res Ipsa]

The RCP average had Trump + 0.2 in North Carolina
Biden +0.9 in Florida
Trump +1.0 in Georgia
Biden +0.9 in Arizona

All within the margin of error I think.

Of course, Wisconsin and Michigan were obviously way out of whack.

I'm wondering if voter suppression is taken into account in polling.

It should be reflected in the likely voter screens, although it may not be expressly labeled.

[Doctor CamNC4Me]

It's funny how the GOP was trying to get the drive-through voting tossed out in Texas, in a primarily Democrat stronghold district, while here in Utah, up in Layton county, they didn't say bupkis about the drive-through voting.

Huh. Weird.

- Doc

[subgenius]

It's funny how the GOP was trying to get the drive-through voting tossed out in Texas, in a primarily Democrat stronghold district, while here in Utah, up in Layton county, they didn't say bupkis about the drive-through voting.

Huh. Weird.

- Doc

Yes, weird how politics and elections have a tactical aspect to them...unlike your thread derail here.