I'm less kind to the totality of polling this year. A smaller miss than 2020, but comparable to 2016. A few points in every swing state, even more in the safe states, several points in most senate races, all missing in the same direction? And we're supposed to look to these pollsters again in future elections? But credit where it's due, NYT/Siena nailed T+13 FL, and called TX and AZ shooting to the right despite a lot of pushback. They still whiffed bad in the north, but others were worse. Marist had a Scott +2 poll in FL. Most other pollsters, particularly ones that weighed on recalled vote, were terrible. The worst pollsters were lesser known names like Focaldata, Research Co, RMG, Bullfinch, and random university polls. Quinnipiac was all over the map (like usual). Never trusting St Pete Polls in FL again.
Also, Arizona is already down to Trump +5.6 while it was Trump +6.4 when the chart was made. Pennsylvania also might move left a few tenths after they count the extra votes. With those adjustments, the state level polling wasn't too bad with the worst miss being about 3.4 points.
Edited: I recommend reading the linked article, it's very good.
I'd take the different view from you. With a ~2-3 point miss across the board, polling did really well. We broadly wouldn't be critiquing the polling industry if the miss had been in our favor.
There are inherent inaccuracies to polling, and considering all the fundamental issues that the polling industry faces today, being more accurate than the statistical margin of error is *good*. They did a good job. Yes, there is absolutely room for improvement as the error was all in a single direction, but this is in line with what we would see 10-15 years ago when we had no qualms with the polling industry.
If I had internalized the polling numbers more fully, I wouldn't have been so shell shocked by the election results. Polling told us it was razor close and we could very well lose. The end result was razor close and we, unfortunately and much to the detriment of my ability to be unstressed, lost.
A run down on how polling performed: https://split-ticket.org/2024/11/10/2024-showed-the-value-of-polling/
I'm less kind to the totality of polling this year. A smaller miss than 2020, but comparable to 2016. A few points in every swing state, even more in the safe states, several points in most senate races, all missing in the same direction? And we're supposed to look to these pollsters again in future elections? But credit where it's due, NYT/Siena nailed T+13 FL, and called TX and AZ shooting to the right despite a lot of pushback. They still whiffed bad in the north, but others were worse. Marist had a Scott +2 poll in FL. Most other pollsters, particularly ones that weighed on recalled vote, were terrible. The worst pollsters were lesser known names like Focaldata, Research Co, RMG, Bullfinch, and random university polls. Quinnipiac was all over the map (like usual). Never trusting St Pete Polls in FL again.
Also, Arizona is already down to Trump +5.6 while it was Trump +6.4 when the chart was made. Pennsylvania also might move left a few tenths after they count the extra votes. With those adjustments, the state level polling wasn't too bad with the worst miss being about 3.4 points.
Edited: I recommend reading the linked article, it's very good.
I'd take the different view from you. With a ~2-3 point miss across the board, polling did really well. We broadly wouldn't be critiquing the polling industry if the miss had been in our favor.
There are inherent inaccuracies to polling, and considering all the fundamental issues that the polling industry faces today, being more accurate than the statistical margin of error is *good*. They did a good job. Yes, there is absolutely room for improvement as the error was all in a single direction, but this is in line with what we would see 10-15 years ago when we had no qualms with the polling industry.
If I had internalized the polling numbers more fully, I wouldn't have been so shell shocked by the election results. Polling told us it was razor close and we could very well lose. The end result was razor close and we, unfortunately and much to the detriment of my ability to be unstressed, lost.