Boy, I have to agree wholeheartedly with the idea that journalists are misusing polls! Especially when there are glaring inconsistencies in the crosstabs (I’m lookin’ at *you*, NYT/Siena!)
weird crosstabs don’t necessarily mean the topline is wrong. for example, pollsters often have trouble getting enough GOP and enough Black ppl in samples. when they try to correct they may overrepresent Black Republicans—but the samples of Ds and Rs are still basically accurate so the topline numbers can be right even if the race crosstabs are wonky.
But *there’s* the fundamental problem, Noah: not enough participants from given groups, so they rely more and more on their algorithms to make up for it. To me, that’s why it’s hard to rely on just the topline, as their algorithms can be skewed, which ends with results like a “red wave” that *never* materialized.
but the 2022 polls were very accurate! the problem that year wasn’t that polls said there’d be a red wave; it’s that polls said there would be no red wave, but pundits didn’t believe the polls!
Sorry! You’re right, ie other than Simon Rosenberg, most of the media was playing into the “red wave” narrative and not the polls. It was 2020 that the polling was off significantly and hence why many were skeptical of the 2022 polling and spun their own story instead…
yep. 2020 polls undercounted Rs by about 5 points. then people figured polls were undercounting again in 2022…but they weren’t. (not in 2023 either; polls were pretty on the money then too.)
Whatever the news says about politics and polls is for $$ or to appease the advertisers.
Anytime I hear a poll result, or anything where I don’t have access to the questions and the raw data, I think of the book 1984. All of it is doublespeak.
And please, never ever pay any attention to Nate Silver.
to some degree. journalists also don't always do a great job of explaining how polls work, or of using them responsibly.
Boy, I have to agree wholeheartedly with the idea that journalists are misusing polls! Especially when there are glaring inconsistencies in the crosstabs (I’m lookin’ at *you*, NYT/Siena!)
weird crosstabs don’t necessarily mean the topline is wrong. for example, pollsters often have trouble getting enough GOP and enough Black ppl in samples. when they try to correct they may overrepresent Black Republicans—but the samples of Ds and Rs are still basically accurate so the topline numbers can be right even if the race crosstabs are wonky.
But *there’s* the fundamental problem, Noah: not enough participants from given groups, so they rely more and more on their algorithms to make up for it. To me, that’s why it’s hard to rely on just the topline, as their algorithms can be skewed, which ends with results like a “red wave” that *never* materialized.
but the 2022 polls were very accurate! the problem that year wasn’t that polls said there’d be a red wave; it’s that polls said there would be no red wave, but pundits didn’t believe the polls!
Sorry! You’re right, ie other than Simon Rosenberg, most of the media was playing into the “red wave” narrative and not the polls. It was 2020 that the polling was off significantly and hence why many were skeptical of the 2022 polling and spun their own story instead…
yep. 2020 polls undercounted Rs by about 5 points. then people figured polls were undercounting again in 2022…but they weren’t. (not in 2023 either; polls were pretty on the money then too.)
Whatever the news says about politics and polls is for $$ or to appease the advertisers.
Anytime I hear a poll result, or anything where I don’t have access to the questions and the raw data, I think of the book 1984. All of it is doublespeak.
And please, never ever pay any attention to Nate Silver.