IIRC they always did worse than in polls, am I really remembering wrong?
I’m too lazy to evaluate data for state elections, but in federal ones the AfD got less than expected in 2021 and 2013, but more than expected in 2017. The underestimations were both less than a single percentage point, the overestimation was closer to two.
https://www.wahlrecht.de/umfragen/archiv/2021.htm
So it seems that our pollsters don’t really have much of an accuracy problem and only a small precision problem.
My guess is that you’re comparing it to the highest poll numbers because those obviously made headlines. “AfD slowly lost a few points over the last months” is boring and therefore less likely to be remembered (if that headline even exists).
Yeah, you are right. I did not want you to be right :(
Well, I hope I’m also right about my second paragraph: Time plays a role. We also have a snapshot here and it’s likely to change.
Also: me being right doesn’t necessarily mean you’re wrong. The polls do fairly accurately get what voters would do if the election were next Sunday, but the country and the voters change in run-up to an actual election. People are more interested in politics around that time, parties get more vocal about their ideas and there’s more about it in the media. All of that isn’t good for people who claim to have easy (but useless) ideas for complicated problems. That won’t stop an established demagogic party like the AfD, but I assume (too lazy to relearn R, so take this assumption with a grain of salt, I did not run the numbers), it puts the result closer to long term averages. So it’s indeed unlikely a spike like this will survive until election day. That said, 15% of these people in parliament would be embarrassing enough.
deleted by creator
I’m too lazy to evaluate data for state elections, but in federal ones the AfD got less than expected in 2021 and 2013, but more than expected in 2017. The underestimations were both less than a single percentage point, the overestimation was closer to two. https://www.wahlrecht.de/umfragen/archiv/2021.htm So it seems that our pollsters don’t really have much of an accuracy problem and only a small precision problem.
My guess is that you’re comparing it to the highest poll numbers because those obviously made headlines. “AfD slowly lost a few points over the last months” is boring and therefore less likely to be remembered (if that headline even exists).
deleted by creator
Well, I hope I’m also right about my second paragraph: Time plays a role. We also have a snapshot here and it’s likely to change.
Also: me being right doesn’t necessarily mean you’re wrong. The polls do fairly accurately get what voters would do if the election were next Sunday, but the country and the voters change in run-up to an actual election. People are more interested in politics around that time, parties get more vocal about their ideas and there’s more about it in the media. All of that isn’t good for people who claim to have easy (but useless) ideas for complicated problems. That won’t stop an established demagogic party like the AfD, but I assume (too lazy to relearn R, so take this assumption with a grain of salt, I did not run the numbers), it puts the result closer to long term averages. So it’s indeed unlikely a spike like this will survive until election day. That said, 15% of these people in parliament would be embarrassing enough.
deleted by creator