My fundamental issue with the presidential polls is that they simply didn’t pass the smell test for the last year or so. Consider this Fox News poll (while Fox News is awful, their polling is recognized as very good; their news celebrities just ignore the inconvenient stuff):
In the stretch between Nov 2023 and May 2024, there were several time points where Trump had a four or five point advantage (and Trump has that advantage even when including spoiler candidates). That would mean Trump would win the popular vote by five to seven million votes, depending on turnout.
That’s just not possible–and that’s not me hitting the hopium pipe. Not only has Trump lost the popular vote twice, but Democrats have won more popular votes in the last seven out of eight presidential races. Yet the polling data, were they to be believed, would suggest (if voter turnout remained the same) a twelve to fourteen million vote swing. That really doesn’t pass the smell test. At all.
Mind you, this doesn’t say much about the chances of Biden winning the Electoral College–Biden needs a majority in the ‘right’ states, but those estimates of the popular vote, in hindsight, are surreal. Something seems very broken with polling right now. Certainly, commentators need to understand the limitations of these data–and that includes the recent good news for Democrats.
Aside: One thing worth noting is that the polled respondents did not rank issues favorable to Democrats as more important, so I really don’t understand what is happening.
mikethemadbiologist
There’s still a lot of polling done by phone, and that runs face first into a stark generational divide in telecom habits. People with a land line they actually answer are not exactly an unbiased pool to sample. It got a little better with some cellphone coverage, but still slanted.
Then the internet polls are terrible quality because people mess with them or don’t take them seriously (because of course the internet is for porn and trolling). In-person ‘sample the common man on the street’ voice polls are too expensive and too variable to compensate., since they depend heavily on location and time of day.
At this point Alphabet and Meta probably know better how the next election’s going to go than any of the polling outfits, mainly because they know what people have been searching and watching and buying and they have absolutely no compunction about using every last scrap of data for profit.
Pingback: Mike's Blog Roundup ... from Crooks & Liars Steve in Manhattan - Tom Bettenhausen's
Pingback: Links 7/4/24 | Mike the Mad Biologist
Pingback: “Black Jobs”: Trump Tried to Do a Racism, but His Brain Is Too Broken | Mike the Mad Biologist
Pingback: In Case You Missed It… | Mike the Mad Biologist