I think people need to be really careful with the assumption that the majority of polls moving in one direction = a trend.
There’s a well known phenomena in polling known as herding where polls will tend to cluster in groups because as a pollster it is dangerous to your reputation to be a significant outlier.
The best example I can think of of this is the 2017 UK general election. All of the polls clustered around the same conclusion which was a comfortable victory for May. The only major pollster that got anywhere close to Labour’s 40% vote share was Survation. The rest were all out by at least 5%. They had herded themselves completely in the wrong direction.
I have shared research like this on here before but “perceived momentum” has a measurable impact on polling:
In short, people think that a party is more likely to perform well if polling momentum is in their favour, this makes them more likely to say they’re going to vote for that party even if they won’t (there are plenty of people who just want to be on the winning side). This creates falsity in the results because often these people are answering polls based on perceived social pressure and not their true voting intention. It is a small effect but it can make all the difference and it can generate these weird feedback loops where a candidate seems to be gaining steam in the polls but it is all a complete social illusion.
business.leeds.ac.uk
(This is super interesting btw)
I think the reality is that most US polling methods are simply not fit for purpose. The margin of error extends beyond their usefulness and it is really hard to draw anything concrete for them. They will be within 5% but when the race is going to come down to a 2-5% max lead that renders a lot of them of limited use. Polling in Europe and the UK is better (still not perfect).
What is of most interest to me is the Dems are not releasing polling they’ve funded with a few exceptions. GOP are releasing them several times a day. That’s weird to me and it suggests they are both seeing information that the rest of us aren’t.
x.com