>>3582450
>he doesn't believe pollsters are purposefully influencing their work
this is half true. pollsters are well aware of these biases and well aware of how different polling techniques exploit response biases. they report poll numbers knowing full well that they are misreporting the true opinion, even if the methodology appears correct.
many polls also use completely broken methodology. Reuters/Ipsos is the worst. It is complete garbage. They actually can't use the term "confidence interval" because the methodology is invalid so they replaced this with a completely fictional term called "credibility interval." it is laughable from the perspective of a professional statistician.
Berkeley IGS polls (in California) are similar. Complete fiction. They are going to be 10 points off probably. Republicans are killing on early ballot returns in CA.