[cdcdawg]

She won by 9 points. That wasn’t close. Pollsters are frauds. She was supposed to lose by 5. They just lie. It’s not more complex than that. Quinnipiac had her losing by 12. Only 21 points off.

[Zhang Fei]

[She won by 9 points. That wasn’t close. Pollsters are frauds. She was supposed to lose by 5. They just lie. It’s not more complex than that. Quinnipiac had her losing by 12. Only 21 points off.]

---

I doubt they do. They dazzle us with a blizzard of rationalizations, but what they do is similar to witchcraft. Nobody knows nothing*. The basic assumption of polling is that the profile of the people who respond is similar to the voting population. They are discovering that this is untrue to the point that their numbers are worthless either in themselves or as trend indicators.

* This was what an Oscar-winning scriptwriter said about the movie business. His favorite movies - the ones he was involved in - were commercial duds. His least favorite ones were blockbusters. But it might as well apply to the polling biz. Trying to divine the intentions of a voting population of hundreds of thousands from a random sample of a few hundred may be impossible. The intent is to get a random, representative sample. But the kinds of people who respond to polls may not be representative of the larger milieu.