In Reply to: Re: The stuff about the polls excluding people likely to vote Dem posted by russsmith on October 10, 2024 at 18:39:36
Again, not crazy - small sample sizes lead to misleading results.
Of course, a single poll's cross tabs are not going to be equal to the mean of the cross tabs of multiple polls. That is not necessarily a product of a poor model but frankly, what happens whenever you do random polling of a small number of samples.
A typical poll of 1000 votes is completely wrong around 1 out of 20 times. Blacks make up 14% of the population. How correct do you think a sample of 140 people are going to be in predicting the outcome of 8.5 million voters in MI?