Yeah, but what drives the decision to only have "100" from a group?
I guess I sort of know the answer ... data from the last election I guess ... but doesn't that still bias the pool? Or at least disregard current data?
If lots of people 45-54 are willing to answer the poll, and no one from 25-34 is willing to answer ... is that "new data" about the election that they are throwing away?
It seems to me that when polling, you'd want to get all the data you can get, and then normalize AFTER the polling is complete ... having all the data.
Not exactly the same thing, but I remember in my photo class, the teacher said "Don't edit in the camera ...." meaning take as many shots as you can, and then select from them later ... because you can never select the photo you didn't take.
I guess is surprised me that the "right" mechanics was to NOT gather someone's poll at all....