It was my understanding that no pollster - except maybe a college poli sci class doing a poll as a class project — just calls up people and takes what they get as their sample. They all screen and/ or weight the raw sample to model the electorate that they think will actually vote. Wasn’t that what that recent NYT article said, the one that gave the same raw numbers to 5 pollsters and got 4 different results?
It should be added Gallup exited the polling business in part because the difficulty of obtaining truly random sampling at a time when most people have cell phones and tweet or use IM online.
Take me. I tweet and being a deaf person, I neither have a landline phone or cell phone. I do have Skype/Yahoo IM installed because text chat compensates for my hearing loss. Pollsters can’t reach me and therefore that’s part of the bias built into present polling methods.
When you aren’t reaching Internet-era voters, your polling results are likely to be way off. We’ll see that again in November. Polling is junk science and Ouija boards have more reliability. There we are.