I don't know how most polling firms establish contact with those included in polls. But a few years ago I received a call from Pew Research Center and answered a fair list of questions. They said I was being pre-screened and if selected I'd be included in future public opinion polls. Never heard from them again so I must have been the wrong demographic at that time.
If polling firms are using that approach and then getting agreement from a valid sample of people to participate in future polls, then they might have good samples poll after poll. But if they're trying to use random calls in polls, then it'd definitely be more and more difficult to put together a good sample at any given time due to cell phones, caller ID and all the ways folks now have to avoid calls they don't care to answer.
And I think many of these preference polls by group we see are still largely based on exit polls, which are fairly questionable as to accuracy.
They are propaganda!