Perhaps we are celebrating a bit too hastily to poll numbers we love. A caveat is in order here. I have some experience with survey research, some postgraduate courses, a stint as an ICPSR (interuniversity consortium for political and social research ) representative and currently as a telephoner for a polling subcontractor.
There's many a chance for sampling error in telephone polling these days. Some are new and some existed even back when the fingers did the dialing and recording. I do not know when pollsters give sampling moe (margin of error)if they account for outdated sample lists which include businesses, wrong numbers and dead people. Presumably, but not necessarily, samples are weighted for age, because a majority of responders are retired and quite a few are in their eighties. Stands to reason, many retirees have time on their hands and enjoy talking. And it is expensive to interview the myriad of non-english speakers such as Russian, Chinese and of course Spanish.
As for bias. Much depends on the wording of the question as others have pointed out. Since the majority of polling calls are rejected, (some vehemently and a few obscenely) what can we surmise about those who do respond? As I noted, the elderly respond with relish. More women than men ;white over black. Although most polls include likely voters, based on past record and registration, how truthfully do people respond when asked, "did you vote in the last election.
so, if you rely on polls, find out more about their sample and their methodology...available at mydd.