A bigger effect is that retired people tend to be home, and older people are likelier to do the poll, so polls are weighted by age. Other wise they would just be polls of people who answer the phone the most.

Everyone weights for age, sex and race. Some weight for educational level or income or region in a state or other things.

And as you weight for certain demographics other things fall into place. But there will always be distortions due to who answers phones, for whatever reason.

(There may be some self correction in methods, also. People who don't want to take a political poll are probably, on average, less likely to vote. People dodging bill collectors are probably, on average, somewhat less likely to vote. Not all distortions will run in the same direction so some even out a little.)

Polling works pretty well, but does not work perfectly.

are always a problem for legitimate telephone pollsters (as is the rise in the # of cellphones and the 'registered' vs 'likely' voter problem).

This is why polling is less than an exact science.

See:
http://www.electoral-vote.com/evp2012/Info/polling-faq.html
http://www.huffingtonpost.com/2010/10/05/likely-voters-how-pollsters-choose-them_n_751560.html

Some adjust samples to match census data for income distributions. Since pollsters only get 9% willing to talk to them - probably fewer for robocalls - it is a major mystery why polls so far have been so amazingly accurate.

but if I don't know the number I don't answer. If they want to leave a message ok, but if not It wasn't important.

Even the people who called to ask me if I mailed my absentee ballot yet didn't get an answer. BUT they left a message BEFORE & AFTER they got it.

But I think they just call people & whoever answers they take the data and work with it. Most of my family does the same thing.