http://pjmedia.com/tatler/2012/09/30/we-are-the-91-only-9-of-americans-cooperate-with-pollsters/ This is a brief on the PEW report on the matter ~ may be written at a comprehension level you can deal with. If you can understand this then you may on to the PEW report itself. THIS WAS PUBLISHED LAST MAY ~ the fact you are unaware of it is suggestive of your being a rustic when it comes to statistical analysis.
Why are you obsessing about this? 9% cooperation rate is obviously plenty enough for scientific polling - especially polling that gets the right combination of land line and mobiles. We know this by how accurate it's proven to be in recent elections - despite what people like yourself claim.
I've given you a clear example in the 2012 elections of how accurate polling has become. The state polls, in particular, painted a highly accurate picture. An aggregate of the state polls produced a perfect 50 state prediction rate from people like Nate Silver. Conservative analysts who disputed the polls such as the Dick Morris, that ridiculous "unskewed polls" guy, etc, have admitted they were wrong and the polls and their methodology/models were right.
You're claim that the use of polling to determine public opinion has collapsed is downright silly.