Wouldn't that depend on who and what you are polling and also the make up of your control group? I think if you have a control group polling would be more accurate than random polls...Say for instance you have a poll of 100 people who party in San Francsisco and you ask them how many of them give money to random bums on the street to find them some pot, I'm sure you'd get a fairly accurate poll of people who've done that... fah q, terry. fah q. ;D