Originally Posted by 80ktsClamp

The problem is the selective participation rates on a survey...even online. Usually 97% or so will vote on a contract, but giving the time to put in a detailed enough survey is something only around half will do (for varying reasons).

Unless that selection bias is corrected, unfortunately the closest to statistically valid polling you'll get is via random selection. Even with the phone polling, there is some selection bias which has to be accounted for (since you have to answer the phone and many more in my generation don't bother to answer numbers they don't recognize).

Statistics are messy math...