[ballardw]

Having worked in survey research or with survey data for going on 14 years my employers and most of my clients would fire me for providing data with out confidence limits, sample and response sizes and a least a minimum of methodology. Even one page executive summaries would mention the sample size (number of people that actually responded to the survey) and usually either a cooperation or response rate. It tells a lot about the survey quality if only 1 of 50 people that were contacted actually participate.

I don't even see how most of the referenced surveys contacted people of they were an actual random sample.

The comments about surveying 3 states and projecting to the whole country are correct that it is unlikely they are truly representative of the country. Do not use a survey on smoking conducted in Virginia or Kentucky and report that as the rate for Utah or Idaho.

Won't even go into the subtle or not so subtle biases by the way the questions are phrased in most of these.