>>Statistics still apply.
>
>John, a dozen different surveys with consistent results - that anywhere from 55% to 75% of physicians disagree with the law. In the absence of any serious sampling bias, that is enough (for me) for a conversion/translation that if all physicians in the U.S. were polled, you'd see similar numbers. In the 5 years since the entire discussion of AHCA began, I have yet to see anything of any significance to the contrary.
>
>And Americans right now (even ones who initially supported the law) are drawing their own conclusions as well.

Kevin,

The problem with the is that the populiation is for a variaty of reasons not random. John already explained that in the case how this was setup it relies on physicians to react on the poll.

This is very, very bad practise, because the reason for not replying might be strongly correlated to just on the of the factors you are trying to research.

I've done my part of doing some advanced statistics and can tell you that getting a reliable population to base your statistics upon is one of the most difficult things to do. Random sampling is difficult to achieve, but at least controling the sampling by calling physicians on the Phone and asking them is better than emailing them all (I know that I never answer a survey on email, but I Always take the time to answer a phonecall), but I recognize that busy physicians will be difficult to get hold off in any way you like it.

Only if your population is randomly sampled, you can extrapolate your results to something like 95% confidence intervals. Otherwise, its pure guesswork.