The study was conducted by the American Mosaic Project, a group attached to the sociology deptanrtment of the University of Minnesota, and funded by the David Edelstein Family Foundation.
There is very little I can find about this last oragnisation, strange in that most research organisations tend to heavily publicise there agendas, if only to raise their profiles for funding purposes, and even the most excessive christian fundie groups have web presense, so to find an organisation that does not seem to court web publicity is strange.
To quote any survey and to understand the results one should be aware of where and when the results were collected, it is easy to take the random element away from a survey, you simply have to know a little about the demographics of the population and target those areas here you are going to get the results you want.
For example, target the state of Utah, the chances are that you will hit upon a large number of Mormons, your survey is still random, but it is skewed by the population you have sought answers from.
Anotherway to put this, if I decide I want to do a survey on how trusted gays are in the UK, I could target Brighton on the south coast, this has a dispropotionately large gay population, so even from telephoning random numbers, I am going to get a deisproportionate amount of gays and therefore be able to claim a higher percentage, where as if I were to target the Yorkshire town of Barnsley, I could very well get a different result.
It is also about how the questions were asked and how they were answered and then how they were presented, this can really influence the answers to the point that it makes the survey pointless, a simple example, a survey was conducted into drug use at a UK school, one of the questions was on cocaine use, out of 500 pupils, one admitted cocaine use, when the survey was completed again a year later, two claimed cocaine use, one of our more vociferous tabloids ran with the headline, 'cocaine use doubles in secondary school in a year'. Technically correct but up from one to two is not a story.
What I am getting at here is simple, if someone claims a survey has said something, I want to read the survey, I want to know the funding of that survey, I want to know the questions, and I really want to know about the organisations involved, as much of this can tell me more about the results they have than the results themselves.
The analogy drawn about the KKK to my mind is correct, until you know all the facts oine can only asume that any survey is loaded in favour of the funders.
Edit:
The fact that you have chosen to go into a rant as soon as people challenge you shows that even you are not sure of the validity of both your own question and the development of the survey you claim results from, maybe you need to check the validity of your sources before you post in future, or take the chance that everyone theists and atheists alike will simply think your a bit of a d!ck!