Originally posted by @metal-brain
Here is an excerpt from the link you posted:
"Approximately 6550 people were invited to participate in this survey, which took place in March and April 2012. Question 1 was answered by 1868 respondents"
Most people did NOT participate in the survey. This means the study is incomplete and unreliable. It is kind of like football players donating the ...[text shortened]... scientists participate in the poll the study is deeply flawed and not worth putting faith into.
Most people did NOT participate in the survey. This means the study is incomplete and unreliable.
Wrong.
It can be mathematically shown that a survey that doesn't ask most people (say it asks just 1% ) can, depending on conditions, nevertheless easily result in an estimate (of something) that is reliable within a few percent.
For example, a probability of less than, say, 1%, might creditably be rationally assigned to the estimate (of something) being inaccurate by more than, say, 5%. Why do you think that estimates are sometimes shown with error-bars?
I CERTAINLY should know because I have actual expertise in statistical analysis and I am actually currently doing cutting-edge research in, amongst other things, statistical analysis. I am even writing a book about it.
But you don't actually need to have such expertise to know about how estimations by scientists can be (and often are) reliable because that would be complete overkill; you only have to know the very basics of how science works with estimations which is something any first-year science student should know.