>That was my point though - it doesn't matter whether 1 or a million people respond, the results can never be valid. There is no correlation between my pain problems & computer use. Similarly, other external influences are affecting my currently limited social life. Without asking more questions about the reasons for things, the results of the survey are completely groundless.
Despite my previous protests against chain letters: I think that interviewing many people
can establish a correlation - or the lack of it. If a correlation is found, some interpretation is up to the statistician: whether (1) computer use causes back pain, (2) back pain makes people use computers, or (3) both are caused by some third factor, perhaps not included in the study. Like our "modern way of life".
If your back pain is not related to computers, there probably are, as compensation, other users who suffer from back pain and do not use computers. This means that if a sufficiently large sample is used, meaningful data will be obtained. One problem is that the sample should be randomly chosen. This doesn't fully apply: only computer users will go to the Web site. However, a correlation
may still be obtained between computer use (average hours per day) and back problems. Or other problems.
Hilmar.
Difference in opinions hath cost many millions of lives: for instance, whether flesh be bread, or bread be flesh; whether whistling be a vice or a virtue; whether it be better to kiss a post, or throw it into the fire... (from Gulliver's Travels)