On Fri, 06 Dec 2002 10:38:58 +0000, Robert J. MacG. Dawson wrote: > The problem with *any* hypothesis test for normality (or any other > "gatekeeper test", such as the infamous "F before t") is that it > answers the wrong question.
I agree with you on the essence of what you are saying, but I think your statement is too strong. In my case, I wanted to know whether my data was normally distributed or not because I wanted to set bounds on what would be considered normal (as in healthy) and what not. If the data was normal, I could use the normal distribution to determine these bounds. If it was not normal, I could use the original data for this. Ofcourse, for large N, the results would be the same if the data was normal but I didn't use the normal distribution. Since it showed to be not normally distributed, I calculated my 95% bound based on the data. Whatever is beyond this limit for some new measurement would be considered 'not healthy' with the knowledge that 5% of these people actually are healthy. The reason for testing for normal distribution was that if the data seemed to be normally distributed, a 95% bound based on the normal distribution would probably be more accurate. Another possibility may be that the distribution is supposed to be normal. If it is not, there has probably been some mistakes in the measuring procedure. For example, if some contamination has a large effect on the measurements, it could be that, due to contamination of samples, one tail of the distribution is much stronger than in the normal distribution. Yet another reason for doing this is when you are considering several distributions and you wonder which one may be the most appropriate. A completely other reason for doing this stuff is that it is, to some extent, fun. Koen . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
