In article <[EMAIL PROTECTED]>, Quito Quito <[EMAIL PROTECTED]> wrote: >Any input is welcome.
>[EMAIL PROTECTED] (Quito Quito) wrote in message news:<[EMAIL PROTECTED]>... >> Dear readers: >> Suppose I have computed that the 95% confidence interval for the >> difference, (u1 - u2), of means from two populations is (-3.2, 5.1) >> which contains the value 0, can I say that u1 and u2 are equal to each >> other? If so, what is the confidence level I can have to say so? No amount of information of this type allows you to make any such statement. If you have a prior belief that this is the case, you can use a Bayesian argument. But in most real problems, u1 and u2 are not equal, although they might be close enough that they should be treated as equal. In any case, decision-theoretic approaches, which do not use confidence levels as such, would be needed. The precise meaning of a 95% confidence interval is that the parameter under question will fall in the interval BEFORE the data is used is .95. It means exactly that, and nothing more, and this is assuming the assumptions are correct. Decision approaches are what are needed, not recipes. >> Similarly, I have computed that the 95% confidence interval for the >> ratio, sigma1^2 / sigma2^2, of variances from two population is (0.1, >> 2.9) which contains the value 1, can I say that sigma1^2 and sigma2^2 >> are equal to each other? If so, what is the confidence level I can >> have to say so? >> Or, to determine whether u1 and u2, or sigma1^1 and sigma2^2, are the >> same, do I have to always start with a null hypothesis that u1 = u2, >> or sigma1^1 = sigma2^2? Can't I answer the questions according to >> whether the confidence interval contains 0 (for the mean difference) >> or 1 (for the variance ratio)? If so, how much confidence do I have? >> Thanks for reading replying. >> Roland -- This address is for information only. I do not claim that these views are those of the Statistics Department or of Purdue University. Herman Rubin, Department of Statistics, Purdue University [EMAIL PROTECTED] Phone: (765)494-6054 FAX: (765)494-0558 . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
