On 20 Nov 2002 08:47:09 -0800, [EMAIL PROTECTED] (Paul R Swank) wrote: > As the websites posted below gave me some problems One couldn't be found and > the other froze in the middle), would someone please explain what has
The first URL leads to the second one. Both URLs are rather long; my newsreader wrapped the second URL to an extra line so I couldn't just click on it. > happeded to the non-central t distribution since I last checked. If the > assumptions for t are met, that is, the observations are NID with equal > variances, why would the non-central t be asymmetric. For a long time, I was thinking exactly that same thing -- If you write it as non-central normal divided by chisquared, the numerator is normal, which is symmetric even if non-central. But look at this artificial, numerator-symmetric example: - A variable y has values of 1 with 50% probability, otherwise 0 or 2 with equal probabilities of 25%. - It is divided by chisquared (as with: non-central t). The result for 1 is the reciprocal of chisquared. The result for 2 is a similar shape, but spread more. However, the result for 0 is all folded into ZERO. So, that is where skewness comes from. The site from DW apparently provides descriptions of a dozens of distributions, including (sometimes) a plot of the CDF. For convenience, I am repeating the addresses here. (The second <longer> one is written on one line, as it looks to me before I post it. But your reading might 'wrap' it, even if my posting it doesn't change it from this.) http://www.itl.nist.gov/div898/software/dataplot.html/distribu.htm and http://www.itl.nist.gov/div898/software/dataplot.html/refman2/auxillar/nctpdf.pdf -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
