Oct. 10, 2002
1:18 a.m.
If you do decide to try the chi-square thing, the idea is to find or create a function (perhaps using a lookup table) that takes a chi-square random variable and outputs the associated p-value. The input random variable is the product of the p's or (1-p)'s as the case may be. If the p's are uniformly distributed under the null hypothesis, the product is chi-square with 2n degrees of freedom, where n is the number of terms making up the product. So the inverse chi-square function gives the probability associated with that product. --Gary -- Gary Robinson CEO Transpose, LLC grobinson@transpose.com 207-942-3463 http://www.emergentmusic.com http://radio.weblogs.com/0101454