With Respect I want to compare the effect of multiple independent variables (at maximum one variable with two values and two with three) on a ratio dependent variable. This a situation for factorial analysis of variance, but: 1/I will have very unequal numbers of observations in the cells 2/ The ratio dependent variable is distributed nonnormally as a whole, and probably will be within categories Any suggestions? Has anyone written a book on statistics for nonnormal ratio data? Yours Sincerely, Alan E. Dunne