About my previous posting asking for information about intervention analysis: I just realized that my time series have not only steps in the mean, but also changes in the variance. It also has significant autocorrelation characteristics. I'm thinking of using a combined GARCH, intervention model to take into accound the heteroskedasticity, but I'm not sure that is the best way. I also have a reference to use Bayes regression using Gibbs sampling approach (a way to estimate parameters for heteroskedastic models with AR(k) characteristics). Am I in the right track? Thank you Hugo -- -- Hugo Hidalgo-Leon Water Resources Program. Civil & Environmental Engineering Department. School of Engineering and Applied Science. University of California, Los Angeles. 5731/5732 Boelter Hall. UCLA box 951593 Los Angeles, CA 90095-1593 [EMAIL PROTECTED] (310) 206 8612 Voice (310) 206 7245 Fax ================================================================= Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =================================================================