This sounds like an example of something called "Tobit" regression that is commonly used in econometrics. This can be done in SAS using the LIFEREG procedure, and is described on pages 999 and 1024 of volume 2 of the SAS/STAT manual.

There is an enormous literature on the subject of TOBIT regression in econometrics, but relatively little in biology. I can give you some references if you like.

Good luck, Garnett

Hans.vanOostenbrugge@Alg.VenV.WAU.NL wrote:

> Hello, > > I am doing some statistical analysis of variance on a dataset (catch data from > fisheries) with many zero's (50%) and with non zero values that are > log-normally distributed. From literature I read that the delta distribution > (Pennington, 1983) is a good way to describe these kind of distributions and > to calculate an good estimate of the mean and variance. I want to do a > analysis of variance using this distribution. Has anybody ever tried to > program this in SAS. Or does anybody have an idea how I can simultaneously > analyse the zero-non zero catches (catch probability as presence absence data) > and the non-zero catches (lognormally distributed) in an analysis of variance. > Can anyone of you help me with this?? Thanks very much in advance. > > Hans van Oostenbrugge > Hans.vanoostenbrugge@alg.venv.wau.nl