Hi, I got the following example from MASS (10.2 Classic Nested Designs) "A cooperative trial in analytical chemistry taken from Analytical Methods Committe (1987). Seven specimens were sent to six laboratories, each three times a month apart for duplicate analysis. The response is the concentration of (unspecified) analyte in g/kg". I understand what is written in the book about the linear mixed model (nested). Laboratories and batches are treated as random and to assess components of variation. However, I have an extral question about it. From the analysis, we can evaluate the variance from different labs. Let's say a training program was done to the labs, I want to see whether after the trainning program, the variance from the lab (sigma_L in the book) becomes smaller or not. Basically I need to compare \sigma_L before and after the training program. How can I do that? I think there should already be some model for this type of analysis. Thank you in advance. Dave [[alternative HTML version deleted]]