Hi, all I happen to a question about multivariate meta-analysis. I found the function of mvmeta in gap package, but I don't know for sure how to construct a covariance matrix. Now, I have 20 studies, each having 6 coefficients for the basis functions from an additive model. I can extract the covariance matrix for each coefficient. Could somebody give me a hint how to construct a covariance matrix for mvmeta function? Thanks. Lee [[alternative HTML version deleted]]