-------- Original Message -------- Subject: weighting of averages Date: Fri, 8 Oct 2010 02:39:21 -0400 From: Ondřej Mikula <[email protected]> To: [email protected] Dear list, I want to fit non-linear spatial trend through the set of localities, each represented by the mean PC1 score. The problem is the sample sizes differ widely among localities. Thus weighting of observations defined this way seems appropriate. I wonder, however, how the weights should be defined. My idea is to use inverse of standard errors of locality means (or, assuming the constant within locality variance, square root of sample sizes). What do you think about? Best regards Ondra Mikula (Institute of Animal Physiology and Genetics, Brno, Czech Republic)
