Sum coding doesn't seem to be working: the intercept is the mean of the 'reference' level rather than the grand mean of all levels

Question

I am currently using the robu() function in R to run a meta-regression applying robust variance estimation. This seems to be working ok. However, I am interes