I have this study which reported mean's and SD's of food intake (grams) for 3 independent groups (age 3-6 yrs, 7-10 yrs, 11-15 yrs) but not total food intake (mean and SD for all kids); I know how to combine mean's (m1*n1+m2*n2+m3*n3/n1+n2+n3) but I also need to have the combined SD to use this study's estimate for meta-analysis. Any idea how I can calculate a good estimate of SD for the whole sample? other than contacting the author! :)