i have 4 groups with individual means and sd - pilot study results -- want to estimate sample size ., using gpower -- there for determining effect size we need single composite SD to enter ...
any reliable calculator with reference available...?
If all groups are of similar size, an estimate for the pooled SD is square root of the average of the squared SD values.
If the group sizes are unequal, the most simple solution is to calculate the SD of the residuals of an ANOVA. Otherwise you need to calculate the square root of the pooled variance: