I want to calculate the Etta square effect sizes of the different variables in a three-way ANOVA design. I'm using Matlab's anovan function that outputs the SS of the different variables, SSerror and SStotal. I noticed that the SS of the variables and error do not sum to SStot. From what I read, I suspect that this might be because the output is the type iii sum of squares. If so, is this what I should be using to calculate the effect size or should I be using type i or ii?
Thanks in advance!