by Ravi » Mon Apr 01, 2019 11:19 am
So I think this has to do with how JASP calculates the total sum of squares. We used to calculate the total sum of squares in the same way as JASP (i.e., add the SS of all terms belonging to an individual error term together + the SS of the residual), but somebody sent us an email telling us that this was wrong in case of a RM/Mixed model so we corrected it. To calculate the total sum of squares, we now add all the sums of squares together and to get the eta squared you divide the SS of the term by the total SS.