Fix description and K==4 case #42
Conversation
Fixed two things here: 1. in the function description, line 1 it said that the function would calculate mu's from effect size expressed in Cohen's f, however, the function actually uses eta-squared for this. 2. in the K==4 case, the vector should be c(-2*a, -a, a, 2*a), instead of c(-a, -a, a, a). The latter would assume equal levels to have equal means.
|
Thank you for the pull request. @Lakens, can you verify this? I believe you wrote the original mu_from_ES function? |
|
Hi, this is a helper function that is not used in our paper or app, but that was used in Albers and Lakens, 2018, to set means for a specific simulation. I don't understand the sentence "The latter would assume equal levels to have equal means." (I think it means the -a -a need to be identical, as it was in our simulation, but the proposaed function would make it more generic?). I don't think I made this code either (Casper Albers did) and as far as I know it generates the correct values for the simulation we wanted to run. Maybe @xavfunk can add a bit more explanation of the problem, which error it leads to (or possible which values, which we never used it for, it fails to work for) and how the proposed change in the code fixes the problem that was identified? That would help me to evaluate this. I guess we used this function only very narrowly in situations where it worked, and therefore did not encounter the situation it did not work in. If we can make it work for more feneral calculations, that is useful, perhaps - although the formula really is not meant for general use, as it is limited to k = 4 anyway, just for a very specific simulation we once ran. I also don't remember what paper this was based on, where we got the formula from, or if we created this ourselves. |
Fixed two things here:
in the function description, line 1 it said that the function would calculate mu's from effect size expressed in Cohen's f, however, the function actually uses eta-squared for this.
in the K==4 case, the vector should be c(-2a, -a, a, 2a), instead of c(-a, -a, a, a). The latter would assume equal levels to have equal means.