Dear colleagues,

Earlier I noticed that the traditional t-test statistic seems to give too low estimates. Now I noticed the same with F-test statistic: The estimates by the traditional F-tests are deflated because the embedded eta squared is radically deflated. That is, the explaining power we usually use with GLM-settings, tends to have reduced scale, radically less than 0 to +1. Its scale may be reduced to 0 to +0,40 or even less. This has an effect to the F-statistic as well as Cohen's f effect size.

I would appreciate your time. It would be highly appreciated if you could spend some minutes to read my preprint at Deleted research item The research item mentioned here has been deleted

JMe