In the study by Conti and Musty XXXXXXXXXXon the effects of THC on activity, the means clearly do not increase linearly with dosage. What effect, if any, should this have on any magnitudeof-effect...

In the study by Conti and Musty (1984) on the effects of THC on activity, the means clearly do not increase linearly with dosage. What effect, if any, should this have on any magnitudeof-effect measure?

With four groups you could have the means equally spaced along some continuum, or you could have three means approximately equal to each other and a fourth one different, or you could have two means approximately equal but different from two other approximately equal means, or some other pattern. Using very simple data that you create yourself (holding within-groups variance constant), how does the F statistic vary as a function of the pattern of means?