Q3.18
Experimental design and controls
Cards in this group · 4
- Q3.18.1Without a control group, treatment cannot be separated from the time trend
- Q3.18.2Randomize at the unit where the effect occurs, or the observations are only pseudo-independent
- Q3.18.3Within-subjects designs need order effects, and usually counterbalancing
- Q3.18.4A group difference is a treatment effect only after other variables are controlled