What is the key difference in the designs of an experiment and an observational study? What does that difference have to do with the fact that we can determine causality between explanatory and response variables with experiments but not with observational studies?