MCC-UE 1040 Lecture Notes - Lecture 13: Blind Experiment, Natural Experiment, Exchangeable Random Variables
Document Summary
Experimental studies: investigator actively manipulates the exposure (intervention or treatment_ Take a group of people, give them the exposure, observe the incidence of the outcome of interest risk of exposure (actual risk) Take the same group of people, go back in time w/your time machine & do not allow them to get the exposure, observe the incidence of the outcome. However, we don"t have a time machine. Best alternative: two separate groups - one group will get the exposure & one will not. To achieve exchangeability (comparability) b/w the 2 groups, we use randomization. Rigorously designed & executed experiments among humans where investigators randomly assign participants to either an intervention or control group. Tightly controlled in terms of eligibility, delivery of intervention, and monitoring outcomes. Generally restricted to highly selected groups of individuals. Considered as the gold standard in research. Randomization: way of assigning participants to different intervention arms. Ensure that treatment occurs in an unbiased fashion.