POL101Y1 Lecture Notes - Lecture 19: Causal Inference
Document Summary
Experiment: milk or tea (which should be put into the teacup rst?) Experiments require treatment, control, measurement, and randomization. Treatment: a stimulus, something meant to have an effect on a subject. Treatment: tie (anything that"s thought to affect an outcome) Random assignment of treatment: some students are taught with tie, some without (coin ip to decide) Random = nothing about a subject makes them more likely to be placed in treatment or control. Ex-post measurement of results: need to be able to measure in some quantity, the results of the experiment. To make a causal inference, we must have all three parts (controversial) Causal effect: difference in outcome between two states of the world, treated versus untreated. Fundamental problem of causal inference: we cannot observe any given subject in both its treated and untreated state. Random assignment allows groups whose treated and untreated states are the same in expectation.
