Average Treatment Effect

The average treatment effect (ATE) is the average causal effect across a whole population: the average outcome if everyone were treated minus the average outcome if no one were treated. It answers “on the whole, how much does the treatment move the needle?”

We cannot compute it directly because nobody is both treated and untreated, but a randomized experiment estimates it cleanly by comparing the average outcome of the treatment group (實驗組) and the control group (對照組). Related quantities include the ATT (the effect just among those who actually got treated), which can differ when treatment effects vary across people.