. Good overview — but one thing one ought to pay more attention to is the following: Even if both sampling and assignment are made in an ideal random way, performing standard randomised experiments only gives you averages. The problem here is that although we may get an estimate of the ‘true’ average causal effect, […]