Randomised Experiment

Random assignment of xx, which makes xx independent of the error uu so that differences in outcomes can be read as causal.

When to use

Lecture 1 uses it as the way to estimate causality, as in the health and drug example Health=β0+β1Drug+uHealth=\beta_0+\beta_1 Drug+u; without randomisation, confounders can bias the estimate.

See Lec 1.

Appears in

No references yet.