All simulators
Randomisation Balances Covariates
Thousands of re-assignments: baseline imbalance and estimates under complete, Bernoulli and self-selected assignment.
Day 35: Randomisation & IdentificationBASELINE IMBALANCE IN EARNINGS (treated − control)
ESTIMATED EFFECT (truth = 2)
mean imbalance
-0.02
mean estimate
1.98
truth 2
sd of estimates
2.01
2,000 re-assignments of the same workers. True effect 2. Under randomisation the baseline imbalance and the estimates centre on 0 and 2; under self-selection (low earners enrol more) both shift. Each single experiment still has some chance imbalance.