A clinic tests a new ointment for a painful rash. Because this is a simulation, you set how well the ointment really works — so you always know the right answer. Now build the study out of three parts — a control group, randomization, and blinding — and watch each one delete a specific slab of bias from what the study reports.
Where the improvement comes from
would have healed anywayplacebo responseexpectation & scoringgroups not comparablethe real drug effect