Order Effects & Counterbalancing Lab

A within-subjects psychology experiment asks every participant to complete both a control and a treatment condition — which means the order they're tested in can leak into the results. This simulator runs a virtual repeated-measures study: each participant gets two bars, one per test position, colored by which condition landed there, with height driven by a transparent generative model (baseline ability + true treatment effect + order/practice effect + noise). Switch between a fixed testing order and a counterbalanced design and watch the naive group-level estimate drift away from — or snap back to — the true, ground-truth treatment effect, with live readouts for the estimate, its bias, and the raw position-to-position gap.