Verifiable Randomisation · IBM Quantum
For exploratory studies, psychology experiments and pilot trials. Commit to the list before enrolment, publish its hash, reveal one allocation at a time, and disclose the whole list at the end. Anyone can check the commitment.
Use one number per arm, such as 2:1
0 means simple randomisation. Use a multiple of the ratio sum
Preparing a few more than planned is fine
Leave blank for no stratification. Each stratum gets an independent list
The disclosure time is recorded. That lets a reviewer compare it with enrolment records and check that the next arm was not chosen after looking at the list.
Randomisation is easy to describe. The failure usually happens after the list exists.
If the next arm is predictable, an enrolment decision can be influenced—consciously or not. That selection bias destroys the comparison the randomisation was meant to protect.
Publishing a hash before enrolment creates a public commitment. The list can remain hidden while the commitment makes a later silent replacement detectable.
The expansion after the seed is a published deterministic calculation. That is intentional: an unpredictable seed plus a public rebuild path gives both unpredictability before creation and reproducibility afterwards.
When available, the seed comes from measurements on IBM Quantum hardware. If the hardware pool is empty, the tool says so and falls back to a cryptographically secure random generator; it does not silently relabel the source.
This is not a medical device, certified trial system or substitute for local requirements. Check institutional and regulatory requirements before using it in a regulated study. It is intended for exploratory research, psychology experiments, student projects and pilots.
The same public quantum-bit pool powers Quantum Draw and Quantum Roulette.