What should you do if randomization does not produce equivalent groups

If randomization is used (randomly allocating participants or sites to one or more treatment groups and a control group), then the characteristics of the groups should be compared to check if randomization has in fact produced similar groups.

If there is an imbalance between your two samples, you have several options: to re-run randomization (not recommended), or to include those “unbalanced” variables as controls in the final analysis.


This FAQ refers to the approach Randomized Controlled Trials (RCTs).

Page Credits

Author: Reviewer: Steward: Other Contributors:
Angela Ambroz, Marc Shotland - - -