The Minimization algorithm selects the next treatment arm for a randomized participant by looking at the existing distributions of pre-existing randomizations. The algorithm attempts to minimize the difference between these distributions by assigning the participant to the treatment arm with the lowest number of participants who fit the stratification factors. In this way, the algorithm can ensure an even distribution of participants by stratification factor to all treatment arms.
The Minimization algorithm also introduces a chance for a random treatment arm selection which is set to 30% by default.
The Chance of Random Treatment Arm Selection can be altered when setting up a new randomization scheme, and changed to any integer between 0 (Never Random) and 100 (Always Random).
Under the Minimization algorithm, only one list of randomization treatment arm assignments exists, unless site is used as a stratification factor. This is to ensure that within a site, the distribution of participants across treatment arms is balanced.
When a subject is randomized, the next treatment assignment is dynamically generated based on the distribution of stratification factors of previous participants on the list. Since it is not possible to know the distributions in advance, it is not possible to generate a pre-determined list of treatment assignments at the start of a study when using the Minimization algorithm.