An investigation of the adequacy of a randomization.

A randomization actually performed in an individual-based double-blind intervention trial in cancer epidemiology is investigated concerning the resulting imbalance of treatment assignment. Complete randomization, permuted block procedures, and adaptive urn models are simulated in order to assess how representative the achieved distribution is for the procedure used and how other procedures would have performed on the given study population. The achieved randomization is deemed to be satisfactory and the different procedures are discussed in light of different logistic requirements. It is argued that such investigations should be carried out when debating the implications of a given randomization.