Beyond Randomization
Author:
- 1 December 2002
- journal article
- Published by SAGE Publications in The Journal of Alternative and Complementary Medicine
- Vol. 8 (6) , 765-772
Abstract
Conventional opinion states that randomization is the appropriate way to allocate participants to treatments in biomedical studies. The most frequently cited justification is that it is the only method that guarantees balance between treatment groups with respect to all prognostic factors, whether measured or not. Here it is shown by simple arguments that this justification is false and misleading. Design-adaptive allocations are an alternative to randomization that are easy to implement, and virtually guarantee better balance than randomization, for both measured and unmeasured factors. The fraction of studies that will exhibit severe imbalance under randomization is not always trivial. For this reason, in small studies or studies with subgroup analyses or many prognostic factors, design-adaptive allocation is an attractive alternative to randomization. These considerations are particularly relevant to complementary and alternative medicine studies, where resources are relatively scarce, and otherwise underpowered studies might lead to premature termination of promising research paths.Keywords
This publication has 10 references indexed in Scilit:
- Randomization, balance, and the validity and efficiency of design-adaptive allocation methodsJournal of Statistical Planning and Inference, 2001
- Some large trial properties of minimum likelihood allocationJournal of Statistical Planning and Inference, 1983
- A program for balancing the allocation of subjects to treatment in a clinical trialComputers and Biomedical Research, 1982
- A Treatment Allocation Procedure for Sequential Clinical Trials1980
- Restricted Randomization Designs in Clinical TrialsBiometrics, 1979
- The role of prognostic stratification in preventing the bias permitted by random allocation of treatmentJournal of Chronic Diseases, 1976
- Sequential Treatment Assignment with Balancing for Prognostic Factors in the Controlled Clinical Trial1975
- Nearly Optimal Allocation of Experimental Units Using Observed Covariate ValuesTechnometrics, 1974
- Minimization: A new method of assigning patients to treatment and control groupsClinical Pharmacology & Therapeutics, 1974
- Streptomycin Treatment of Pulmonary Tuberculosis: A Medical Research Council InvestigationBMJ, 1948