Abstract

Randomized control trials (RCTs) are regarded as the gold standard for intervention research. The randomization process is intended to establish comparability between groups, so that the study outcomes can be attributable to the intervention, rather than group differences. The purpose of this paper is to emphasize the inherent risks of conducting multiple tests in the establishment of equivalency at baseline while omitting the cumulative effect of small group differences in RCTs. Randomization does not thoroughly prevent differences in group averages at the specific characteristic level. Any baseline differences that benefit the intervention group when accumulated over multiple categories of demographic characteristics described herein as cumulative inequivalence can significantly impact the internal validity of RCTs. This paper describes a procedure for assessing for cumulative inequivalence, as well as procedures such as re-randomization prior to intervention to establish comparability and thus promote cumulative equivalence of RCTs.

This work is licensed under a Creative Commons Attribution 4.0 License.