Knight's approval should be withheld unless he is able to provide an editorial policy stating that repeated studies are required for publication. Although certain journals may indeed require repetition for publication, the information given here is not enough to justify the use of additional animals in these circumstances. Because the biostatistician is satisfied with the proposed research, it would be unnecessary for the committee to approve repetition. When a committee considers a research proposal, it should measure the work against the 3Rs to see if replacement, reduction or refinement has been implemented. A well-planned and well-executed study with an appropriate number of animals as determined by a power analysis should give journal editors and reviewers, as well as Knight, confidence in the results.

The number of animals necessary to attain a high level of confidence (power) can be calculated if the standard deviation, expected effect and error are known. Knight seems to have provided satisfactory proof of such to the IACUC. Using more animals increases power up to a point, but surpassing the required number of animals is a waste of resources and time.

Knight's argument that he must repeat his studies twice to show that the results are reproducible is a mistake that might be made by many researchers who are not confident about statistics. A lack of understanding or a distrust of statistical methodology may negatively influence research design. Although Knight may object to the biostatistician's confidence in the size of his study groups and questioning of his need for repetition, he should understand that the biostatistician is better qualified to advise him on the appropriate design of the study. Proof that his work is repeatable should be undertaken independently. The purpose of publication is to promulgate the methods and results to the scientific community, allowing other researchers to independently replicate the results, which adds to the validity of the findings.

Return to Protocol Review