Abstract:
|
Ethical guidelines for biomedical research require that subject numbers be determined by what is 'justifiable, reasonable, and ethical', with statistical justification as appropriate. However, substantive clarification is lacking. For most biomedical researchers who are not statisticians, 'statistical' justification implies power calculations. However, 'statistics' also refers to the collection, analysis, interpretation, and communication of data in a much broader sense. I surveyed directions for calculating preclinical subject numbers for 25 academic institutions. Less than half provided information on relevant basic statistics, statistical resources, or even simple tracking metrics; 25% described power calculations, but most of these had at least one major error. Power calculations presented in 50 protocol submissions at this institution showed profound misunderstanding of basic concepts and major calculation errors. Quantitative claims need quantitative justification, but this must be accurate and informed. Investigators must be educated in how to use simple math and quantifiable metrics more effectively, and more thoroughly instructed in the basics of hypothesis testing.
|