No Estimation without Inference

The International Society of Physiotherapy Journal Editors (ISPJE) recently published an editorial warning that many their journals would soon prohibit the use null hypothesis tests and instead require authors to interpret 95% confidence intervals relative clinically important values. Although I encourage reporting discussing uncertainty in context a research question, ISPJE’s proposed ban is illogical there are several instances flawed statistical reasoning editorial. In brief, editorial: (1) fails adequately grapple with inherent connection between testing estimation, (2) presents misleading arguments about perceived flaws tests, (3) alternative is, itself, form test – minimal effects albeit done informally. If editorials’ taken at face value, then will lower literacy our field readers have understanding p-values. Further, if put into practice, fear could decrease scientific integrity as it removes quantitative benchmarks favor more subjective interpretation intervals. Ultimately, think concerns led valid, but those problems result questionable practices stemming from poor methodological training for authors, reviewers, editors. These only be fixed through better continuing education, not banning statistically valid methods.