August 2019; Vol. 8, No. 2: Pages 19-23
What information is provided from non-significant findings and how can this be improved?
Scott J. DankelAbstract
Objectives: To clarify what information is provided from non-significant findings and explain possible additional/alternative tests to help make these findings more informative. Design & Methods: The design of this manuscript was to first clarify what information is provided from non-significant findings and detail why this may be different than what is commonly thought. Next, information is given as to why it may be particularly important for non-significant findings to be further examined within the field of exercise science given that small sample sizes are often employed. Lastly, a brief overview of two possible ways in which researchers can make non-significant findings more informative is provided. Results & Conclusions: Non-significant findings alone do not provide strong support that a given intervention did not have an effect. Researchers may wish to instead use a Bayesian statistical approach capable of quantifying evidence for both the null and alternative hypotheses. For researchers who prefer to use frequentist statistical approaches, a test for statistical equivalence may be used when there is no statistical difference present. These approaches may provide more insight into whether non-significant findings are due to uncertainty in the data or support for the null hypothesis.
Received June 29, 2019; accepted August 1, 2019