Commentary: Improving our statistical inferences requires meta-research

In the past few years, there has been renewed interest in the use of thresholds for declaring statistical significance and, if we decide to use thresholds, what alpha value is optimal. Should we lower the threshold for claiming statistical significance fromP< 0.05 toP< 0.005?1 Should we justify the selected alpha level for every study separately?2 Or should we abandon statistical significance altogether?3 Whereas these proposals may appear at odds, they align in terms of their goal —to improve the robustness of our inferences. However, empirical data that sheds light on the potential costs and benefits of these approaches remains scarce.
Source: International Journal of Epidemiology - Category: Epidemiology Source Type: research