All Times EDT
Keywords: Estimation, Null-Hypothesis Significance Testing, P Values, Data Visualization
The two-groups design is the most frequently used experimental paradigm in biomedical research. It is most commonly analysed with null-hypothesis significance testing (NHST). The massive shortcomings of NHST, however, have been pointed out by several critics for decades. Many now agree that NHST fosters a dichotomizing perspective that hampers quantitative reasoning. Estimation statistics, with its emphasis on effect sizes and confidence intervals, is a robust and intuitive alternative. We created DABEST, a suite of software packages and webapp, enabling nonexperts to generate estimation graphics easily. These tools display all data points, and employ bootstrap techniques to produce robust non-parametric confidence intervals. The evolution of statistics has been linked with development and adoption of quantitative displays that support complex visual reasoning. The rapid adoption of DABEST indicates its effectiveness to remedy the statistical and visualization shortcomings in biomedical research.