Getting over ANOVA: Estimation graphics for multi-group comparisons

Abstract Data analysis in experimental science mainly relies on null-hypothesis significance testing, despite its well-known limitations. A powerful alternative is estimation statistics, which focuses on effect-size quantification. However, current estimation tools struggle with the complex, multi-group comparisons common in biological research. Here we introduce DABEST 2.0, an estimation framework for complex experimental designs, including shared-control, repeated-measures, two-way factorial experiments, and meta-analysis of replicates. Competing Interest Statement The authors have declared no competing interest. Footnotes * New Supplementary Notes added addressing the importance of scientific judgment; the marginal nature of DABEST's confidence intervals and trade-offs in multiple-comparisons correction; discussion of bootstrap resampling including BCa interval estimation; and a conceptual comparison of delta-delta analysis with two-way ANOVA interaction terms. * Introduction now includes recent literature on NHST limitations and multiple comparisons, and clarification that estimation graphics structurally encourage focused, pre-specified comparisons. * Repeated-measures section expanded to explain more clearly the distinct purposes of the two estimation plot panels: observed-value dispersion versus effect-size precision. * Proportions section now leads with the finding from a systematic review of 70 articles, of which none reported numerical effect sizes for Fisher's exact test results. * Mini-meta section extended to note that pooling and averaging replicates improves precision. * Author list and acknowledgements updated.