Compatibility and S values versus traditional reporting in epidemiological studies

Purpose of review This opinion review examines recent literature questioning the traditional use of P values and null hypothesis significance testing (NHST) in biomedical research. It explores frameworks that reinterpret the P value as a continuous measure of model-data compatibility and introduces the S value (surprisal) as a more informative alternative. Recent findings Growing criticism targets the dichotomization of results into ‘significant’ and ‘nonsignificant’, which oversimplifies statistical evidence and ignores biomedical complexity. Researchers advocate for viewing the P value as a graded indicator of compatibility between data and model, rather than a threshold-based decision rule. Others note that the P value's scale is nonlinear, limiting interpretability. The S value, defined as the negative base-2 logarithm of the P value, offers a more intuitive measure of refutational evidence. Such insights have been incorporated into novel proposals such as compatibility and surprisal intervals, context-specific evidentiary targets – which better reflect clinical nuance, patient variability, and loss acceptability. Summary Recent methodological developments encourage a shift from dichotomous testing to continuous, context-sensitive reasoning. Interpreting results through compatibility and surprisal promotes more transparent and nuanced inference, better suited to biomedical complexity. These approaches aim to make authors’ prior beliefs more explicit, allowing for interpretations that preserve a high degree of independence and adaptability to specific situations.