Navigating between contradictory results is not rare in the practice of evidence-based medicine. Recently, two papers published in the same year and in the same journal investigated the same research question with the same dataset and reached divergent results regarding the benefits of retrieval bag use during laparoscopic appendectomy.1 The two studies reached contrasting conclusions, one found that these bags actually reduce the risk of infection2 while the other study found no support for a difference.3 Likewise, a multitude of network meta-analyses about the treatment of psoriasis reached divergent conclusions on the best drug to use,4 the best drug always being the one of the drug manufacturer in case of industry-funded meta-analyses. Implementing the findings of medical research for decision-making in clinical practice is quite challenging when scientific results stand on such unstable ground. One reason, among others, is analytical flexibility that represents the variability in results arising from ‘researcher degrees of freedom’ (ie, uncertain decisions researchers have to make in study design, data collection and data analysis5). Analytical flexibility arises, for instance, when researchers have to choose among multiple justifiable methods, models or measurements. Given this analytical variability and under the pressure to publish, researchers may try different analysis strategies and selectively report the most impressive, desirable, publishable result.6 Not surprisingly, reported results may be, on average, inflated.7 The range of results arising from analytical flexibility can be explored using a generalisation of sensitivity analyses in which all uncertain analytical and methodological choices are systematically varied to estimate how much different results can be, that is, the vibration of effects (VoE). This framework reports the range of effect sizes that can be obtained within the same study due to the various analytical and methodological choices that can be made,7 providing researchers …

中文翻译：

---

什么是效果器的振动？


在循证医学实践中，在相互矛盾的结果之间进行选择并不罕见。最近，在同一年同一期刊上发表的两篇论文使用相同的数据集调查了同一研究问题，并就腹腔镜阑尾切除术中使用取回袋的益处得出了不同的结果。 1 这两项研究得出了相反的结论，一项研究发现这些研究袋子实际上降低了感染风险2，而另一项研究没有发现差异的证据。3 同样，大量关于治疗牛皮癣的网络荟萃分析对于最佳药物使用得出了不同的结论，4 最好的药物总是行业资助的荟萃分析中的药品制造商之一。当科学结果站在如此不稳定的基础上时，将医学研究结果应用于临床实践决策是相当具有挑战性的。其中一个原因是分析灵活性，它代表了“研究人员自由度”（即研究人员在研究设计、数据收集和数据分析中必须做出的不确定决策）所产生的结果的可变性。例如，当研究人员必须在多种合理的方法、模型或测量中进行选择时，就会出现分析灵活性。鉴于分析的可变性以及在发表的压力下，研究人员可能会尝试不同的分析策略，并有选择地报告最令人印象深刻、最理想、可发表的结果。6 毫不奇怪，报告的结果平均而言可能会被夸大。7 分析灵活性产生的结果范围可以使用敏感性分析的概括来探索，其中所有不确定的分析和方法选择都系统地变化，以估计结果的差异程度，即效应振动 (VoE)。该框架报告了由于可以做出的各种分析和方法选择而在同一研究中可以获得的效应大小范围，7为研究人员提供了……