Systematically estimating ‘replicability’ by using Fear Conditioning Research as a case example (FEAR-REP)

行为科学、社会科学和认知科学目前在实证研究结果的可重复性方面面临着激烈的争论。在过去的几年里，方法稳健性的重要性以及测量、处理和分析异质性的丰富性在恐惧条件反射研究领域得到了强调和讨论。然而，缺乏对“可复制性”的正式和全面的调查。fear - rep项目的总体目标是通过使用恐惧条件反射研究领域作为案例，从元科学的角度系统地调查“可复制性”。FEAR-REP项目将通过对代表不同样本和程序规范的大量研究的重新分析，系统地评估对测量挑战（可靠性、聚合效度、假阳性风险、WP 1）的直接复制和推广性的见解。此外，通过将“消极情绪”与恐惧学习之间的关联作为案例，将系统地研究不同样本量下效应大小的变化（WP2）。这将为我们提供可靠的交叉研究估计，从而为未来的样本量和功率计算提供信息。最后,根据数据处理和分析异质性（在没有正式理论的情况下），将通过多元宇宙/规范曲线分析来研究其通用性和稳健性，该分析有望推进“负面情绪”和“恐惧学习”之间关联的累积科学，并提供对这种关联的稳健性的见解，以对抗程序和分析异质性，这将有助于评估方法上的“密切性”泛化的尝试。

The behavioral, social and cognitive sciences currently face an intense debate on the replicability of empirical findings. During the past years, the importance of methodological robustness as well as the abundance of measurement, processing and analytical heterogeneity has been highlighted and discussed - also in the field of fear conditioning research. Yet, a formal and comprehensive investigation on ‘replicability’ is lacking. The overarching aim of the FEAR-REP project is to systematically investigate ‘replicability’ from a meta-scientific perspective by using the field of fear conditioning research as a case example. The FEAR-REP project will systematically gauge insights into measurement challenges (reliability, convergent validity, risk for false positives, WP 1) for direct replication and generalizability through re-analyses across a large number of studies representing a diverse set of sample and procedural specifications. Furthermore, by employing the association between ‘negative emotionality’ and fear learning as a case example, the variation in effect sizes at different sample sizes will be systematically investigated (WP2). This will inform future sample size and power calculations by providing us with robust cross-study estimates. Finally, generalizability and robustness in light of data processing and analytical heterogeneity (in absence of formal theories) will be investigated though multiverse/specification curve analyses which is expected to advance cumulative science on the association between ‘negative emotionality’ and ‘fear learning’ and provide insights into the robustness of this association against procedural and analytical heterogeneity that will help to evaluate methodological “closeness” of replication and generalization attempts.