Learning from Outliers and Anomalies
区分了离群值（极端数据点）和异常（极端模式）两种认知立场，分别对应频率论和贝叶斯框架，帮助学者和政策制定者从极端组织绩效中提取有效洞见。
While management scholarship has frequently taken an interest in extreme organizational performance, methodologies that draw appropriate inferences from extreme outcomes are a recent development. How can academics and policy-makers learn from exceptional entrepreneurs, leaders, and firms? In this article, we propose that answers to this question hinge on the epistemic stance we take toward extremes—in particular, whether we view them as outliers (extreme data points) or anomalies (extreme patterns). A focus on outliers encourages in-depth examination of empirical cases, with an eye toward distinctive mechanisms that drive their performance. Quantitative assessment proceeds within a frequentist framework that is sensitive to the effect of adding and removing outliers from a larger sample or population. A focus on anomalies emphasizes in-depth examination of parameters or processes, with an eye toward spatial, temporal, industry, and firm-level contexts that may shift performance outcomes. Quantitative assessment proceeds within a Bayesian framework that updates prior knowledge with novel contexts that generate anomalous patterns. Drawing largely on examples from the entrepreneurship literature, as well as simulated data, we highlight conditions under which a focus on outliers or anomalies may be most informative for organizational research and policy.