Exploratory data analysis as a foundation of inductive research
Exploratory data analysis as a foundation of inductive research
复制标题
DOI：
10.1016/j.hrmr.2016.08.003
复制
发表时间：
2017-06-01
影响因子：
11.4
通讯作者：
Woo, Sang Eun
中科院分区：
文献类型：
--
作者：
Jebb, Andrew T.;Parrigon, Scott;Woo, Sang Eun
在各个学科中，当演绎和归纳方法之间取得平衡时，科学进步最大化。为了促进组织科学中的这种平衡，必须进一步鼓励旨在发现现象的严格归纳研究。为此，本文讨论了探索性数据分析（EDA）的逻辑和方法，即涉及发现、探索和经验性检测数据中现象的分析模式。我们开始首先描述EDA的历史和概念背景。然后，我们讨论了与EDA及其与科学可信度的关系有关的两个问题。首先，我们认为，EDA通过要求交叉验证和强调数据模式的自然不确定性来促进基于复制的科学。其次，我们明确EDA与其他被认为在科学上有问题的探索性实践（例如，"p-hacking"、"data fishing"和"data-dredging"）。在本文的下一节中，我们为EDA提出了最后一个论点：它有助于最大限度地提高数据的价值。为了说明这一点，我们提出了几个图形化的方法来检测数据模式，并为感兴趣的读者提供进一步的技术参考。(C)2016 Elsevier Inc. All rights reserved.
Across academic disciplines, scientific progress is maximized when there is a balance between deductive and inductive approaches. To promote this balance in organizational science, rigorous inductive research aimed at phenomenon detection must be further encouraged. To this end, the present article discusses the logic and methods of exploratory data analysis (EDA), the mode of analysis concerned with discovery, exploration, and empirically detecting phenomena in data. We begin by first describing the historical and conceptual background of EDA. We then discuss two issues related to EDA and its relationship to scientific credibility. First, we argue that EDA fosters a replication-based science by requiring cross-validation and by emphasizing the natural uncertainty of data patterns. Second, we clarity that EDA is distinguishable from other exploratory practices that are considered scientifically questionable (e.g., "p-hacking", "data fishing" and "data-dredging"). In the following section of the paper, we present a final argument for EDA: that it helps maximize the value of data. To illustrate this point, we present several graphical methods for detecting data patterns and provide references to further techniques for the interested reader. (C) 2016 Elsevier Inc. All rights reserved.