Binned scatterplots: A simple tool to make research easier and better

• 成果类型: Article
• 署名作者: Starr, Evan; Goldfarb, Brent
• 署名单位: University System of Maryland; University of Maryland College Park
• 刊物名称: STRATEGIC MANAGEMENT JOURNAL
• ISSN/ISSBN: 0143-2095
• DOI: 10.1002/smj.3199
• 发表日期: 2020
• 页码: 2261-2274
• 关键词: binned scatterplots; data visualization; graphical analysis; least squares regression; Nonparametric analysis

摘要

Research Summary We seek to diffuse a graphical tool-binned scatterplots-which we argue can dramatically improve the quality and speed of research in strategic management. In contrast to the current practice of showing plots of predicted values, binned scatterplots graph the nonparametric relationship between two variables, either unconditionally or conditional on a set of controls, for multiple subgroups. This allows researchers to quickly detect the shape of that relationship, examine outliers, and assess which part of the support may be driving a relationship. We propose that the adoption of binned scatterplots will lead to the identification of new and interesting phenomena, raise the credibility of empirical research, and help create richer theories. Managerial Summary Regression analysis often assumes linear or quadratic relationship forms between two variables. We seek to diffuse a graphical tool-binned scatterplots-that allow the researcher and reader to evaluate whether such assumptions are maintained throughout the data. For example, binned scatterplots may clarify that a regression relationship is nonlinear, or driven by an exceptional firm or a small set of firms. We propose that using binned scatterplots will improve the transparency and quality of empirical work, as well as aid in the development of new phenomena and enhance theory development.