模糊集定性比较分析(模糊集定性比较分析法百度百科)

知识是为了预见，预见是为了权力。--孔德

说明

作者简介：

Evan J. Douglas. Professor of Entrepreneurship in the Department of Business Strategy & Innovation, Griffith Business School

埃文·j·道格拉斯。格里菲斯商学院(Griffith Business School)商业战略与创新系的创业学教授

Dean A. Shepherd is the David H. Jacobs Chair in Strategic Entrepreneurship, Kelley School of Business, Indiana University, USA. He is the former Editor-in-Chief of the Journal of Business Venturing

迪恩·谢博德是美国印第安纳大学凯利商学院战略创业学的大卫·H·雅各布斯讲席教授。他是Journal of Business Venturing的前主编。

Catherine Prentice, Griffith University · Department of Marketing. Associate professor or marketing. Used to be the chief editor of JBV.

凯瑟琳·普伦蒂斯，格里菲斯大学·营销系副教授。

关键词：模糊集，创业，非对称数据，异常值，组态

绝大多数创业理论都是被概念化可以通过多元回归分析和结构方程模型等对称定量方法进行检验。这些传统的对称方法是通过测试自变量和因变量之间的关系来解释创业现象的。对称方法要求数据符合限制性假设，包括正态分布数据、对称数据关系和变量的独立性，这些要求限制了这些方法处理我们观察到的一些复杂创业行为的能力。事实上，传统方法无法反映复杂的创业异质性这一重要的方面。

创业中出现的五个主要问题对传统的对称方法来说是棘手的，使得这些问题的研究较少受到关注。

首先，大多数关键创业变量呈现出高偏度的分布，而不是围绕均值的正态分布，这就需要对数据进行操作并消除异常观测值，然而，偏度和异常值对于理解创业背景可能很重要。

第二，创业行为的异质性反映了创业者及创业企业之间的差异，而传统的方法旨在捕捉所有案例的共性，并必然抑制可能观察到的异质性的案例间差异。

第三，创业现象通常以关系不对称为特征，这意味着在不同的情况下，一个先行变量可能与结果正相关，也可能与结果负相关。对称方法旨在发现单一的“净效应模型”，该模型强调发现的主导关系，忽略数据中的任何少数关系。

第四，创业结果，无论是在创业者、公司还是制度的层面上，都需要考虑到自变量之间的相互依赖性之后进行分析，然而，传统的方法将创业现象解释为分离地，考虑为前因变量的线性可加性影响，即独立于其他前因变量的影响。

最后，我们观察到创业者和组织可以通过各种途径获得创业成果，但传统方法只提供了单一的主导净效应解释。因此，对称方法不能揭示我们在实践中观察到的创业者异质性的重要方面（因为这不是这些方法的目的）。

模糊集定性比较分析（fsQCA）提供了一种深入挖掘数据以揭示关于创业现象复杂性的精细细节的方法。fsQCA方法兼容于数据不对称性，变量的潜在相互依赖性，识别非对称的数据关系，并揭示了同一结果的多个等效性路径。FsQCA检验案例内前因变量（称为条件）之间的关系，并分析因变量（称为结果）与特定条件组合（称为组态）间关系。它发现多个案例的共同组态，这些不同的共同组态构成特定结果的特定途径。因此，FsQCA是对传统对称方法的补充，它增加了关于创业现象的更细致的理解，并为溯因分析提供了经验基础，即它可以揭示令人惊讶的经验发现，从而激发新理论构建的尝试。

本文概述了fsQCA方法，论证了该方法在给定数据集上可以提供的新信息，为未来的创业研究提供了一个全面的方案，即可以利用fsQCA补充传统方法，从而为未来的理论建设提供新的信息。不同于“一刀切”的政策，本文从多方面为创业公共政策提供了可能性。同样，这也表明教育者可以超越典型创业者的概念，鼓励那些不符合经典模式的人创业，而那些未来的创业者和投资者应该认识到，展现出各种不同组态的个人也可以创业，并有可能取得创业成功。

英文原文：

‍‍The great majority of entrepreneurship theory has been conceptualized to be tested using symmetric quantitative methods, such as multiple regression analysis and structural equation modeling. These traditional symmetric methods test relationships betweenproposed independent and dependent variables to explain entrepreneurial phenomena. Symmetric methods require the data to conform to restrictive assumptions, includingnormally distributed data, symmetric data relationships, and independence of the variables, and these restrictions limit the ability of these methods to deal with some of the complexity of entrepreneurial behavior that we observe. In effect, the complexity of entrepreneurial phenomena exceeds the capability of traditional methods to reflect important aspects of its heterogeneity.

‍

Five main issues occur in entrepreneurship that are problematic for traditional symmetric methods, such that these issues have been afforded less research attention. First, the majority of key entrepreneurship variables exhibit highly skewed distributions, rather than being normally distributed around their means, necessitating data manipulation and the elimination of outlierobservations, yet the skew and the outliers may be important in understanding the entrepreneurship context. Second, the heterogeneity of entrepreneurial behavior is reflective of inter-case differences amongst entrepreneurs and their ventures, whereas traditional methods aredesigned to capture the commonalities across all the cases, and necessarily suppress inter-case differences that may be causal for the observed heterogeneity. Third, entrepreneurial phenomena are often characterized by relationship asymmetry, meaning that an antecedent variable may be both positively and negatively associated with the outcome, for different cases. Symmetric methods find the single “net-effects model” which highlights the dominant relationship found and ignores any minority relationships that lie within the data. Fourth, entrepreneurial outcomes, whether at the inpidual, firm or institutional level of analysis, tend to be pursued after taking into account the interdependencies between and among the antecedent variables, yet traditional methods explain entrepreneurship phenomenon as the linear additive impact of the antecedent variables considered discretely – i.e. independently of the effect of other antecedent variables. Finally, we observe entrepreneurs and organizations taking a variety of pathways to entrepreneurial outcomes, yet traditional methods offer a single dominant net-effects explanation. Thus, symmetric methods cannot reveal important aspects of entrepreneurial heterogeneity (because it is not the method’s purpose) that we observe in practice.

Fuzzy-set qualitative comparative analysis (fsQCA) provides a method to dig deeper into the data to reveal finer-grained detail about the complexity of entrepreneurial phenomenon. The fsQCA method accommodates data asymmetry, recognizes the potential interdependence of antecedent variables, identifies asymmetric data relationships, and reveals multiple equally-effective pathways to the same outcome, if they exist. FsQCA examines the within-case relationships among the antecedent variables (referred to as conditions), and characterizes cases as having a particular combination of conditions (known as a configuration) that associates with the dependent variable (known as the outcome). It discovers the configuration common to multiple cases who take a particular pathway to a given outcome, as distinct from those who take otherpathways to the same outcome. FsQCA is thus complementary to traditional symmetric methods, adding finer-grained detail about entrepreneurial phenomena and providing an empirical basis for abduction, i.e., it can reveal surprising empirical findings to provoke new theory building efforts.

‍‍In this paper we outline the fsQCA method, demonstrate the additional information this method can provide from a given data set, and provide a comprehensive agenda for future entrepreneurship research where fsQCA can be used to complement traditional methods and thereby provide new information for future theory building. This paper provides motivation for entrepreneurial public policy on multiple fronts, rather than for “one-size-fits-all” policies. Itsimilarly suggests that educators can go beyond the notion of the archetypical entrepreneur to encourage entrepreneurship by those who do not fit the classical mold, and that would-be entrepreneurs and investors should recognize that inpiduals exhibiting a variety of different configurations can act entrepreneurially and potentially achieve entrepreneurial success.

‍

总结

本文概述了fsQCA方法，论证了该方法在给定数据集上可以提供附加信息，为未来的创业研究提供了一个全面的方案，即可以利用fsQCA补充传统方法，从而为未来的理论建设提供新的信息。不同于“一刀切”的政策，本文从多方面为创业公共政策提供了可能性。同样，这也表明教育者可以超越典型企业家的概念，鼓励那些不符合经典模式的人创业，而那些未来的企业家和投资者应该认识到，展现出各种不同配置的个人也可以创业，并有可能取得创业成功。