社会杂志

• 论文 •    下一篇

走向开源的社会学——定量分析中的复制性研究

*作者1:陈云松牛津大学社会学系，纳菲尔德学院| 作者2:吴晓刚香港科技大学社会科学部。     

  1. *作者1:陈云松牛津大学社会学系，纳菲尔德学院| 作者2:吴晓刚香港科技大学社会科学部
  • 出版日期:2012-05-20 发布日期:2012-05-20
  • 通讯作者: *作者1:陈云松牛津大学社会学系，纳菲尔德学院。 E-mail: yunsong2000@gmail.com
  • 作者简介:*作者1:陈云松牛津大学社会学系，纳菲尔德学院(Author 1: Chen Yunsong, Department of Sociology & Nuffield College, The University of Oxford) Email: yunsong2000@gmail.com; 作者2:吴晓刚香港科技大学社会科学部（Author 2: Wu Xiaogang, Division of Social Science, Hong Kong University of Science and Technology）Email: sowu@ust.hk摘要：社会科学定量分析亟待建立一个透明和开源的学术机制，让研究数据和模型公开共享，使研究成果可以得到他人的验证和进一步拓展。这个学术机制的核心就是倡导“复制性研究”。本文结合近年来西方社会科学界围绕复制性研究的实践和讨论，提出了复制性研究对于社会科学特别是中国社会学定量分析的重要意义、操作可能性以及可能存在的争议。在此基础上，本文通过两个实证定量分析的例子，为如何进行复制性研究提供了简明的范本。最后，结合中国实际提出了就复制性研究所应达成的共识和建议。本文所论及的复制性研究，对社会科学定量分析其他领域也同样具有参考价值。

Towards an OpenSource Social Science: The Replication Study in Quantitative Social Analysis

*Author 1: Chen Yunsong, Department of Sociology & Nuffield College, The University of Oxford|Author 2: Wu Xiaogang, Division of Social Science, Hong Kong University of Science and Technology   

  1. *Author 1: Chen Yunsong, Department of Sociology & Nuffield College, The University of Oxford| Author 2: Wu Xiaogang, Division of Social Science, Hong Kong University of Science and Technology
  • Online:2012-05-20 Published:2012-05-20
  • Contact: *Author 1: Chen Yunsong, Department of Sociology & Nuffield College, The University of Oxford E-mail: yunsong2000@gmail.com
  • About author:*作者1:陈云松牛津大学社会学系，纳菲尔德学院(Author 1: Chen Yunsong, Department of Sociology & Nuffield College, The University of Oxford) Email: yunsong2000@gmail.com; 作者2:吴晓刚香港科技大学社会科学部（Author 2: Wu Xiaogang, Division of Social Science, Hong Kong University of Science and Technology）Email: sowu@ust.hk摘要：社会科学定量分析亟待建立一个透明和开源的学术机制，让研究数据和模型公开共享，使研究成果可以得到他人的验证和进一步拓展。这个学术机制的核心就是倡导“复制性研究”。本文结合近年来西方社会科学界围绕复制性研究的实践和讨论，提出了复制性研究对于社会科学特别是中国社会学定量分析的重要意义、操作可能性以及可能存在的争议。在此基础上，本文通过两个实证定量分析的例子，为如何进行复制性研究提供了简明的范本。最后，结合中国实际提出了就复制性研究所应达成的共识和建议。本文所论及的复制性研究，对社会科学定量分析其他领域也同样具有参考价值。

PDF

695

摘要：

摘要：社会科学定量分析亟待建立一个透明和开源的学术机制，让研究数据和模型公开共享，使研究成果可以得到他人的验证和进一步拓展。这个学术机制的核心就是倡导“复制性研究”。本文结合近年来西方社会科学界围绕复制性研究的实践和讨论，提出了复制性研究对于社会科学特别是中国社会学定量分析的重要意义、操作可能性以及可能存在的争议。在此基础上，本文通过两个实证定量分析的例子，为如何进行复制性研究提供了简明的范本。最后，结合中国实际提出了就复制性研究所应达成的共识和建议。本文所论及的复制性研究，对社会科学定量分析其他领域也同样具有参考价值。

关键词: 开源社会学 , 复制性研究 , 研究方法 , 定量分析

Abstract:

The development of quantitative analysis in social science calls for the establishment of an open and transparent mechanism to allow research data and models to be shared to the community, and the research findings to be verify and elaborated by others. The core in this practice is replication studies in social sciences. In this essay, we review the discussion and practices of replication studies in western social sciences and highlight its significance in promoting quantitative social science research in China.  Why is the replication study necessary? First, the collectivity of social science research suggests the research process should be transparent; Second, the principle of parsimony in quantitative social analysis requires the verification of research process; Third, the issue of endogeneity calls for extension of the past quantitative social analysis; Four, as a matter of fact, the rate of successful verification in published social science literature is low. Fifth, replication study is important to the teaching and training of quantitative social science; and finally, replication study can facilitate the professionalization of quantitative sociology and make us of latedevelopment advantages. Why is replication study feasible? First, the nature of quantitative data suggests that the results be verifiable; second, the development of techniques in quantitative social research makes it possible to extend the replications; third, internet provides an easy platform for sharing data and working programs; third, the practice of replication studies in other disciplines such as economics in other countries has provided valuable experience; and finally, replicable studies can increase the academic reputation of the author and the journal. Notwithstanding what are said above, there are many concerns that need to be addressed in promoting replication study. The first is the ownership of the data analyzed. The author may not have rights to share the data with other colleagues. The second issue is that it may bring extra burden to the authors, reviewers, and replicators. There might also be issues related to the sharing of the programs, the choice between the sharing of original data and working data, and results from replication studies may lead to pressures on the authors and personal attacks. We use two empirical examples as the template how to conduct the replication and the extension. The first example by Chen, Volker and Flap (2011), using more sophisticated modeling to replicate a study by two Dutch sociologists on how the use of social capital affects the attainment of the first job in East Germany. The second example is a rejoinder to the replication results on how selectivity of entry into the private sector affects earnings inequality in urban China. Both examples show how the replication and extensions help to advance the knowledge in the field. The paper concludes with the consensus and practical protocol that should be established in the academia of sociology.

Key words: opensource sociology  , replication study ,   research methodology  ,  quantitative analysis