多维极化的测度:新的方法及其应用  

Multidimensional Polarization:New Measurement and Its Application

作　　者：洪兴建 董君[3] Hong Xingjian;Dong Jun

机构地区：[1]浙江财经大学数据科学学院 [2]浙江工商大学 [3]浙江工商大学统计与数学学院

出　　处：《统计研究》2020年第9期24-33,共10页Statistical Research

基　　金：国家社科基金重点项目“城乡共享发展成果的统计测度研究”(16ATJ002);浙江省A类一流学科(浙江财经大学统计学18A001)。

摘　　要：鉴于现有多维极化测度方法未能体现维度之间的相关性,本文引入余弦相似度,并基于相对剥夺理论提出了新的多维极化测度方法。该方法还可以进行静态和动态分解,能够分析每个维度及其变动对多维极化的贡献率。运用中国家庭追踪调查(CFPS)数据,本文从收入、财产、交通通信支出和受教育水平四个维度实证分析了我国城乡居民家庭之间的多维极化。实证结果表明,城乡长期多维极化呈现明显下降趋势,其中教育、交通通信支出和收入三个维度的极化效应是主要原因,而财产极化效应不利于多维极化的下降。As the existing measurements of multidimensional polarization fail to reflect the correlation between dimensions,this paper introduces cosine similarity,and proposes a new multidimensional polarization measurement based on the relative deprivation theory.The method can also be decomposed statically and dynamically,and can analyze the contribution rate of each dimension and its change to multidimensional polarization.Using the data from China Family Panel Studies(CFPS),this paper empirically analyzes the multidimensional polarization between urban and rural households in China from four dimensions:income,properties,expenditure on transportation and communications,and educational level.The empirical results show that the long-term multidimensional polarization shows a significant downward trend,and the polarization effect of education,expenditure on transportation and communications,and income is the main reason,but the polarization effect of properties promotes multidimensional polarization.

关 键 词：多维极化 测度方法 维度分解 城乡多维极化 

分 类 号：C81[社会学—统计学]