抽样调查中对定性变量之间关联分析的方法选择  

Choosing A Method of Analyzing the Relation Among Non-Metric Variables Based Sampling

作　　者：张士玉[1] 郝旭光[2] 

机构地区：[1]北京联合大学管理学院信息管理与电子商务系,北京100101 [2]对外经济贸易大学国际商学院人力资源系,北京100029

出　　处：《数理统计与管理》2014年第6期1038-1047,共10页Journal of Applied Statistics and Management

基　　金：国家社会科学基金资助面上项目(05BJL028);国家社会科学基金资助重点项目(11AZD010)

摘　　要：在社会和行为科学研究中大量通过抽样调查收集数据并对其进行分析,其中对定性变量之间的关联分析是许多研究项目的重要内容。面对同样的数据有不同可供选择的方法,而方法选取的正确与否,直接关系到研究结果的质量和正确性。本文从所研究变量的性质和研究深度这两个方面考察三大类方法,即:非参数统计、参数统计和关联规则发现方法,指出在不同情况下选择不同方法的思路和原则。这对于项目研究及调查和数据分析服务都具有一定的应用价值。There are many data be collected by sampling and be analyzed in social science and behavioral science research, it is very important that analyzing the relation among the non-metric variables in many research objects. There are many methods can be selected for the same data, but the method be selected is or not correct relate to the quality and efficacy of the result of reach. Three types of method are studied in this paper, such as the method of non-parametric statistics, parametric statistics and association rules. The idea and principle be put forward used for the selecting correct method according the property of the variables and the reach depth. It has some application value for project research and sampling service.

关 键 词：抽样调查 定性变量 关联 数据分析 方法 

分 类 号：C812[社会学—统计学] O212[理学—概率论与数理统计]