教育和心理研究中的多层线性模型  

HIERARCHICAL LINEAR MODEL IN THE STUDY OF EDUCATION AND PSYCHOLOGY

作　　者：刘红云[1] 孟庆茂[1] 

机构地区：[1]北京师范大学心理系,北京100875

出　　处：《心理科学进展》2002年第2期213-219,共7页Advances in Psychological Science

摘　　要：多层线性模型是分析具有层次结构数据的一种新型统计分析技术,与传统统计方法相比,具有模型假设与实际更吻合、结果解释更合理等特点。近年来这一方法的应用逐渐在社会科学的研究中受到重视。文章从多层线性模型的基本假设入手,较系统地介绍了模型参数估计和假设检验的方法,并通过一个具体例子将这一方法与传统回归分析方法相比,进一步说明了多层线性模型在分析具有层次结构数据时的优点。Hierarchical Linear Model (HLM) is a new tool for analyzing hierarchical structure data. It is superior to traditional statistical methods in that HLM based hypothesis is more compatible with facts and that the results generated from the model are reasonable. Therefore, HLM has been applied more and more wildly in social science. This paper is an introduce to HLM which introduces the fundamental hypothesis of the model, parameter estimator and hypothesis testing methods, as well as its advantages over the traditional regression method in dealing with hierarchical structure data.

关 键 词：层线性模型 回归分析 最小二乘估计 迭代最小二乘估计 

分 类 号：B841.2[哲学宗教—基础心理学]