分层线性模型层-1自变量中心化问题研究综述

Centering of Level-1 Independent Variables in Hierarchical Linear Models

层-1自变量中心化方法的选择是分层线性模型中的一个关键问题,它直接影响到模型参数的估计和解释。国外已有很多学者对其进行了理论和实践方面的研究,但是在实证研究中选择适当的中心化方法仍然存在困难,为能够给研究人员提供一点参考依据,在给出两层模型自变量的三种主要中心化方法以后,回顾了分层线性模型中层-1自变量中心化产生的影响,总结了具体应用中不同中心化方法的适用情况和如何进行选择的一些建议。

It is one of the key questions in hierarchical linear models that how a researcher chooses to scale the level - 1 independent variables, which directly influences the estimation and interpretation of the parameters. The issue of centering in theory and practice has been discussed by many researchers. Unfortunately, there are some difficulties in empirical research about centering decisions. In order to provide a certain reference basis for researchers, three different scaling options will be considered. The influence about the centering approach of level - 1 predictors in HIM is reviewed, and a overview of the linkage between substantive research questions and the form of centering is given, and a number of practical recommendation are provided.