非线性模型连续调节变量检验的实现方法

An Implement Method on Testing Continuous Moderator Variable in Non-linear Model

针对非线性模型中调节作用检验不能只依赖交叉项系数的问题,目前存在两个解决办法:构造更好的统计量和通过图形直观表现。但是已有研究对于非线性模型中的离散型和连续型调节变量没有作进一步区分,尤其是非线性模型中连续型调节变量的检验一直缺乏简便有效的方法。文章从模型和数据两个维度划分了四种调节变量的类型,梳理了不同类型调节变量的检验方法和存在的问题,针对连续型调节变量检验的问题,提出了对连续变量进行多点分割的处理方式,并采用三维作图的方法,直观展现了调节作用的连续变化。

With respect to the fact that the regulating effect verification in nonlinear models cannot only depend on the cross term coefficient, there are two solutions, namely constructing a more accurate statistics and using graphs to illustrate the moderator effect. However, the existing researches do not further differentiate the discrete-type and successive-type regulated variables in the nonlinear models, especially in which the verification of the successive-type regulated variables lack convenient and effective means all the way. This paper marks off four lypes of moderator variables from the two dimensions of models and data, and combs the test methods and problems of different moderator variables. Aiming at the problems in the successive variables test, the paper puts forward a process mode of the successive variables muhipoint segmentation, and also adopts three-dimensional graphics to il- lustrate the continuous change of the moderator effect.