基于方差分析的回归元边际贡献的实证研究

The Empirical Study on the Marginal Contribution of Regression Based on the Analysis of Variance

论文主要对回归元的边际贡献进行了研究分析,首先具体详细地阐述了边际贡献的概念,然后简要地介绍了方差分析方法,紧接着选取世界上66个国家的儿童死亡率、成年女性识字率和人均国民收入三个因素的数据,通过这样一个实例从方差分析的视角对回归元的边际贡献进行了具体分析。通过单因素分析、双因素分析并利用F检验分析法,得出结论:只要新增加的回归元能增加调整后的拟合优度值,且在统计上是高度显著的,即使调整后的拟合优度值稍微低点也应该把它保留在模型当中。

This paper studies and analyzes the marginal contribution of the regressor.At the first, the article illus-trates the concept of marginal contribution in detail and introduces briefly the analysis of variance, and then the ar-ticle selects the data of three factors such as the children mortality, the adult female literacy rate and pgnp from 66 countries in the world.At last this paper analyzes specially the marginal contribution of regressor through an exam-ple from the perspective of variance analysis.By using the single factor analysis, double factor analysis and the F-test analysis, the article draws the following conclusion： it should also retain the new regressor in the model even when the adjusted goodness of fit values is a little low, as long as the new regressor could improve the adjusted goodness of fit values and it is highly significant in statistics.