摘要
论证线性回归分析指标R^2拟合意义的局限性,并质疑R^2检验的正当性。通过经济计量学课程实验资料揭示R^2的局限,实证分析经济计量学软件Eviews R^2计算公式的误区。此外,提出并建议采用残差误差率指标作为回归分析拟合优度的可决系数较为适宜。
This paper points out the imperfections of R2in its meaning of good- ness of fit for linear regression, and also questions the proper conduct of R2about its statistical test. With curriculum cases of econometrics we verify its imperfec- tions and confirm the calculation errors made by Eviews. We suggest a new error ratio to describe the goodness of fit to determine a linear regression model.
出处
《数量经济技术经济研究》
CSSCI
北大核心
2013年第9期152-160,共9页
Journal of Quantitative & Technological Economics