线性回归方程中多重共线性诊断方法及其实证分析.pdf

( ) , ( 74 ) 2008(2) Journal of Huazhong Agricultural University( Social Sciences E it ion) 马雄威 ( 华中农业大学 经济管理学院, 湖北武汉430070) , , , ; ; : O212 : A : 1008- 3456( 2008) 02- 0078- 04 Diagnosis and Empirical Analysis on Multicollinearity in Linear Regression Model Ma Xiong-w ei (College of Economics and M anagement, H uaz hong A gricultural University , Wuhan, H ubei, 430070) Abstract With the popularity of computer application, multiple regression analysis has been exten- sively applie in pro uction, as w ell as scientific research in practice. But in the application of multiple re- gression analysis, the application con ition that there is no linear relationship betw een in epen ent varia- bles is apt to be overlooke ; therefore , the obtaine result may become incorrect an even be far from the original relationship among the variables. T herefore, it is necessary to analyze the causes an the im- pact of multicollinearity in the linear regression mo el, an intro uces some of the iagnosis metho s of multicollinearity. Then this paper intro uces the theory of the ri ge regression mo el, an the scope of its application, as w ell as a vantages an isa vantages. T hen w ith the example of pork prices factors analy- sis an the ai of SAS program, this paper objectively evaluates the characteristics of the metho an its strength an w eakness. Key words multiple regression analysis; linear regression mo el; multicollinearity x 1, , xm , 0 , 1 , , m , 1x 1 2 x2 mx m = 0 ( 1. 1) , , X k , , : (0 - j xj )