6多元回归分析进一步问题.ppt

Chapter Outline 本章大纲 Effects of Data Scaling on OLS Statistics 数据的测度单位换算对OLS统计量的影响 More on Functional Form 对函数形式的进一步讨论 More on Goodness-of-Fit and Selection of Regressors 拟合优度和回归元选择的进一步探讨 Prediction and Residual Analysis 预测和残差分析 Lecture Notes 课堂笔记 Effects of Redefining variables 重新定义变量的影响 Estimated coefficients 估计系数 R squared R 平方 t statistics t 统计量 Functional form 函数形式 Logarithmic form 对数函数形式 Models with Quadratics 含二次式的模型 Models with interaction terms 含交叉项的模型 Redefining Variables重新定义变量 Why would we want to do so? 为什么我们想这样做？ Often, data scaling is used to reduce the number of zeros after a decimal point in an estimated coefficient, so that the results appear prettier. 数据测度单位变换经常被用于减少被估参数小数点后的零的个数，这样结果更好看一些。 Since this is mainly an action of decoration, we expect nothing essential should change. 既然这样做主要为了好看，我们希望本质的东西不改变。 Redefining Variables: An example重新定义变量：一个例子 Consider a model relating infant birth weight to cigarette smoking and family income: 以下模型反映了婴儿出生体重与孕妇吸烟量和家庭收入之间的关系： (1) Consider the following rescaling: 考虑如下单位变换： (2) Birth weight is changed from ounces to pounds 出生体重单位由盎司变为磅 (3)Number of cigarettes is changed to packs of cigrattes 香烟的支数变为包数 The estimation results is presented in the following table. 估计结果列于下表 Table 6.1 Impact of changing the scale of the dependent variable改变被解释变量测度单位的影响 Since 1lbs = 16 oz, the dependent variable is transformed by dividing 16. 因为1磅＝16盎司，被解释变量被除以16。 We compare columns (1) and (2). 比较第1列与第2列。 The estimated coefficients in (1)/16 = those in (2). (1)中被估参数/16＝ (2)中被估参数 The standard errors of estimated coefficients in (1)/16=those in (2) (1)中被估参数的标准差/16＝ (2)中被估参数的标准差 Impact of changing the scale of the dependent variable改变被解释变量测度单位的影响 The t statistics in (1) and (2) are identical. (1)和(2)中 t 统计量相同 The R squared are identical. R平方相同 SSR in (1) /(16*16)=SSR in (2) (1)中SSR/（16*16）＝ (2)中SSR SER (standard error) in (1)/16=SER in (2) (1)中SER(标准差)/16＝ (2)中SER Impact of changing the scale of the indep