異方差實(shí)驗(yàn)報(bào)告

1、精選優(yōu)質(zhì)文檔-傾情為你奉上附件二：實(shí)驗(yàn)報(bào)告格式（首頁(yè)） 山東輕工業(yè)學(xué)院實(shí)驗(yàn)報(bào)告 成績(jī) 課程名稱 計(jì)量經(jīng)濟(jì)學(xué) 指導(dǎo)教師 實(shí)驗(yàn)日期 2013.5.18 院（系） 商學(xué)院會(huì)計(jì)系 專業(yè)班級(jí) 會(huì)計(jì) 實(shí)驗(yàn)地點(diǎn) 實(shí)驗(yàn)樓二機(jī)房 學(xué)生姓名 學(xué)號(hào) 同組人 無(wú) 實(shí)驗(yàn)項(xiàng)目名稱 異方差的檢驗(yàn) 一、 實(shí)驗(yàn)?zāi)康暮鸵?、理解異方差的含義后果、2、 學(xué)會(huì)異方差的檢驗(yàn)與加權(quán)最小二乘法要求熟悉基本操作步驟，讀懂各項(xiàng)上機(jī)榆出結(jié)果 的含義并進(jìn)行分析 3、掌握異方差性問(wèn)題出現(xiàn)的來(lái)源、后果、檢驗(yàn)及修正的原理，以及相關(guān)的Eviews操 作方法 4、練習(xí)檢查和克服模型的異方差的操作方法。 5、掌握異方差性的檢驗(yàn)及處理方法 6、用圖示法、斯皮

2、爾曼法、戈德菲爾德、white驗(yàn)證法，驗(yàn)證該模型是否存在異方差 二、 實(shí)驗(yàn)原理1、異方差的檢驗(yàn)出消除方法2、運(yùn)用EVIEWS軟件及普通最小二乘法進(jìn)行模型估計(jì)3、檢驗(yàn)?zāi)Ｐ偷漠惙讲钚圆?duì)其進(jìn)行調(diào)整三、 主要儀器設(shè)備、試劑或材料Eviews軟件、課本教材、電腦 四、 實(shí)驗(yàn)方法與步驟一、準(zhǔn)備工作。建立工作文件，并輸入數(shù)據(jù)，用普通最小二乘法估計(jì)方程（操作步驟與方法同前），得到殘差序列。1、CREATE U 1 31 回車2、DATA Y X 回車 輸入數(shù)據(jù)obsYX1264877721059210390995441311050851221097961071191274061274785031349994

3、31142691058815522118981673012950176631377918575148191963515122221163161702228801715782412718165425604191400265002018292676021220028300222017274302321052956024160028150252250321002624203250027257035250281720335002919003600030210036200312800382003、LS Y C X 回車用最小二乘法進(jìn)行估計(jì)出現(xiàn)Dependent Variable: YMethod: Le

4、ast SquaresDate: 05/18/13 Time: 11:19Sample: 1 31Included observations: 31VariableCoefficientStd. Errort-StatisticProb. C-700.411116.6679-6.003460X0.0.18.195750R-squared0. Mean dependent var1266.452Adjusted R-squared0. S.D. dependent var846.757S.E. of regression244.4088 Akaike info criterion13.8979S

5、um squared resid Schwarz criterion13.99042Log likelihood-213.418 F-statistic331.0852Durbin-Watson stat1. Prob(F-statistic)0用普通最小二乘法進(jìn)行估計(jì)，估計(jì)結(jié)果如下=700.41+0.Xi R2=0.92 =0.92 F=335.82 t=(-6.0) (18.2) 括號(hào)內(nèi)為t統(tǒng)計(jì)量。1=0.說(shuō)明在其他因素不變的情況下，可支配收入每增長(zhǎng)1元，個(gè)人儲(chǔ)蓄平均增長(zhǎng)0.元。=0.92 , 擬合程度較好。在給定=0.05時(shí)，t=18.2 > =2.055 ，拒絕原假設(shè)，說(shuō)明銷售收

6、入對(duì)銷售利潤(rùn)有顯著性影響。F=335.82 > = 4.18 ,表明方程整體顯著。(一).圖示檢驗(yàn)法分別繪制X、Y坐標(biāo)系散點(diǎn)圖，命令如下：Scat x yGenr e2=resid2Scat x e2出現(xiàn)可以看出，隨著可支配收入x的增加，儲(chǔ)蓄y的離散程度增加，表明隨機(jī)誤差項(xiàng)ui存在異方差性。（二）斯皮爾曼等級(jí)相關(guān)系數(shù)檢驗(yàn)命令scat x e2sort x data x dd1（輸入1-31）ls y c x 出現(xiàn)Dependent Variable: YMethod: Least SquaresDate: 05/18/13 Time: 11:25Sample: 1 31Included

7、observations: 31VariableCoefficientStd. Errort-StatisticProb.  C-700.4110116.6679-6.0.0000X0.0.18.195750.0000R-squared0.    Mean dependent var1266.452Adjusted R-squared0.    S.D. dependent var846.7570S.E. of regression244.4088    

8、Akaike info criterion13.89790Sum squared resid.    Schwarz criterion13.99042Log likelihood-213.4175    F-statistic331.0852Durbin-Watson stat1.    Prob(F-statistic)0.sort x data x dd1ls y c x genr e1=abs(resid)sort e1 data e1 dd2genr r=1-6*s

9、um(dd2-dd1)2)/(313-31)genr z=r*sqrt(30)出現(xiàn)3.326即等級(jí)相關(guān)數(shù)是顯著的，說(shuō)明儲(chǔ)蓄計(jì)量模型的隨機(jī)誤差項(xiàng)存在異方差性。R=0. z=3. 給定顯著性水平=0.05，查正態(tài)分布表，得,因?yàn)閆=3.331.96，所以拒絕H0,接受H1,即等級(jí)相關(guān)系數(shù)是顯著的，說(shuō)明儲(chǔ)蓄計(jì)量模型的隨機(jī)誤差項(xiàng)存在異方差性。(三)、Goldfeld-Quant檢驗(yàn)命令sort x smpl 1 11 ls y c x Dependent Variable: YMethod: Least SquaresDate: 05/18/13 Time: 11:36Sample: 1 11Incl

10、uded observations: 11VariableCoefficientStd. Errort-StatisticProb.  C-744.6351195.4108-3.0.0041X0.0.5.0.0003R-squared0.    Mean dependent var331.3636Adjusted R-squared0.    S.D. dependent var260.8157S.E. of regression129.4724    A

11、kaike info criterion12.72778Sum squared resid.9    Schwarz criterion12.80012Log likelihood-68.00278    F-statistic31.58011Durbin-Watson stat1.    Prob(F-statistic)0.smpl 21 31 ls y c x Dependent Variable: YMethod: Least SquaresDate: 05/18/1

12、3 Time: 11:39Sample: 21 31Included observations: 11VariableCoefficientStd. Errort-StatisticProb.  C666.3811911.25850.0.4832X0.0.1.0.1352R-squared0.    Mean dependent var2152.909Adjusted R-squared0.    S.D. dependent var354.4462S.E. of regression327.7

13、867    Akaike info criterion14.58557Sum squared resid.0    Schwarz criterion14.65791Log likelihood-78.22063    F-statistic2.Durbin-Watson stat2.    Prob(F-statistic)0.計(jì)算F=rss2rss1=0.2960>F0.05(9,9)=3.18，說(shuō)明儲(chǔ)蓄計(jì)量模型的隨機(jī)誤差項(xiàng)

14、存在異方差.記下第一個(gè)殘差平方和：.9 記下第二個(gè)殘差平方和：.0。計(jì)算F=6.41，給定顯著性水平=0.05，查F分布表V1=V2=11-2=9,F0.05(9,9)=3.18,因?yàn)镕=6.413.18，所以接受備擇假設(shè)，即儲(chǔ)蓄計(jì)量模型的隨機(jī)誤差項(xiàng)存在異方差性。(四).White檢驗(yàn)命令smpl 21 31 ls y c x smpl 1 31 ls y c x出現(xiàn)Dependent Variable: YMethod: Least SquaresDate: 05/18/13 Time: 11:43Sample: 1 31Included observations: 31VariableCo

15、efficientStd. Errort-StatisticProb.  C-700.4110116.6679-6.0.0000X0.0.18.195750.0000R-squared0.    Mean dependent var1266.452Adjusted R-squared0.    S.D. dependent var846.7570S.E. of regression244.4088    Akaike info criterion13.89

16、790Sum squared resid.    Schwarz criterion13.99042Log likelihood-213.4175    F-statistic331.0852Durbin-Watson stat1.    Prob(F-statistic)0.White Heteroskedasticity Test:F-statistic5.    Probability0.Obs*R-squared9. 

17、   Probability0.Test Equation:smpl 1 31 ls y c xDependent Variable: RESID2Method: Least SquaresDate: 05/18/13 Time: 11:46Sample: 1 31Included observations: 31VariableCoefficientStd. Errort-StatisticProb.  C19975.9882774.930.0.8111X-2.8.-0.0.7879X20.0.0.0.4135R-squared0. 

18、;   Mean dependent var55881.73Adjusted R-squared0.    S.D. dependent var77875.67S.E. of regression67748.39    Akaike info criterion25.17675Sum squared resid1.29E+11    Schwarz criterion25.31553Log likelihood-387.2397 

19、60;  F-statistic5.Durbin-Watson stat2.    Prob(F-statistic)0.Obs*R-squared=9.>X2(0.05)=6.0，所以結(jié)論是該回歸模型中存在異方差.因?yàn)門R2=31×0.2936=9.1，所以結(jié)論是該回歸模型中存在異方差.其中obs*R-squared等于9.表示的就是統(tǒng)計(jì)量TR2的值。(五)克服異方差命令smpl 1 31 ls y c x Dependent Variable: YMethod: Least SquaresDate: 05/18/1

20、3 Time: 11:52Sample: 1 31Included observations: 31VariableCoefficientStd. Errort-StatisticProb.  C-700.4110116.6679-6.0.0000X0.0.18.195750.0000R-squared0.    Mean dependent var1266.452Adjusted R-squared0.    S.D. dependent var846.7570S.E. of regressi

21、on244.4088    Akaike info criterion13.89790Sum squared resid.    Schwarz criterion13.99042Log likelihood-213.4175    F-statistic331.0852Durbin-Watson stat1.    Prob(F-statistic)0.genr ww=1/abs(resid)ls (w=1/x) y c xDependent Variable: YMethod: Least SquaresDate: 05/18/13 Time: 12:03Sample: 1 31Included observations: 31VariableCoefficientStd. Errort-StatisticProb.  C-700.4110116.6679-6.0.0000X0.0.18.195750.0000R-squared0.    Mean dependent var1266.