【文章內容簡介】 b(Fstatistic)White檢驗White Heteroskedasticity Test:Fstatistic ProbabilityObs*Rsquared ProbabilityTest Equation:Dependent Variable: RESID^2Method: Least SquaresDate: 12/26/13 Time: 10:34Sample: 1 29Included observations: 29VariableCoefficientStd. ErrortStatisticProb. CXX^2Rsquared Mean dependent varAdjusted Rsquared . dependent var. of regression Akaike info criterionSum squared resid+08 Schwarz criterionLog likelihood FstatisticDurbinWatson stat Prob(Fstatistic) （） （） （） T=29所以該回歸模型不存在異方差。（4）戈德菲爾德夸特檢驗第一個樣本輸出Dependent Variable: YMethod: Least SquaresDate: 12/26/13 Time: 10:49Sample: 1 11Included observations: 11VariableCoefficientStd. ErrortStatisticProb. CXRsquared Mean dependent varAdjusted Rsquared . dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood FstatisticDurbinWatson stat Prob(Fstatistic)殘差平方和=第二個樣本輸出Dependent Variable: YMethod: Least SquaresDate: 12/26/13 Time: 10:50Sample: 19 29Included observations: 11VariableCoefficientStd. ErrortStatisticProb. CXRsquared Mean dependent varAdjusted Rsquared . dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood FstatisticDurbinWatson stat Prob(Fstatistic)殘差平方和=提出原假設，：備擇假設，：互不相同。構造F統(tǒng)計量給出顯著性水平=，查F分布表=9，因為F=，所以接受原假設，即城鎮(zhèn)居民人均生活費計量模型的隨機誤差不存在異方差。實驗九P158第六章第3題Dependent Variable: YMethod: Least SquaresDate: 12/26/13 Time: 11:43Sample: 1975 1994Included observations: 20VariableCoefficientStd. ErrortStatisticProb. CXRsquared Mean dependent varAdjusted Rsquared . dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood FstatisticDurbinWatson stat Prob(Fstatistic)（1）線性回歸模型 （）（） = DW= T=20所以回歸方程擬合效果較好，但是DW值比較低。（2）殘差圖LM檢驗BreuschGodfrey Serial Correlation LM Test:Fstatistic ProbabilityObs*Rsquared ProbabilityTest Equation:Dependent Variable: RESIDMethod: Least SquaresDate: 12/06/13 Time: 16:38Presample missing value lagged residuals set to zero.VariableCoefficientStd. ErrortStatisticProb. CXRESID(1)Rsquared Mean dependent varAdjusted Rsquared . dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood FstatisticDurbinWatson stat Prob(Fstatistic)Obs*RsquaredLM（BG）自相關檢驗輔助回歸式估計結果是=20*=DW==LM=，所以LM檢驗結果也說明的誤差項存在一階正自相關。假設 至少一個不等于0。,與相比，=，=，所以拒絕，接受，所以該誤差項存在一階自相關。（4）已知DW=，若給定，查表得DW檢驗臨界值。因為DW=， 依據(jù)判別規(guī)則，認為誤差項存在嚴重的自相關。估計得自相關系數(shù)。對原變量做廣義差分變換。令 以、（1976~1994年）為樣本再次回歸，得Dependent Variable: Y1Method: Least SquaresDate: 12/06/13 Time: 16:09Sample(adjusted): 1976 1994Included observations: 19 after adjusting endpointsVariableCoefficientStd. ErrortStatisticProb. CX1Rsquared Mean dependent varAdjusted Rsquared . dependent var. of regression Akaike info criterionSum squared resid Schwarz criterionLog likelihood FstatisticDurbinWatson stat Prob(Fstatistic) （）（）DW=，查臨界值表，若給定，查表得DW檢驗臨界值。因為DW=， 依據(jù)判別規(guī)則，認為誤差項不存在自相關。殘差圖如下：殘差圖%。實驗十P159第六章第4題Dependent Variable: YMethod: Least SquaresDate: 12/09/13 Time: 08:41Sample: 1960 2001Included observations: 42VariableCoefficientStd. ErrortStatisticProb. CGDPRsquared Mean dependent varAdjusted Rsquared . dependent var. of regression Akaike info criterionSum squared resid+08 Schwarz criterionLog likelihood FstatisticDurbinWatson stat Prob(Fstatistic)（1）線性回歸模型 （）（） = DW= T=42所以回歸方程擬合效果較好，但是DW值比較低。（2）LM檢驗：BreuschGodfrey Serial Correlation LM Test:Fstatistic ProbabilityObs*Rsquared ProbabilityTest Equation:Dependent Variable: RESIDMethod: Least SquaresDate: 12/09/13 Time: 08:51Presample missing value lagged residuals set to zero.VariableCoefficientStd. ErrortStatisticProb. CGDPRESID(1)Rsquared Mean dependent varAdjusted Rsquared . dependent var. of regression Akaike info criterionSum squared resid51414932 Schwarz criterionLog likelihood FstatisticDurbinWatson stat Prob(Fstatistic)假設 至少一個不等于0。,與相比，=，=，所以拒絕，接受，所以該誤差項存在一階自相關。（4）已知DW=，若給定，查表得DW檢驗臨界值。因為DW=， 依據(jù)判別規(guī)則，認為誤差項存在嚴重的自相關。估計得自相關系數(shù)。對原變量做廣義差分變換。令