Model Fit StatisticsInterceptIntercept andCriterion Only CovariatesAIC 144.206 59.848SC 147.230 147.540-2 Log L 142.206 1.848Testing Global Null Hypothesis:BETA=0Test Chi-Square DF Pr > ChiSqLikelihood Ratio 140.3582 28
来源:学生作业帮助网 编辑:六六作业网 时间:2024/11/28 15:36:33
Model Fit StatisticsInterceptIntercept andCriterion Only CovariatesAIC 144.206 59.848SC 147.230 147.540-2 Log L 142.206 1.848Testing Global Null Hypothesis:BETA=0Test Chi-Square DF Pr > ChiSqLikelihood Ratio 140.3582 28
Model Fit Statistics
Intercept
Intercept and
Criterion Only Covariates
AIC 144.206 59.848
SC 147.230 147.540
-2 Log L 142.206 1.848
Testing Global Null Hypothesis:BETA=0
Test Chi-Square DF Pr > ChiSq
Likelihood Ratio 140.3582 28
Model Fit StatisticsInterceptIntercept andCriterion Only CovariatesAIC 144.206 59.848SC 147.230 147.540-2 Log L 142.206 1.848Testing Global Null Hypothesis:BETA=0Test Chi-Square DF Pr > ChiSqLikelihood Ratio 140.3582 28
从这两部分结果可推断出这是用SAS做Logistic回归的结果,该模型共有28个变量,有152个样本.
第一部分是一些模型拟合统计量.
(1)Intercept Only那一列是零模型对应的拟合统计量,Intercept and Covariates那一列是model语句所描述的模型对应的拟合统计量.
(2)-2LogL=142.206类似于多元线性回归的总平方和(TSS),-2LogL=1.848类似于多元线性回归的残差平方和(ESS),第二部分的似然比检验卡方统计量140.3582=142.206-1.848.
(3)AIC(即Akaike's information criterion)、SC(即Schwartz criterion)均为对-2LogL的修正.
这三个统计量越小,说明模型对数据拟合得越好!
第二部分是回归方程的显著性检验.
假设检验的原假设是——该模型和零模型效果一样,即所有自变量的回归系数均为0!似然比检验的p值