6.随机效应的显著性检验
#使用nlme的lme函数拟合多层模型,通过summary()获得的输出结果中不包含对于随机效应的显著性的检验结果。
- Model3.6 <-
- lme(
- fixed = geread ~ gevocab + senroll + gevocab * senroll,
- random = ~ 1 | school,
- data =
- Achieve
- )
- summary(Model3.6)
复制代码 #结果如下
Linear mixed-effects model fit by REML
Data: Achieve
AIC BIC logLik
43175.57 43219.02 -21581.79
Random effects:
Formula: ~1 | school
(Intercept) Residual
StdDev: 0.316492 1.940268
Fixed effects: geread ~ gevocab + senroll + gevocab * senroll
Value Std.Error DF t-value p-value
(Intercept) 1.7477004 0.17274011 10158 10.117513 0.0000
gevocab 0.5851202 0.02986497 10158 19.592189 0.0000
senroll 0.0005121 0.00031863 158 1.607242 0.1100
gevocab:senrol l -0.0001356 0.00005379 10158 -2.519975 0.0118
Correlation:
(Intr) gevocb senrll
gevocab -0.782
senroll -0.958 0.735
gevocab:senroll 0.752 -0.960 -0.766
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-3.1228018 -0.5697103 -0.2090374 0.3187827 4.4358936
Number of Observations: 10320
Number of Groups: 160
#为了获得随机效应的显著性检验的结果,须使用intervals函数
#结果如下,若置信区间下限(lower)与上限(upper)之内不包含0,则相应的效应值(est.)显著
Approximate 95% confidence intervals
Fixed effects:
lower est. upper
(Intercept) 1.4090956090 1.7477003581 2.086305e+00
gevocab 0.5265789741 0.5851202223 6.436615e-01
senroll -0.0001172069 0.0005121095 1.141426e-03
gevocab:senroll -0.0002410031 -0.0001355577 -3.011228e-05
attr(,"label")
[1] "Fixed effects:"
Random Effects:
Level: school
lower est. upper
sd((Intercept)) 0.2646107 0.316492 0.3785455
Within-group standard error:
lower est. upper
1.913779 1.940268 1.967123