解释广义线性混合模型 (GLMM) 中的独立分类变量
Interpreting independent categorical variable in a generalized linear mixed model (GLMM)
我使用具有准泊松回归的广义线性混合模型 (GLMM)，并使用惩罚拟似然法（即 glmmPQL）用多元正态随机效应拟合模型。输出结果如下：
收入变量有 3 类，低收入、中低收入、中高收入。在输出中，低收入似乎是参考类别，但我不知道我应该如何解释和报告它。
非常感谢您。
Linear mixed-effects model fit by maximum likelihood
Data: my_scaled_data
AIC BIC logLik
NA NA NA
Random effects:
Formula: ~1 | country
(Intercept) Residual
StdDev: 1.191246 7.062197
Variance function:
Structure: fixed weights
Formula: ~invwt
Fixed effects: protests ~ stringency + cpi + income
Value Std.Error DF t-value p-value
(Intercept) 3.993691 0.3732307 428 10.700329 0.0000
stringency 0.152788 0.0322449 428 4.738373 0.0000
cpi -0.509498 0.3093523 428 -1.646984 0.1003
incomelower middle income -0.028550 0.2156300 428 -0.132403 0.8947
incomeupper middle income -0.528267 0.2520429 428 -2.095941 0.0367
Correlation:
(Intr) strngn cpi incmlmi
stringency -0.005
cpi 0.065 -0.311
incomelower middle income -0.302 -0.089 0.056
incomeupper middle income -0.244 -0.060 -0.004 0.539
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-1.6874331 -0.4638920 -0.1344516 0.2557120 10.2539363
Number of Observations: 444
Number of Groups: 12
Income variable has 3 categories, low income, lower middle income, upper middle income. In the output, low income appears to be reference category but I don't know how should I interpret and report this.
这是处理分类回归变量的常规方法。每个估计都被解释为与参考水平的对比。因此，incomelower middle income 的线性预测变量比参考水平低 0.028550，incomeupper middle income 的线性预测变量比参考水平低 0.528267。
我使用具有准泊松回归的广义线性混合模型 (GLMM)，并使用惩罚拟似然法（即 glmmPQL）用多元正态随机效应拟合模型。输出结果如下：
收入变量有 3 类，低收入、中低收入、中高收入。在输出中，低收入似乎是参考类别，但我不知道我应该如何解释和报告它。
非常感谢您。
Linear mixed-effects model fit by maximum likelihood
Data: my_scaled_data
AIC BIC logLik
NA NA NA
Random effects:
Formula: ~1 | country
(Intercept) Residual
StdDev: 1.191246 7.062197
Variance function:
Structure: fixed weights
Formula: ~invwt
Fixed effects: protests ~ stringency + cpi + income
Value Std.Error DF t-value p-value
(Intercept) 3.993691 0.3732307 428 10.700329 0.0000
stringency 0.152788 0.0322449 428 4.738373 0.0000
cpi -0.509498 0.3093523 428 -1.646984 0.1003
incomelower middle income -0.028550 0.2156300 428 -0.132403 0.8947
incomeupper middle income -0.528267 0.2520429 428 -2.095941 0.0367
Correlation:
(Intr) strngn cpi incmlmi
stringency -0.005
cpi 0.065 -0.311
incomelower middle income -0.302 -0.089 0.056
incomeupper middle income -0.244 -0.060 -0.004 0.539
Standardized Within-Group Residuals:
Min Q1 Med Q3 Max
-1.6874331 -0.4638920 -0.1344516 0.2557120 10.2539363
Number of Observations: 444
Number of Groups: 12
Income variable has 3 categories, low income, lower middle income, upper middle income. In the output, low income appears to be reference category but I don't know how should I interpret and report this.
这是处理分类回归变量的常规方法。每个估计都被解释为与参考水平的对比。因此，incomelower middle income 的线性预测变量比参考水平低 0.028550，incomeupper middle income 的线性预测变量比参考水平低 0.528267。