从逻辑回归模型 (R::lme4) 中提取具有 95% CI 的预测变量效应答案

【问题标题】：Extracting a predictor's effect with 95% CIs from a logistic regression model (R::lme4)从逻辑回归模型 (R::lme4) 中提取具有 95% CI 的预测变量效应
【发布时间】：2021-09-25 00:20:35
【问题描述】：

我有一个逻辑回归模型，使用 logit 链接。如何在“y”（包括 95% CI）的概率尺度上提取预测变量的“x”效应？预测变量“x”是一个连续变量。

数据

library(tidyverse)
n = 100
a = tibble(y = rep(c("pos", "neg", "neg", "neg"), length.out = n), x = rep(3, length.out = n), group = rep(letters[1:7], length.out = n))
b = tibble(y = rep(c("pos", "pos", "neg", "neg"), length.out = n), x = rep(2, length.out = n), group = rep(letters[1:7], length.out = n))
c = tibble(y = rep(c("pos", "pos", "pos", "neg"), length.out = n), x = rep(1, length.out = n), group = rep(letters[1:7], length.out = n))
d = rbind(a, b)
df = rbind(d, c)
df = df %>% mutate(y = as.factor(y))
df

[![在此处输入图片描述][1]][1]

型号

library("lme4")
m = glmer(
  y ~ x + (x | group), 
  data = df, 
  family = binomial(link = "logit"))
m

总结

Generalized linear mixed model fit by maximum likelihood (Laplace Approximation) [glmerMod]
 Family: binomial  ( logit )
Formula: y ~ x + (x | group)
   Data: df
      AIC       BIC    logLik  deviance  df.resid 
 373.5635  392.0824 -181.7817  363.5635       295 
Random effects:
 Groups Name        Std.Dev.  Corr
 group  (Intercept) 0.000e+00     
        x           3.961e-09  NaN
Number of obs: 300, groups:  group, 7
Fixed Effects:
(Intercept)            x  
      2.197       -1.099  
optimizer (Nelder_Mead) convergence code: 0 (OK) ; 0 optimizer warnings; 1 lme4 warnings

【问题讨论】：

标签： r regression lme4

【解决方案1】：

你可以使用confint()，更多细节见the help page，在你的例子中，你没有模拟组的随机效应，所以(x|groups)可能是不可估计的。

这是示例数据集的示例，我将因变量 Reaction 离散化以与二项式一起使用：

library(lme4)
df = sleepstudy
df$Reaction = ifelse(sleepstudy$Reaction>300,1,0)
m = glmer(Reaction ~ Days + (Days | Subject), df,family="binomial") 

                 2.5 %     97.5 %
.sig01       0.3926123  4.5486979
.sig02      -0.9169549  1.0000000
.sig03       0.0000000  0.8148731
(Intercept) -5.9765500 -2.1041927
Days         0.3960485  1.0942084

【讨论】：