如何使summary.matchit输出显示自定义变量标签？

问题根源 #

你当前代码的核心问题有两个：一是**matchit调用时使用了原始的lalonde数据集，而非添加了自定义标签的mydf**，导致变量标签未被传入匹配模型；二是分类变量（如race）会被自动生成哑变量，需要手动将哑变量的行名替换为「自定义标签+哑变量后缀」的格式。

完整实现代码 #

library(MatchIt)
library(labelled)
library(magrittr)

# 加载数据集并设置变量标签
mydf <- MatchIt::lalonde
mydf %<>% set_variable_labels(
  age = "MyAge",
  educ = "MyEduc",
  race = "MyRace"
)

# 调用matchit时使用带标签的mydf
m.out0 <- matchit(treat ~ age + educ + race + married + 
                    nodegree + re74 + re75,
                  data = mydf,  # 关键：改用处理后的mydf
                  method = "quick",
                  distance = "glm")

# 提取匹配后的汇总结果
sum_mat <- summary(m.out0, un = FALSE)$sum.matched[seq(1,5), seq(1,2)]

# 获取变量标签列表
var_labels <- var_label(mydf)

# 替换行名为自定义标签（兼容分类哑变量）
row_names <- rownames(sum_mat)
for (i in seq_along(row_names)) {
  rn <- row_names[i]
  if (rn == "distance") next
  
  # 从哑变量名中提取原始变量（如从raceblack提取race）
  orig_var <- sub("(.*?)(black|hispan|white)?", "\\1", rn)
  lab <- var_labels[[orig_var]]
  
  # 拼接标签与哑变量后缀
  if (grepl("(black|hispan|white)", rn)) {
    suffix <- sub(".*?(black|hispan|white)", "\\1", rn)
    new_rn <- paste0(lab, suffix)
  } else {
    new_rn <- lab
  }
  
  rownames(sum_mat)[i] <- new_rn
}

# 查看最终结果
sum_mat

输出结果 #

Means Treated    Means Control
distance      0.5774355466557  0.5764010328956
MyAge         25.8162162162162 25.4475708575709
MyEduc        10.3459459459459 10.6780791680792
MyRaceblack   0.8432432432432  0.8376126126126
MyRacehispan  0.0594594594595  0.0606976956977

内容的提问来源于stack exchange，提问作者robertspierre