R语言将18水平教育因子重编码为7分类时触发recursive indexing报错

你代码首行的取子集语法错误，是触发Error in .subset2(x, i, exact = exact) : recursive indexing failed at level 2的根本诱因：

• R语言中，用[[ ]]从数据框提取列时，括号内必须传入列名的字符串常量，而bes$education本身是education列存储的18个水平的取值向量，不是列名。相当于你把一长串分类标签塞给了索引位置，R会尝试逐层递归匹配这些标签对应的列，最终因为找不到匹配项抛出递归索引失败的错误。
• 退一步讲，就算你把取列语法修正为bes[["education"]]，原代码的逻辑也存在缺陷：管道仅将提取出来的向量传入recode()处理，处理完成后结果是一个独立的临时向量，没有通过赋值回写到bes数据框的对应列里，你后续查看原数据时根本看不到变量的变化，这也是你反复尝试不同函数都没有得到预期效果的重要原因。

就算修正了上述取列逻辑，以下细节问题也会导致重编码后大量值变为NA：

• 新分类标签和你最初的目标定义不匹配：原代码里将研究生学历标为Postgrad、其他类标为Other，和你要求的7个水平（no qualification、GCSE D-G、GCSE A*-C、A Level、Undergraduate、Postgraduate、其他）不一致。
• 匹配字符串里残留网页转义字符：原代码写的City&Guilds是网页爬取过程中生成的HTML转义内容，实际因子水平存储的是City&Guilds，用带&的字符串完全匹配不上真实水平。
• 字符串存在多余的尾随空格：原代码里'A level or equivalent '末尾多了一个空格，如果实际因子水平没有这个空格，也会匹配失败。

# 加载dplyr包，提供recode、mutate等函数
library(dplyr)

# 对education变量执行重编码，结果回写到原数据框
bes <- bes %>%
  mutate(
    # 直接在数据框上下文内引用education列，不需要手动写取子集语法
    education = recode(
      education,
      "No qualification" = "no qualification",
      "GCSE D-G, CSE grades 2-5, O level D-E" = "GCSE D-G",
      "Youth training certificate, skill seekers" = "GCSE D-G",
      "Clerical and commercial qualifications" = "GCSE D-G",
      "GCSE A*-C, CSE grade 1, O level grade A-C" = "GCSE A*-C",
      "Scottish Standard grades, Ordinary bands" = "GCSE A*-C",
      "Recognised trade apprenticeship" = "GCSE A*-C",
      "City&Guilds level 1, NVQ/SVQ 1 and equivalent" = "GCSE A*-C",
      "A level or equivalent" = "A Level",
      "Scottish Higher or equivalent" = "A Level",
      "City&Guilds level 2, NVQ/SVQ 2 and equivalent" = "A Level",
      "HNC/HND, City&Guilds level 4, NVQ/SVQ 4/5" = "A Level",
      "ONC/OND, City&Guilds level 3, NVQ/SVQ 3" = "A Level",
      "Univ/poly diploma" = "Undergraduate",
      "First degree" = "Undergraduate",
      "Nursing qualification" = "Undergraduate",
      "Teaching qualification" = "Undergraduate",
      "Postgraduate degree" = "Postgraduate",
      "Other technical, professional or higher qualification" = "其他"
    ) %>%
      # 按照你指定的顺序设置因子水平，避免后续分析、绘图时水平顺序混乱
      factor(levels = c("no qualification", "GCSE D-G", "GCSE A*-C", "A Level", "Undergraduate", "Postgraduate", "其他"))
  )

如果运行代码后发现有大量值被识别为NA，先执行以下命令打印出education列实际存储的所有因子水平，和你写的匹配字符串逐行比对，找出转义字符、尾随空格、拼写不一致的问题：

levels(bes$education)

内容的提问来源于stack exchange，提问作者Kuddiking