如何将多列种族人口统计变量合并为单列并标记混血情况？

我正在处理一份调查原始数据导出文件，目前处于数据清洗阶段。部分人口统计变量的结构不利于分析，具体来说，受访者可选择多个种族选项，数据导出后每个种族对应一列，选中的列填种族名称，未选的为NA。

复现数据集的R代码：

ID <- c(rep(c(1:8), 1))
White <- c("White",NA,NA,NA,NA,NA,"White","White")
Asian <- c(NA,NA,NA,NA,NA,"Asian",NA,"Asian")
SouthAfrican <- c(NA,"SouthAfrican",NA, NA,NA, NA, NA, "SouthAfrican")
Hispanic <- c(NA, NA, NA, NA, "Hispanic", "Hispanic", NA, NA)
WestAsian <- c(NA, NA, NA, NA, NA, NA, "WestAsian", NA)
PreferNotToAnswer <- c(NA, NA,"PreferNotToAnswer", "PreferNotToAnswer", NA, NA, NA, NA)

df <- data.frame(ID, White, Asian, SouthAfrican, Hispanic, WestAsian, PreferNotToAnswer)

我希望创建一个Race列，将各分散列中的值合并到该列中：

• 若受访者仅选择一个种族选项，直接填入该种族名称
• 若选择多个种族选项（如ID为8的受访者），则编码为“Mixed”

预期结果对应的R代码：

ID <- c(rep(c(1:8), 1))
White <- c("White",NA,NA,NA,NA,NA,"White","White")
Asian <- c(NA,NA,NA,NA,NA,"Asian",NA,"Asian")
SouthAfrican <- c(NA,"SouthAfrican",NA, NA,NA, NA, NA, "SouthAfrican")
Hispanic <- c(NA, NA, NA, NA, "Hispanic", "Hispanic", NA, NA)
WestAsian <- c(NA, NA, NA, NA, NA, NA, "WestAsian", NA)
PreferNotToAnswer <- c(NA, NA,"PreferNotToAnswer", "PreferNotToAnswer",NA, NA, NA, NA)
Race <- c("White","SouthAfrican","PreferNotToAnswer", "PreferNotToAnswer","Hispanic", "Mixed", "Mixed", "Mixed")

df <- data.frame(ID, White, Asian, SouthAfrican, Hispanic,WestAsian,PreferNotToAnswer,Race)

方法一：使用dplyr + tidyr（推荐，代码简洁易读） #

library(dplyr)
library(tidyr)

df_updated <- df %>%
  rowwise() %>%
  mutate(
    # 收集当前行非NA的种族值
    selected_races = list(na.omit(c_across(White:PreferNotToAnswer))),
    # 根据选中数量生成Race列
    Race = case_when(
      length(selected_races) == 1 ~ selected_races[[1]],
      length(selected_races) > 1 ~ "Mixed"
    )
  ) %>%
  ungroup() %>%
  select(-selected_races) # 删除临时辅助列

方法二：基础R实现（无需额外安装包） #

# 逐行处理，生成Race列
df$Race <- apply(df[, -1], 1, function(row) {
  non_na_vals <- na.omit(row)
  if (length(non_na_vals) == 1) {
    non_na_vals[1]
  } else if (length(non_na_vals) > 1) {
    "Mixed"
  } else {
    NA # 处理未选择任何选项的情况（若存在）
  }
})

结果验证 #

运行任意一种方法后，数据集的Race列将与预期结果完全匹配。

内容的提问来源于stack exchange，提问作者Darko