如何在R语言中从5级变量生成多个3级变量并保留原变量

Got it, let's solve this problem! You want to retain your original 5-level variables while generating new 3-level variables with the _3l suffix—mirroring the behavior of Stata's recode ... gen(...) command. Here's a clean, modern R implementation using dplyr:

Step 1: Reproduce Your Data #

First, let's recreate your original dataset to work with:

library(dplyr)

# Original 5-level data
df <- structure(
  list(
    Question1 = c("I", "5", "4", "4"), 
    Question2 = c("I", "5", "4", "4"), 
    Question3 = c("I", "3", "2", "4")
  ), 
  class = c("tbl_df", "tbl", "data.frame"), 
  row.names = c(NA, -4L)
)

Step 2: Generate New 3-Level Variables (Keep Originals) #

Use mutate(across(...)) (the updated replacement for mutate_at) to batch-process your variables, recode them, and create new columns without overwriting the originals:

df_updated <- df %>%
  mutate(
    # Target the variables you want to recode
    across(
      c(Question1, Question2, Question3),
      # Define your recoding rules (matches your Stata logic)
      ~recode(
        .,
        '1' = '1', '2' = '1',   # 5-level 1/2 → 3-level 1
        '3' = '2',             # 5-level 3 → 3-level 2
        '4' = '3', '5' = '3',  # 5-level 4/5 → 3-level 3
        'I' = NA_character_    # 'I' maps to NA
      ),
      # Auto-name new columns with _3l suffix
      .names = "{col}_3l"
    )
  )

Step 3: Verify the Result #

When you print df_updated, you'll get exactly the structure you're looking for:

print(df_updated)

Output:

# A tibble: 4 × 6
  Question1 Question2 Question3 Question1_3l Question2_3l Question3_3l
  <chr>     <chr>     <chr>     <chr>        <chr>        <chr>       
1 I         I         I         NA           NA           NA          
2 5         5         3         3            3            2           
3 4         4         2         3            3            1           
4 4         4         4         3            3            3           

Bonus: Batch-Process Variables by Pattern #

If you have many variables to recode (like Stata's varlist range), you can use a pattern match instead of listing each variable. For example, if all target variables start with "Question":

df_updated <- df %>%
  mutate(
    across(
      starts_with("Question"),
      ~recode(., '1'='1', '2'='1', '3'='2', '4'='3', '5'='3', 'I'=NA_character_),
      .names = "{col}_3l"
    )
  )

This approach keeps your original data intact while efficiently generating the new 3-level variables you need.

内容的提问来源于stack exchange，提问作者JohnD