Working with invalid names in R

Blog post
Author

Steve Simon

Published

April 9, 2025

The R programming language has a few rules about names that you can assign for various objects. This can cause some issues with data frames and tibbles that you might be importing or exporting. There is a fairly standard, but somewhat awkward way to handle this.

The rules for names in R are fairly restrictive. A name cannot include blanks. It cannot include most special characters, such as the dash (-) or the slash (/). The two exceptions are the dot (.) and the underscore (_). Any combination of those two symbols plus any number (0-9) or any letter (A-Z, a-z) is fine, with one exception. You can’t start a name with a number. So a1 is okay, but 1a is not.

Invalid names during data import

You may encounter invalid names when you import data from a text file, from a spreadsheet, or from other statistical software. You can live with the invalid names with the careful use of backticks and quote marks. You can also change the names to names that are valid. I’ll show both approaches, though I strongly prefer changing to valid names.

When you import a text file or a spreadsheet, the file will often include the names of each variable as the first line. These names might not respect the limitations that R imposes.

Less common, but still a problem is the names from files created by other statistical software. The rules that those programs have for variable names may not work with R.

Use backticks or quote marks

You can still work with data where one or more variables have invalid names by surrounding the name with backticks or quote marks.

Let’s work with a small file with two invalid names. First load the tidyverse library. I’ll also need the gt package for the last example.

library(gt)
library(tidyverse)

Read in the file using the read_csv function in the readr library (included when you loaded tidyverse) and display the results.

beatles_1 <- read_csv(
  file = "http://pmean.com/new-images/25/invalid-names.csv",
  col_names = TRUE,
  col_types = "cc")
beatles_1
# A tibble: 4 × 2
  `last-name` `1st-name`
  <chr>       <chr>     
1 Harrison    George    
2 Lennon      John      
3 McCartney   Paul      
4 Starkey     Richard

The results display just fine, but if you try to work with individual variables, it all falls apart.

beatles_1$last-name
Warning: Unknown or uninitialised column: `last`.
Error: object 'name' not found

You can work with this variable by surrounding it with backticks

beatles_1$`last-name`
[1] "Harrison"  "Lennon"    "McCartney" "Starkey"

or with quotes.

beatles_1[ ,"last-name"]
# A tibble: 4 × 1
  `last-name`
  <chr>      
1 Harrison   
2 Lennon     
3 McCartney  
4 Starkey

Sometimes the backticks and the quote marks are interchangeable

beatles_1$"last-name"
[1] "Harrison"  "Lennon"    "McCartney" "Starkey"

and sometimes they are not. With a bit of trial and error, you can figure out what works and what doesn’t.

beatles_1[ ,`last-name`]
Error: object 'last-name' not found

The rename function

Always having to surround your names with backticks or quotes get very tedious very quickly. It’s a good idea to change the names right away. The rename function in dplyr is an easy way to do this.

beatles_1 |>
  rename(
    last_name = `last-name`,
    first_name = `1st-name`) -> beatles_2

beatles_2
# A tibble: 4 × 2
  last_name first_name
  <chr>     <chr>     
1 Harrison  George    
2 Lennon    John      
3 McCartney Paul      
4 Starkey   Richard

You could also use the names function, part of base R.

names(beatles_1) <- c("last_name", "first_name")
beatles_1
# A tibble: 4 × 2
  last_name first_name
  <chr>     <chr>     
1 Harrison  George    
2 Lennon    John      
3 McCartney Paul      
4 Starkey   Richard

Once you have converted to valid names, you don’t have to use backticks and quote marks.

The name_repair argument

Most (but not all) of the functions in the readr library have a name_repair argument. The default value for the name_repair argument is “unique”. This will change the name of a variable to avoid the unpleasant prospect of having two variables with the same name, but it won’t fix names that fail to follow the rules of R.

You can, however, use the “universal” value for the name_repair. This will modify any variables to insure unique names, but will also slightly modify the names to avoid any conflicts with

beatles_3 <- read_csv(
  file = "invalid-names.csv",
  col_names = TRUE,
  col_types = "cc",
  name_repair = "universal")
Error: 'invalid-names.csv' does not exist in current working directory ('C:/Users/steve/git/qblog4/posts/invalid-names').
beatles_3
Error: object 'beatles_3' not found

The universal repair places dots strategically to make the variable name valid. The results may look a bit ugly at times, but that’s a small price to pay.

Replace with generic names

You can tell R to ignore the names given in the first line. If you do this, R will create generic names. Remember to skip to the second line of the file, or R will use the variable names as the first row of the data frame or tibble.

beatles_4 <- read_csv(
  file = "invalid-names.csv",
  col_names = FALSE,
  skip = 1,
  col_types = "cc")
Error: 'invalid-names.csv' does not exist in current working directory ('C:/Users/steve/git/qblog4/posts/invalid-names').
beatles_4
Error: object 'beatles_4' not found

I generally discourage the use of generic names like X1 and X2. But it does work.

Introducing invalid names during output

There are a few times when you might want to deliberately create invalid names. One reason is that you are outputtng a dataset using a fancy format and want the headings to look nice.

You should look at the gt library, which does an awesome job of creating nice looking tables. Here’s some sample code.

beatles_1 |>
  gt() |> 
  tab_header(
    title = "The Fab Four",
    subtitle = "In alphabetical order")
The Fab Four
In alphabetical order
last_name first_name
Harrison George
Lennon John
McCartney Paul
Starkey Richard

The table looks nice, except for the names in each column. This table would be greatly improved using mixed capitalization and spaces. You can do this with the rename function and backticks.

beatles_1 |>
  rename(
    `Last Name` = last_name,
    `First Name` = first_name) |>
  gt() |> 
  tab_header(
    title = "The Fab Four",
    subtitle = "In alphabetical order")
The Fab Four
In alphabetical order
Last Name First Name
Harrison George
Lennon John
McCartney Paul
Starkey Richard

Now doesn’t that look so much nicer!

Summary

You might encounter invalid names for variables in a data frame or tibble when you import data. You can accommodate those invalid names using backticks and/or quote marks. It is usually better to convert to valid names. You can do this several ways.

You may sometimes want to create invalid names intentionally to make nicer looking tables. Use the rename function and backticks to accomplish this.