Recoding Multiple Likert Scales-- Any packages?

Question

At the moment, I'm recoding my Likert type scales into numeric values. However, I have many items from a number of different scales in this dataset. For example: Q2.1_1: Q2.1_16 is one scale with a Likert scale that has a different key than the other surveys. At the moment, I'm manually entering each recode like this:

final$Q2.1_1rc <- as.numeric(recode(
    final$Q2.1_1,
    "Very slightly or not at all" = 1,
    "A little"                    = 2,
    "Moderately"                  = 3,
    "Quite a bit"                 = 4,
    "Extremely"                   = 5
))

I then C/P and continue to change the name of variable.. However, I have a large dataset so doing this manually will be cumbersome. Can anyone help me out with a shorter way to code this? Are there packages that help with this? Perhaps a function?

Thanks!

markdly markdly · Accepted Answer · 2017-09-30T23:39:44

One way you could recode the multiple variables at once is to use the mutate_at function from the dplyr package.

Example data

library(dplyr)
set.seed(123)
resp <- c("Very slightly or not at all", "A little", "Moderately", "Quite a bit", "Extremely")
final <- tibble(Q2.1_1 =  sample(resp, 6, replace = TRUE),
                Q2.2_1 =  sample(resp, 6, replace = TRUE))

Solution
Assuming the variables you want to start with in final all have "Q2" at the start of their variable name, you could do this:

final %>% 
  mutate_at(vars(starts_with("Q2")), 
            funs("rc" = recode(.,
                               "Very slightly or not at all" = 1,
                               "A little"                    = 2,
                               "Moderately"                  = 3,
                               "Quite a bit"                 = 4,
                               "Extremely"                   = 5)))

#> # A tibble: 6 x 4
#>                        Q2.1_1     Q2.2_1 Q2.1_1_rc Q2.2_1_rc
#>                         <chr>      <chr>     <dbl>     <dbl>
#> 1                    A little Moderately         2         3
#> 2                 Quite a bit  Extremely         4         5
#> 3                  Moderately Moderately         3         3
#> 4                   Extremely Moderately         5         3
#> 5                   Extremely  Extremely         5         5
#> 6 Very slightly or not at all Moderately         1         3

As stated in the official documentation, the best place to get started with dplyr is the data import chapter of the R for Data Science book while specific examples related to the use of mutate_at can be found here.

Recoding Multiple Likert Scales-- Any packages?

1 Answers