Recode multiple categorical variables to new variables

[Polly]

I have a dataset with 11 variables describing ‘reasons for using e-cigarettes’ (ecig2crav, ecig2quit, ecig2symp, smokefree, exterior, bothering, rednoquit, red2quit, toxic5, cheaper5, cantstop), all are factor variables with 4 levels:

1=Not at all true,

2=Not very true,

3=Somewhat true,

4=Very true.

I want a function to create a new (labelled) factor variable which collapses these four categories into two: Not true (levels 1 and 2) and True (levels 3 and 4). How can I do this in a quick way in R? I want to create a new factor variable for each of the original variables. At the moment I have the code for the first two variables as

data %>%
mutate(ecig2crav_rec=recode(ecig2crav, "Not at all true"="Not true", "Not very true"="Not true", "Somewhat true"="True", "Very true"="True", .default = NA_character_),
ecig2quit_rec=recode(ecig2quit, "Not at all true"="Not true", "Not very true"="Not true", "Somewhat true"="True", "Very true"="True", .default = NA_character_))

I wonder if there is a way to avoid to write the recoding rules for each variable as they all share the same rules. Thanks.

Polly

[andresrcs]

You can use mutate_at() or the new across() to recode several columns at the same time.

dplyr 1.0.0: working across columns

If you need more specific help, please provide a proper REPRoducible EXample (reprex) illustrating your issue.

[Polly]

Thank you Andresrcs for your reply. I tried mutate_at(), but it does not generate new variables. Instead, it overwrites the original variables. But I want new variables rather than overwriting the existing variables.

Polly

[andresrcs]

@Polly

I tried mutate_at(), but it does not generate new variables.

You can also generate new variables, simply pass your function as a named list. As I said, if you need more specific help, please provide a proper reprex.

[Polly]

Here is some code to generate some sample data.

mydata <- data.frame(ecig2crav = c("Not at all true", "Not very true", "Somewhat true", "Very true"), ecig2quit = c("Not very true", "Somewhat true", "Very true", NA), ecig2symp = c("Somewhat true", "Very true", "Not at all true", "Not very true"))

mydata

## ecig2crav ecig2quit ecig2symp
## 1 Not at all true Not very true Somewhat true
## 2 Not very true Somewhat true Very true
## 3 Somewhat true Very true Not at all true
## 4 Very true <NA> Not very true

Now I want to recode each of the three variables with the same rules ("Not at all true" = "Not true"; "Not very true" = "Not true"; "Somewhat true" = "True"; "Very true" ="True", NA=NA) to NEW variables. My previous approach was as below:

mydata %>%
mutate(ecig2crav_rec=recode(ecig2crav, "Not at all true"="Not true", "Not very true"="Not true", "Somewhat true"="True", "Very true"="True", .default = NA_character_),
ecig2quit_rec=recode(ecig2quit, "Not at all true"="Not true", "Not very true"="Not true", "Somewhat true"="True", "Very true"="True", .default = NA_character_), 
ecig2symp_rec=recode(ecig2symp, "Not at all true"="Not true", "Not very true"="Not true", "Somewhat true"="True", "Very true"="True", .default = NA_character_)). 

In this approach, I am repeating the rules for each variable which I would like to avoid. Thanks.

Polly

[andresrcs]

As I said, use a named list

library(dplyr)

mydata <-
    data.frame(
        ecig2crav = c("Not at all true", "Not very true", "Somewhat true", "Very true"),
        ecig2quit = c("Not very true", "Somewhat true", "Very true", NA),
        ecig2symp = c("Somewhat true", "Very true", "Not at all true", "Not very true")
    )

mydata %>% 
    mutate_all(list(rec = ~ recode(.,
                                 "Not at all true"="Not true",
                                 "Not very true"="Not true",
                                 "Somewhat true"="True",
                                 "Very true"="True",
                                 .default = NA_character_)
                    )
               )
#>         ecig2crav     ecig2quit       ecig2symp ecig2crav_rec ecig2quit_rec
#> 1 Not at all true Not very true   Somewhat true      Not true      Not true
#> 2   Not very true Somewhat true       Very true      Not true          True
#> 3   Somewhat true     Very true Not at all true          True          True
#> 4       Very true          <NA>   Not very true          True          <NA>
#>   ecig2symp_rec
#> 1          True
#> 2          True
#> 3      Not true
#> 4      Not true

^{Created on 2020-06-01 by the reprex package (v0.3.0)}

[Polly]

Hi Andresrcs,

I copied your code to my Rstudio and got an error message as below. Am I doing anything wrong? Please bare with me, I am fairly new to R. Do I need to create a list before using the mutate_all? What if I only want to recode the first two variables? Thanks.

Error in recode(., Not at all true = "Not true", Not very true = "Not true", : unused arguments (Not at all true = "Not true", Not very true = "Not true", Somewhat true = "True", Very true = "True", .default = NA)

[Polly]

I figured out the error. Maybe my dplyr package was out of date. I reinstalled the package and it works now. Thank you very much. You have been very helpful!

Polly

[andresrcs]

If your question's been answered (even by you!), would you mind choosing a solution? It helps other people see which questions still need help, or find solutions if they have similar problems. Here’s how to do it:

FAQ: How do I mark a solution?

If your question has been answered, don't forget to mark the solution! How do I mark a solution? Find the reply you want to mark as the solution and look for the row of small gray icons at the bottom of that reply. Click the one that looks like a box with a checkmark in it: [image] Hovering over the mark solution button shows the label, "Select if this reply solves the problem". If you don't see the mark solution button, try clicking the three dots button ( ••• ) to expand the full set of options. When a solution is chosen, the icon turns green and the hover label changes to: "Unselect if this reply no longer solves the problem". Success! [solution_reply_author] …

[Polly]

Can I add one more question to this. After the recoding is done, how can I crosstabulate multiple tables at the same time to check my recoding has been done correctly. At the moment, I have to write one line of code for each pair of variables which I don't think is very efficient, particularly I have 11 pairs of variables that I would like to check. Is there a faster way to do this? Thanks.

with(data, table(ecig2crav, ecig2crav_rec))
with(data, table(ecig2quit, ecig2quit_rec))
with(data, table(ecig2symp, ecig2symp_rec))

[phiggins]

Hi Polly,

This is a nice use case for the check_recode() function from the {finalfit} package.

Details on this package can be found here

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(finalfit)

mydata <-
  data.frame(
    ecig2crav = c("Not at all true", "Not very true", "Somewhat true", "Very true"),
    ecig2quit = c("Not very true", "Somewhat true", "Very true", NA),
    ecig2symp = c("Somewhat true", "Very true", "Not at all true", "Not very true")
  )

mydata %>% 
  mutate_all(list(rec = ~ recode(.,
                                 "Not at all true"="Not true",
                                 "Not very true"="Not true",
                                 "Somewhat true"="True",
                                 "Very true"="True",
                                 .default = NA_character_)
  )
  )->
mydata2

mydata2 %>% 
  check_recode()
#> $index
#> # A tibble: 3 x 2
#>   var1      var2         
#>   <chr>     <chr>        
#> 1 ecig2crav ecig2crav_rec
#> 2 ecig2quit ecig2quit_rec
#> 3 ecig2symp ecig2symp_rec
#> 
#> $counts
#> $counts[[1]]
#>         ecig2crav ecig2crav_rec n
#> 1 Not at all true      Not true 1
#> 2   Not very true      Not true 1
#> 3   Somewhat true          True 1
#> 4       Very true          True 1
#> 
#> $counts[[2]]
#>       ecig2quit ecig2quit_rec n
#> 1 Not very true      Not true 1
#> 2 Somewhat true          True 1
#> 3     Very true          True 1
#> 4          <NA>          <NA> 1
#> 
#> $counts[[3]]
#>         ecig2symp ecig2symp_rec n
#> 1 Not at all true      Not true 1
#> 2   Not very true      Not true 1
#> 3   Somewhat true          True 1
#> 4       Very true          True 1

^{Created on 2020-06-05 by the reprex package (v0.3.0)}

[Polly]

Hi Phiggins,

This worked perfectly. Thank very much!

Polly