PNTA unanswered to miss tutorial

Purpose

This vignette focuses on the to_binary() and pnta_unaswered_to_miss() functions. The purpose of this section is to help users identify when to use pnta_unaswered_to_miss(), how to get the data in a format the function can use, and finally, how to apply the function.

When to use

We use pnta_unanswered_to_miss() when working with a select all that apply question that has a column for prefer not to answer. This is most commonly a column filled with “Yes” and “No”, “0” and “1”, or “Prefer not to answer” and “NA”. The issue arises when someone marks the prefer not to answer option along with at least one other of the options of the select all that apply question. When someone selects the prefer not to answer option, we want to mark all other cells of that row in the columns relevant to the prefer not to answer question as NA.

Lets take a look at what this may look like

data <- bns2_pkg_data %>% select(q14_1:q14_10) %>% mutate(pnta = bns2_pkg_data$q14_30)
data[8:12, 7:10]
#> # A tibble: 5 × 4
#>   q14_8 q14_9 q14_10  pnta
#>   <chr> <chr> <chr>  <dbl>
#> 1 No    Yes   Yes        1
#> 2 No    No    Yes        0
#> 3 No    Yes   Yes        1
#> 4 No    No    Yes        0
#> 5 No    Yes   Yes        0

As we can see from the above snippet there are “Yes” and “No” responses in the same row someone marked prefer not to answer (pnta == 1). Before we can change these rows to NA we want the data to be in a 1/0 format because the pnta_unanswered_to_miss() function is not compatible with non-binary data.

Changing the columns from yes/no to 1/0 

df_converted <- to_binary(data, these.cols = "q14_", prefix = TRUE, yesno = TRUE)
# View the converted dataframe side-by-side
old <- bns2_pkg_data |> dplyr::select(q14_1, q14_4)
new <- df_converted |> dplyr::select(q14_1, q14_4)
cbind(old, new) %>% head(10)
#>    q14_1 q14_4 q14_1 q14_4
#> 1    Yes   Yes     1     1
#> 2     No   Yes     0     1
#> 3    Yes   Yes     1     1
#> 4    Yes   Yes     1     1
#> 5     No   Yes     0     1
#> 6     No   Yes     0     1
#> 7   <NA>  <NA>    NA    NA
#> 8     No   Yes     0     1
#> 9     No   Yes     0     1
#> 10    No   Yes     0     1

Applying the pnta function

Each column of our mark all that apply question is now 1/0, which allows us to use pnta.unanswered.to.miss() to set all cells of a row in relevant columns of the select all that apply question to NA when the “Prefer not to answer” option was selected.

df_unanswered_to_miss <- pnta.unanswered.to.miss(data = df_converted,
                                  these.col = "q14_",
                                  pnta = df_converted$pnta,
                                  prefix = TRUE)

new <- df_unanswered_to_miss %>%  select(q14_3:q14_5, pnta)
new %>% head(10)
#> # A tibble: 10 × 4
#>    q14_3 q14_4 q14_5  pnta
#>    <dbl> <dbl> <dbl> <dbl>
#>  1     0     1     1     0
#>  2     0     1     0     0
#>  3     0     1     0     0
#>  4     0     1     0     0
#>  5     0     1     1     0
#>  6     0     1     0    NA
#>  7    NA    NA    NA     1
#>  8    NA    NA    NA     1
#>  9     0     1     0     0
#> 10    NA    NA    NA     1

Success! All rows where pnta == 1 are NA.