Skip to contents

Split a packed multi-select column into one column per answer

Source: R/recode.R
split_multi.Rd

Turns the single cell a spreadsheet export gives a "tick all that apply" question ("Speed; Drivers; Betting") into the one-column-per-option layout the rest of the package expects, so calc_percentage_multi(), crosstab() and the plots all work on it.

Usage

split_multi(data = NULL, column, split = "auto", prefix = NULL)

Arguments

data

A data frame. If omitted, the session default (use_dataset()) is used.

column

The packed column (unquoted).

split

The delimiter between answers inside a cell. "auto" (default) detects ";", "|" or ",", in that order. Pass a string to force one.

prefix

Prefix for the new column names. Defaults to the column's own name followed by an underscore, e.g. motivations_.

Value

The data frame with one new column per distinct answer, each holding the answer text for respondents who chose it and "" for those who did not, ordered most-chosen first. The packed column is kept.

Details

Google Forms, Google Sheets and most survey platforms export a multi-select question as one cell per respondent holding every answer they ticked, joined by a delimiter. That shape cannot be counted directly, because a respondent who ticked three options is one row, not three. This function unpacks it into the indicator layout the package's own datasets use.

The new columns are named prefix plus the answer text verbatim, which keeps the labels readable all the way through to a chart: calc_percentage_multi() strips the prefix back off and uses what remains as the option label. Answer text is rarely a syntactic R name, so refer to the new columns with backticks if you need them individually. Surrounding whitespace is trimmed and empty answers are dropped, so "Speed; ; Drivers" yields two options. When no cell contains the delimiter, each cell is treated as a single answer, which is the right answer for a multi-select question everyone happened to answer once.

If you only want the percentages, skip this step: calc_percentage_multi() detects a packed column and splits it for you.

See also

calc_percentage_multi(), na_blank().

Other recode: add_region(), bin_numeric(), clean_label(), drop_items(), ensure_numeric(), generation_scheme(), na_blank(), nps_group(), recode_age(), recode_generation(), recode_likert(), recode_region()

Examples

packed <- data.frame(
  respondent = 1:4,
  motivations = c("Speed; Drivers", "Speed", "", "Betting; Speed")
)
split_multi(packed, motivations)
#> # A tibble: 4 × 5
#>   respondent motivations      motivations_Speed motivations_Betting
#>        <int> <chr>            <chr>             <chr>              
#> 1          1 "Speed; Drivers" "Speed"           ""                 
#> 2          2 "Speed"          "Speed"           ""                 
#> 3          3 ""               ""                ""                 
#> 4          4 "Betting; Speed" "Speed"           "Betting"          
#> # ℹ 1 more variable: motivations_Drivers <chr>

split_multi(packed, motivations) %>%
  calc_percentage_multi("motivations_", id = respondent, sort = "desc")
#> # A tibble: 3 × 3
#>   option      n   pct
#>   <fct>   <int> <dbl>
#> 1 Speed       3   100
#> 2 Betting     1    33
#> 3 Drivers     1    33