library(YouAnalyser)
#>
#> ── Welcome to YouAnalyser! ─────────────────────────────────────────────────────
#> ✔ Package loaded successfully!
#> Type `?YouAnalyser` to see the documentation.
#> Visit the package's website for more information:
#> <https://eguizarrosales.github.io/YouAnalyser/>
library(haven)Labelled Survey Data
In a best case scenario, survey data is collected in a format that
preserves the metadata of the survey, such as variable labels and value
labels. This is often the case when data is collected using YouGov’s own
Gryphon or other software solutions like
Quantilope, Qualtrics or
SurveyMonkey, which can export data in formats that
retain this information, most commonly SPSS .sav files.
In this case, not only the names of the variables (e.g. “F600”) and the values (e.g. “1”, “2”, “3”) are preserved, but also the labels that describe what these variables and values represent:
- Variable labels: These provide a human-readable description of what each variable represents. For example, a variable named “F600” might have a label like “Wie attraktiv finden Sie die BKW als Arbeitgeberin?”, which makes it clear what the question was participants answered.
- Value labels: These provide a human-readable description of what each value represents. For example, a value of “1” for the variable “F600” might have a label like “1 - Überhaupt nicht attraktiv”, which makes it clear what the response option was for participants.
When you import such a labelled dataset into R using the
haven::read_sav() function, you can easily access both the
variable labels and value labels. The YouAnalyser package
provides functions to work with this metadata, allowing you to quickly
understand and analyze your survey data without having to manually
decode the variable and value names.
# Provide a file path to your labelled survey data in SPSS .sav format.
# Here we accesss the example data shipped with the `YouAnalyser` package.
file_path <- system.file(
"extdata",
"bkw.sav",
package = "YouAnalyser",
mustWork = FALSE
)
# Read in the data using haven::read_sav() and filter/select relevant variables
myData <- haven::read_sav(file = file_path) |>
# Filter out missing values for the dependent variable F600
dplyr::filter(!is.na(F600)) |>
# Select only the dependent variable F600 and the independent variables F800
dplyr::select("F600", dplyr::starts_with("F800_"))
# Inspect the structure of the data for the first two variables. Take note of
# the attributes "@label" (variable label) and "@labels" (value labels).
str(myData[, 1:2])
#> tibble [1,171 × 2] (S3: tbl_df/tbl/data.frame)
#> $ F600 : dbl+lbl [1:1171] 5, 3, 6, 2, 4, 5, 5, 5, 5, 5, 5, 4, 3, 4, 4, 5, 4, 4,...
#> ..@ label : chr "Wie attraktiv finden Sie die BKW als Arbeitgeberin?"
#> ..@ format.spss: chr "F8.2"
#> ..@ labels : Named num [1:7] 1 2 3 4 5 6 7
#> .. ..- attr(*, "names")= chr [1:7] "1 - Überhaupt nicht attraktiv" "2" "3" "4" ...
#> $ F800_1: dbl+lbl [1:1171] 6, 6, 7, 5, 4, 6, 4, 5, 7, 6, 7, 7, 3, 4, 6, 7, 5, 4,...
#> ..@ label : chr "Sicherheit und langfristige Stabilität des Arbeitgebers"
#> ..@ format.spss: chr "F8.2"
#> ..@ labels : Named num [1:7] 1 2 3 4 5 6 7
#> .. ..- attr(*, "names")= chr [1:7] "Überhaupt nicht gut 1" "2" "3" "4" ...There are different ways you can inspect the variable and value labels in R, e.g.:
# Access variable label
attr(myData$F600, "label")
#> [1] "Wie attraktiv finden Sie die BKW als Arbeitgeberin?"
# Access value labels
attr(myData$F600, "labels")
#> 1 - Überhaupt nicht attraktiv 2
#> 1 2
#> 3 4
#> 3 4
#> 5 6
#> 5 6
#> 7 - Sehr attraktiv
#> 7
# Prettier with the haven::print_labels() function
haven::print_labels(myData$F600)
#>
#> Labels:
#> value label
#> 1 1 - Überhaupt nicht attraktiv
#> 2 2
#> 3 3
#> 4 4
#> 5 5
#> 6 6
#> 7 7 - Sehr attraktivYouAnalyser provides the
dp_inspect_codebook() function to print a codebook of your
labelled data to your console, which includes the variable labels and
value labels in a tidy format:
# Print codebook for the first two variables
dp_inspect_codebook(myData[, 1:2])
#> data (1171 rows and 2 variables, 2 shown)
#>
#> ID | Name | Label | Type | Missings | Values | Value Labels | N
#> ---+--------+----------------+---------+----------+--------+--------------------+------------
#> 1 | F600 | Wie attraktiv | numeric | 0 (0.0%) | 1 | 1 - Überhaupt... | 43 ( 3.7%)
#> | | finden Sie die | | | 2 | 2 | 50 ( 4.3%)
#> | | BKW als | | | 3 | 3 | 130 (11.1%)
#> | | Arbeitgeberin? | | | 4 | 4 | 461 (39.4%)
#> | | | | | 5 | 5 | 303 (25.9%)
#> | | | | | 6 | 6 | 109 ( 9.3%)
#> | | | | | 7 | 7 - Sehr... | 75 ( 6.4%)
#> ---+--------+----------------+---------+----------+--------+--------------------+------------
#> 2 | F800_1 | Sicherheit und | numeric | 0 (0.0%) | 1 | Überhaupt nicht... | 11 ( 0.9%)
#> | | langfristige | | | 2 | 2 | 19 ( 1.6%)
#> | | Stabilität des | | | 3 | 3 | 66 ( 5.6%)
#> | | Arbeitgebers | | | 4 | 4 | 301 (25.7%)
#> | | | | | 5 | 5 | 270 (23.1%)
#> | | | | | 6 | 6 | 322 (27.5%)
#> | | | | | 7 | Sehr gut 7 | 182 (15.5%)
#> ---------------------------------------------------------------------------------------------Convert Unlabelled to Labelled Data
Sometimes, your data comes in an unlabelled format, for example if
you receive a .csv or a .xlsx file. Then, this
data looks like this:
# Provide a file path to your unlabelled survey data in Excel .xlsx format.
# Here we accesss the example data shipped with the `YouAnalyser` package.
file_path <- system.file(
"extdata",
"bkw_unlabelled.xlsx",
package = "YouAnalyser",
mustWork = TRUE
)
# Read in the unlabelled data using openxlsx::read.xlsx() and convert it to a tibble
myData_unlabelled <- openxlsx::read.xlsx(file_path) |>
tibble::as_tibble()
# Inspect the structure of the data for the first two variables.
# Note that there are no variable labels or value labels.
str(myData_unlabelled[, 1:2])
#> tibble [1,171 × 2] (S3: tbl_df/tbl/data.frame)
#> $ F600 : num [1:1171] 5 3 6 2 4 5 5 5 5 5 ...
#> ..- attr(*, "label")= chr "Wie attraktiv finden Sie die BKW als Arbeitgeberin?"
#> ..- attr(*, "format.spss")= chr "F8.2"
#> $ F800_1: num [1:1171] 6 6 7 5 4 6 4 5 7 6 ...
#> ..- attr(*, "label")= chr "Sicherheit und langfristige Stabilität des Arbeitgebers"
#> ..- attr(*, "format.spss")= chr "F8.2"
# `dp_inspect_codebook()` also confirms that there are no variable labels or
# value labels in the unlabelled data.
dp_inspect_codebook(myData_unlabelled)
#> data (1171 rows and 15 variables, 15 shown)
#>
#> ID | Name | Label | Type | Missings | Values | N
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 1 | F600 | Wie attraktiv | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | finden Sie die | | | |
#> | | BKW als | | | |
#> | | Arbeitgeberin? | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 2 | F800_1 | Sicherheit und | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | langfristige | | | |
#> | | Stabilität des | | | |
#> | | Arbeitgebers | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 3 | F800_2 | Karriere- und | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | Entwicklungsmöglichkeiten | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 4 | F800_3 | Sinnvolle | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | Tätigkeit und | | | |
#> | | gesellschaftlicher | | | |
#> | | Beitrag | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 5 | F800_4 | Gute | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | Zusammenarbeit | | | |
#> | | und Teamkultur | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 6 | F800_5 | Vereinbarkeit | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | von Beruf und | | | |
#> | | Privatleben | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 7 | F800_6 | Moderne | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | Arbeitsumgebung | | | |
#> | | und | | | |
#> | | Technologien | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 8 | F800_7 | Attraktive | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | Vergütung und | | | |
#> | | Zusatzleistungen | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 9 | F800_8 | Gute Führung | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | und | | | |
#> | | wertschätzende | | | |
#> | | Unternehmenskultur | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 10 | F800_9 | Verantwortung | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | und | | | |
#> | | Gestaltungsspielraum | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 11 | F800_10 | Zukunftsorientierung | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | und | | | |
#> | | Nachhaltigkeit | | | |
#> | | des | | | |
#> | | Unternehmens | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 12 | F800_11 | Internationales | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | Arbeitsumfeld | | | |
#> | | und kulturelle | | | |
#> | | Vielfalt | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 13 | F800_12 | Innovationskultur | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | und | | | |
#> | | Veränderungsbereitschaft | | | |
#> | | des | | | |
#> | | Unternehmens | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 14 | F800_13 | Arbeitszeit- | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | und | | | |
#> | | Arbeitsortflexibilität | | | |
#> ---+---------+---------------------------+---------+----------+--------+-----
#> 15 | F800_14 | Diversität, | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | Gleichstellung | | | |
#> | | und Inklusion | | | |
#> -----------------------------------------------------------------------------
# The missing variable and value labels are also confirmed by the haven::is.labelled()
# function, which returns FALSE for all variables in the unlabelled data.
haven::is.labelled(myData_unlabelled)
#> [1] FALSEIn this case, you can use the dp_convert_to_labelled()
function to convert your unlabelled data into a labelled format, which
will allow you to work with variable and value labels in R and in
YouAnalyser.
To do this, you need to provide the data as well as a
codebook that contains the variable labels and value
labels for your data. The codebook should be in a tidy format, with one
row per variable and columns for the variable name, variable label, and
value labels. This codebook can be created manually using a spreadsheet
software like Excel. Use the following code to save a copy of the
codebook template provided with YouAnalyser:
# Choose interactively where the template should be copied to
file_path <- ya_choose_file_path("my_codebook.xlsx")
# Copy the template
dp_copy_codebook_template(file_path)Keep the first row unchanged, as it contains the required column names. Then, fill in the codebook with the variable names, variable labels and value labels for your data. Import the codebook into R and inspect it to make sure it looks correct.
# Choose the codebook file you just edited
codebook_file_path <- ya_choose_file()
# Read in the codebook
myCodebook <- openxlsx::read.xlsx(codebook_file_path) |>
tibble::as_tibble()
# Display the first few rows of the codebook
head(myCodebook, n = 3 * 7) |>
print(n = Inf)
# Display the first few rows of the codebook
head(myCodebook, n = 3 * 7) |>
print(n = Inf)
#> # A tibble: 21 × 4
#> variable variable_label value value_label
#> <chr> <chr> <dbl> <chr>
#> 1 F600 Wie attraktiv finden Sie die BKW als Arbeitgeberi… 1 1 - Überha…
#> 2 F600 Wie attraktiv finden Sie die BKW als Arbeitgeberi… 2 2
#> 3 F600 Wie attraktiv finden Sie die BKW als Arbeitgeberi… 3 3
#> 4 F600 Wie attraktiv finden Sie die BKW als Arbeitgeberi… 4 4
#> 5 F600 Wie attraktiv finden Sie die BKW als Arbeitgeberi… 5 5
#> 6 F600 Wie attraktiv finden Sie die BKW als Arbeitgeberi… 6 6
#> 7 F600 Wie attraktiv finden Sie die BKW als Arbeitgeberi… 7 7 - Sehr a…
#> 8 F800_1 Sicherheit und langfristige Stabilität des Arbeit… 1 Überhaupt …
#> 9 F800_1 Sicherheit und langfristige Stabilität des Arbeit… 2 2
#> 10 F800_1 Sicherheit und langfristige Stabilität des Arbeit… 3 3
#> 11 F800_1 Sicherheit und langfristige Stabilität des Arbeit… 4 4
#> 12 F800_1 Sicherheit und langfristige Stabilität des Arbeit… 5 5
#> 13 F800_1 Sicherheit und langfristige Stabilität des Arbeit… 6 6
#> 14 F800_1 Sicherheit und langfristige Stabilität des Arbeit… 7 Sehr gut 7
#> 15 F800_2 Karriere- und Entwicklungsmöglichkeiten 1 Überhaupt …
#> 16 F800_2 Karriere- und Entwicklungsmöglichkeiten 2 2
#> 17 F800_2 Karriere- und Entwicklungsmöglichkeiten 3 3
#> 18 F800_2 Karriere- und Entwicklungsmöglichkeiten 4 4
#> 19 F800_2 Karriere- und Entwicklungsmöglichkeiten 5 5
#> 20 F800_2 Karriere- und Entwicklungsmöglichkeiten 6 6
#> 21 F800_2 Karriere- und Entwicklungsmöglichkeiten 7 Sehr gut 7Once your codebook is ready, you can use the
dp_convert_to_labelled() function to convert your
unlabelled data into a labelled format:
# Convert unlabelled data to labelled format using the codebook
myData_labelled <- dp_convert_to_labelled(
data = myData_unlabelled,
codebook = myCodebook
)
# Confirm that labelling worked (only show the first 3 variables for brevity)
dp_inspect_codebook(myData_unlabelled[, 1:3])
#> data (1171 rows and 3 variables, 3 shown)
#>
#> ID | Name | Label | Type | Missings | Values | N
#> ---+--------+---------------------------+---------+----------+--------+-----
#> 1 | F600 | Wie attraktiv | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | finden Sie die | | | |
#> | | BKW als | | | |
#> | | Arbeitgeberin? | | | |
#> ---+--------+---------------------------+---------+----------+--------+-----
#> 2 | F800_1 | Sicherheit und | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | langfristige | | | |
#> | | Stabilität des | | | |
#> | | Arbeitgebers | | | |
#> ---+--------+---------------------------+---------+----------+--------+-----
#> 3 | F800_2 | Karriere- und | numeric | 0 (0.0%) | [1, 7] | 1171
#> | | Entwicklungsmöglichkeiten | | | |
#> ----------------------------------------------------------------------------
dp_inspect_codebook(myData_labelled[, 1:3])
#> data (1171 rows and 3 variables, 3 shown)
#>
#> ID | Name | Label | Type | Missings | Values
#> ---+--------+---------------------------+---------+----------+-------
#> 1 | F600 | Wie attraktiv | numeric | 0 (0.0%) | 1
#> | | finden Sie die | | | 2
#> | | BKW als | | | 3
#> | | Arbeitgeberin? | | | 4
#> | | | | | 5
#> | | | | | 6
#> | | | | | 7
#> ---+--------+---------------------------+---------+----------+-------
#> 2 | F800_1 | Sicherheit und | numeric | 0 (0.0%) | 1
#> | | langfristige | | | 2
#> | | Stabilität des | | | 3
#> | | Arbeitgebers | | | 4
#> | | | | | 5
#> | | | | | 6
#> | | | | | 7
#> ---+--------+---------------------------+---------+----------+-------
#> 3 | F800_2 | Karriere- und | numeric | 0 (0.0%) | 1
#> | | Entwicklungsmöglichkeiten | | | 2
#> | | | | | 3
#> | | | | | 4
#> | | | | | 5
#> | | | | | 6
#> | | | | | 7
#> ---------------------------------------------------------------------
#>
#> ID | Value Labels | N
#> ---+--------------------+------------
#> 1 | 1 - Überhaupt... | 43 ( 3.7%)
#> | 2 | 50 ( 4.3%)
#> | 3 | 130 (11.1%)
#> | 4 | 461 (39.4%)
#> | 5 | 303 (25.9%)
#> | 6 | 109 ( 9.3%)
#> | 7 - Sehr... | 75 ( 6.4%)
#> ---+--------------------+------------
#> 2 | Überhaupt nicht... | 11 ( 0.9%)
#> | 2 | 19 ( 1.6%)
#> | 3 | 66 ( 5.6%)
#> | 4 | 301 (25.7%)
#> | 5 | 270 (23.1%)
#> | 6 | 322 (27.5%)
#> | Sehr gut 7 | 182 (15.5%)
#> ---+--------------------+------------
#> 3 | Überhaupt nicht... | 11 ( 0.9%)
#> | 2 | 34 ( 2.9%)
#> | 3 | 70 ( 6.0%)
#> | 4 | 417 (35.6%)
#> | 5 | 316 (27.0%)
#> | 6 | 234 (20.0%)
#> | Sehr gut 7 | 89 ( 7.6%)
#> -------------------------------------Summary
If your data is unlabelled (.csv, .xlsx),
use the following code snippets to convert it to a labelled format:
# Read in your .xlsx data
file_path <- ya_choose_file()
myData_unlabelled <- openxlsx::read.xlsx(file_path) |>
tibble::as_tibble()
# Create a codebook for your data using the provided template
file_path <- ya_choose_file_path("my_codebook.xlsx")
dp_copy_codebook_template(file_path)
#### EDIT THE CODEBOOK TEMPLATE IN EXCEL ####
# After having edited the codebook template in Excel, you can read it back into R
file_path <- ya_choose_file()
myCodebook <- openxlsx::read.xlsx(file_path) |>
tibble::as_tibble()
# Use the unlabelled data and the codebook to convert the unlabelled data into a labelled format
myData_labelled <- dp_convert_to_labelled(
data = myData_unlabelled,
codebook = myCodebook
)
# Confirm that the data is now labelled
dp_inspect_codebook(myData_labelled)
# Save the labelled data as a .sav file to preserve the variable and value labels for future use
file_path <- ya_choose_file_path("my_labelled_data.sav")
haven::write_sav(myData_labelled, file_path)