Convert a vector of class factor or numeric with value labels (e.g.,
haven_labelled)
to a dichotomous factor vector. A dichotomous factor vector is a vector of
class factor with only two bipolar levels (e.g., "Agree" and "Disagree")
Details
make_dicho was designed to work on any vector that is of class factor,
haven_labelled,
or numeric with value labels. If the vector is numeric with no value
labels, the function will return an error. This is because the function first
converts the vector to a factor using make_factor(). Then, it removes the
first word if there are multiple words in the factor level.
The resulting factor levels default to alphabetical but if you want to
reverse them, just set flip_levels = TRUE in the function.
In addition, this function adds two new attributes. The first attribute,
transformation, indicates the data transformation that the original vector
underwent to create this new vector. The second attribute, label, contains
the variable label that was found in the original variable. However, if the
original vector did not have a variable label, then this attribute will not
show up.
Examples
# load the libraries
library(dplyr)
library(adlgraphs)
# Check a labelled variable -------------------------------------------------
# convert "top" into a dichotomous factor
new_df <- test_data %>%
mutate(top_f2 = make_dicho(top)) %>%
# keep only two relevant variables
select(top, top_f2)
# compare "top" to "top_f2"
# we can see all response with "agree" in the name are now "agree" and
# all those with "disagree" are now "disagree"
new_df
#> # A tibble: 250 × 2
#> top top_f2
#> <dbl+lbl> <fct>
#> 1 1 [Strongly agree] Agree
#> 2 2 [Somewhat agree] Agree
#> 3 2 [Somewhat agree] Agree
#> 4 3 [Somewhat disagree] Disagree
#> 5 2 [Somewhat agree] Agree
#> 6 4 [Strongly disagree] Disagree
#> 7 2 [Somewhat agree] Agree
#> 8 2 [Somewhat agree] Agree
#> 9 2 [Somewhat agree] Agree
#> 10 4 [Strongly disagree] Disagree
#> # ℹ 240 more rows
# show the attributes
attributes(new_df$top_f2)
#> $levels
#> [1] "Agree" "Disagree"
#>
#> $class
#> [1] "factor"
#>
#> $transformation
#> Converting 'top' to a dichotomous factor with 'Agree' as the reference level
#>
#> $label
#> [1] "An ideal society requires some groups to be on top and others to be on the bottom"
#>
# Check a factor variable ---------------------------------------------------
new_df <- test_data %>%
mutate(
# convert it to a factor
top_f = make_factor(top),
# convert the dichotomous factor
top_f2 = make_dicho(top_f)
) %>%
# select the three variables to compare them
select(top, top_f, top_f2)
# compare the three variables
new_df
#> # A tibble: 250 × 3
#> top top_f top_f2
#> <dbl+lbl> <fct> <fct>
#> 1 1 [Strongly agree] Strongly agree Agree
#> 2 2 [Somewhat agree] Somewhat agree Agree
#> 3 2 [Somewhat agree] Somewhat agree Agree
#> 4 3 [Somewhat disagree] Somewhat disagree Disagree
#> 5 2 [Somewhat agree] Somewhat agree Agree
#> 6 4 [Strongly disagree] Strongly disagree Disagree
#> 7 2 [Somewhat agree] Somewhat agree Agree
#> 8 2 [Somewhat agree] Somewhat agree Agree
#> 9 2 [Somewhat agree] Somewhat agree Agree
#> 10 4 [Strongly disagree] Strongly disagree Disagree
#> # ℹ 240 more rows
# show the attributes for the new variable
attributes(new_df$top_f2)
#> $levels
#> [1] "Agree" "Disagree"
#>
#> $class
#> [1] "factor"
#>
#> $transformation
#> Converting 'top_f' to a dichotomous factor with 'Agree' as the reference level
#>
#> $label
#> [1] "An ideal society requires some groups to be on top and others to be on the bottom"
#>
# Show it with flipped levels -----------------------------------------------
# let's do the same thing but let's flip the levels now
new_df <- test_data %>%
mutate(
# show it without flipping the levels
top_f2 = make_dicho(top),
# show it with the levels being flipped
top_f2_flip = make_dicho(top, flip_levels = TRUE)
) %>%
# keep only relevant variables
select(top, top_f2, top_f2_flip)
# compare them
new_df
#> # A tibble: 250 × 3
#> top top_f2 top_f2_flip
#> <dbl+lbl> <fct> <fct>
#> 1 1 [Strongly agree] Agree Agree
#> 2 2 [Somewhat agree] Agree Agree
#> 3 2 [Somewhat agree] Agree Agree
#> 4 3 [Somewhat disagree] Disagree Disagree
#> 5 2 [Somewhat agree] Agree Agree
#> 6 4 [Strongly disagree] Disagree Disagree
#> 7 2 [Somewhat agree] Agree Agree
#> 8 2 [Somewhat agree] Agree Agree
#> 9 2 [Somewhat agree] Agree Agree
#> 10 4 [Strongly disagree] Disagree Disagree
#> # ℹ 240 more rows
# They look the same but if we check the levels of the factor we can see
# that they are in different orders
attributes(new_df$top_f2)
#> $levels
#> [1] "Agree" "Disagree"
#>
#> $class
#> [1] "factor"
#>
#> $transformation
#> Converting 'top' to a dichotomous factor with 'Agree' as the reference level
#>
#> $label
#> [1] "An ideal society requires some groups to be on top and others to be on the bottom"
#>
attributes(new_df$top_f2_flip)
#> $levels
#> [1] "Disagree" "Agree"
#>
#> $class
#> [1] "factor"
#>
#> $transformation
#> Converting 'top' to a dichotomous factor and reordering the factor levels so that 'Disagree' is the reference level
#>
#> $label
#> [1] "An ideal society requires some groups to be on top and others to be on the bottom"
#>
# ----------------------------------------------------------------------------
# function also works inside dplyr::across()
# Create new columns using `across()`
new_df <- test_data %>%
dplyr::mutate(
# use this example if you don't want to flip the factor levels
dplyr::across(
c(top:deserving),
make_dicho,
.names = "{col}_f2"
),
# if you want to flip the factor levels, follow this example
dplyr::across(
c(top:deserving),
~make_dicho(., flip_levels = TRUE),
.names = "{col}_f2_flip"
)
) %>%
# select the variables with "f2" in the name
select(contains("f2"))
# show that the function worked properly by creating two new sets of variables
new_df
#> # A tibble: 250 × 9
#> edu_f2 top_f2 inferior_f2 dominate_f2 deserving_f2 top_f2_flip
#> <fct> <fct> <fct> <fct> <fct> <fct>
#> 1 No College Degree Agree Disagree Agree Agree Agree
#> 2 No College Degree Agree Disagree Agree Agree Agree
#> 3 At Least a Bachelor'… Agree Agree Agree Agree Agree
#> 4 No College Degree Disag… Disagree Agree Agree Disagree
#> 5 No College Degree Agree Disagree Agree Disagree Agree
#> 6 At Least a Bachelor'… Disag… Disagree Agree Agree Disagree
#> 7 At Least a Bachelor'… Agree Disagree Agree Agree Agree
#> 8 At Least a Bachelor'… Agree Disagree Agree Agree Agree
#> 9 No College Degree Agree Disagree Disagree Agree Agree
#> 10 No College Degree Disag… Disagree Agree Agree Disagree
#> # ℹ 240 more rows
#> # ℹ 3 more variables: inferior_f2_flip <fct>, dominate_f2_flip <fct>,
#> # deserving_f2_flip <fct>
# show the underlying structure of the entire df
str(new_df)
#> tibble [250 × 9] (S3: tbl_df/tbl/data.frame)
#> $ edu_f2 : Factor w/ 2 levels "No College Degree",..: 1 1 2 1 1 2 2 2 1 1 ...
#> ..- attr(*, "label")= chr "What is the highest level of school you have completed or the highest degree you have received?"
#> ..- attr(*, "transformation")= 'glue' chr "Recoded 'edu' as a factor and set the levels based on their order.\nThe data transformation is as follows:\nWha"| __truncated__
#> $ top_f2 : Factor w/ 2 levels "Agree","Disagree": 1 1 1 2 1 2 1 1 1 2 ...
#> ..- attr(*, "transformation")= 'glue' chr "Converting 'top' to a dichotomous factor with 'Agree' as the reference level"
#> ..- attr(*, "label")= chr "An ideal society requires some groups to be on top and others to be on the bottom"
#> $ inferior_f2 : Factor w/ 2 levels "Agree","Disagree": 2 2 1 2 2 2 2 2 2 2 ...
#> ..- attr(*, "transformation")= 'glue' chr "Converting 'inferior' to a dichotomous factor with 'Agree' as the reference level"
#> ..- attr(*, "label")= chr "Some groups of people are simply inferior to other groups"
#> $ dominate_f2 : Factor w/ 2 levels "Agree","Disagree": 1 1 1 1 1 1 1 1 2 1 ...
#> ..- attr(*, "transformation")= 'glue' chr "Converting 'dominate' to a dichotomous factor with 'Agree' as the reference level"
#> ..- attr(*, "label")= chr "No one group should dominate in society"
#> $ deserving_f2 : Factor w/ 2 levels "Agree","Disagree": 1 1 1 1 2 1 1 1 1 1 ...
#> ..- attr(*, "transformation")= 'glue' chr "Converting 'deserving' to a dichotomous factor with 'Agree' as the reference level"
#> ..- attr(*, "label")= chr "Groups at the bottom are just as deserving as groups at the top"
#> $ top_f2_flip : Factor w/ 2 levels "Disagree","Agree": 2 2 2 1 2 1 2 2 2 1 ...
#> ..- attr(*, "transformation")= 'glue' chr "Converting 'top' to a dichotomous factor and reordering the factor levels so that 'Disagree' is the reference level"
#> ..- attr(*, "label")= chr "An ideal society requires some groups to be on top and others to be on the bottom"
#> $ inferior_f2_flip : Factor w/ 2 levels "Disagree","Agree": 1 1 2 1 1 1 1 1 1 1 ...
#> ..- attr(*, "transformation")= 'glue' chr "Converting 'inferior' to a dichotomous factor and reordering the factor levels so that 'Disagree' is the reference level"
#> ..- attr(*, "label")= chr "Some groups of people are simply inferior to other groups"
#> $ dominate_f2_flip : Factor w/ 2 levels "Disagree","Agree": 2 2 2 2 2 2 2 2 1 2 ...
#> ..- attr(*, "transformation")= 'glue' chr "Converting 'dominate' to a dichotomous factor and reordering the factor levels so that 'Disagree' is the reference level"
#> ..- attr(*, "label")= chr "No one group should dominate in society"
#> $ deserving_f2_flip: Factor w/ 2 levels "Disagree","Agree": 2 2 2 2 1 2 2 2 2 2 ...
#> ..- attr(*, "transformation")= 'glue' chr "Converting 'deserving' to a dichotomous factor and reordering the factor levels so that 'Disagree' is the reference level"
#> ..- attr(*, "label")= chr "Groups at the bottom are just as deserving as groups at the top"
# show how the levels are flipped when "flip_levels = TRUE"
levels(new_df$top_f2)
#> [1] "Agree" "Disagree"
levels(new_df$top_f2_flip)
#> [1] "Disagree" "Agree"