In Bart6114/licorice: Plot Likert-Like Data

licorice is an R package that eases the plotting of Likert-like data. It has been heavily inspired by the likert package from Bryer and Speerschneider.

licorice makes use of the ggplot2 plotting engine in such a way that interference by the licorice package in terms of theming is kept to a minimum (the graphs diplayed belowed are themed using the ggthemr package). When using the licorice function a ggplot2 object is returned which can then be added upon to your liking.

library(knitr)
opts_knit$set(upload.fun = imgur_upload, base.url = NULL)

library(ggthemr)
ggthemr("flat")

Installation

For now, no CRAN version exists and you'll have to install from GitHub using devtools.

devtools::install_github("Bart6114/licorice")

Preparing the data

The licorice function expects a given structure of data. The example pisatest dataset can be used as a reference.

A minimal requirement is the presence of the question, response and count column. Additionally a group column can be added.

library(licorice)

head(pisatest)

For example the gapsample dataset is not structured as it should be. Some simple preparations make it suited for licorice.

head(gapsample)

library(dplyr)
gap_fixed<-
  gapsample %>%
  group_by(question, response) %>%
  summarise(count = n())

head(gap_fixed)

Plotting the data

Three main plots are available. First a centered plot is shown; here the junction between two categories (which can be controlled by the middle_pos parameter) is centered. If the factor levels of the reponse variable are not set correctly, they can be specified using the answer_order parameter. If a middle_pos value of e.g. 2 is given, the results are centered at the junction between the second and third response type.

my_order<-
  c("Strongly disagree","Disagree", "Agree", "Strongly agree")


licorice(pisatest, answers_order = my_order, middle_pos = 2, type = "center", sort=T)

One can also fill the vertical space using a filled plot (also notice the sort argument).

licorice(pisatest, answers_order = my_order, type = "fill", sort=TRUE)

We can also have a look at the count data.

licorice(pisatest, answers_order = my_order, type = "count")

You can also show plots in combination with each other using existing functionality (the gridExtra library).

library(gridExtra)

grid.arrange(
  licorice(pisatest, my_order, middle_pos = 2.5, type = "center", sort=TRUE),
  licorice(pisatest, my_order, type = "count", sort=TRUE) +
    theme(axis.text.y=element_blank()) +
    scale_fill_discrete(""),
  ncol = 2,
  widths = c(3/4,1/4)
)

Groups (as the countries in the graph above) are shown automatically when a group column is available in the data set. For example, when using the generated gap_fixed data set (where not group column is present), no group is shown;

levels(gap_fixed$response)

licorice(gap_fixed, middle_pos = 4, sort=TRUE) +
  theme(legend.position="right")

Bart6114/licorice documentation built on May 5, 2019, 10:26 a.m.