Best plot to represent categorical data

Question

I have a dataset containing two time-points (T1, T2), two groups (Group A, Group B), and three types of emotional faces (happy, sad, neutral) and reaction time as the dependent variable. I have used a bar plot to represent the data using R. Are there any other graphs that can be used to represent this kind of data? A sample dataset is given below (for each participant, there are three emotion types and two time-points):

participant	emotion	group	time	rt
AA	happy	groupA	t1	47.09
AA	happy	groupA	t2	40.09
AA	sad	groupA	t1	33.99
AA	sad	groupA	t2	33.99
AA	neutral	groupA	t1	104.97
AA	neutral	groupA	t2	98.75
AB	happy	groupB	t1	54.65
AB	happy	groupB	t2	64.65
AB	sad	groupB	t1	53.99
AB	sad	groupB	t2	43.99
AB	neutral	groupB	t1	24.97
AB	neutral	groupB	t2	98.75

Quinten · Answer 1 · 2022-07-30T07:54:13.350

1

An option could be representing your data using geom_point like this:

library(ggplot2)
library(dplyr)
df %>%
  ggplot(aes(x = time, y = rt, color = emotion)) +
  geom_point(alpha = 0.4, size = 2) +
  facet_wrap(~participant)

^{Created on 2022-07-30 by the reprex package (v2.0.1)}

Another option if you only want to represent your categorical data, you could use a mosaic plot like this:

library(vcd)
mosaic(~ participant + emotion + time, data = df)

Please note: this works good if you have different proportion between categorical variables.

edited Jul 30 '22 at 07:54

answered Jul 30 '22 at 07:01

Quinten

35,235
5
20
53

Thank you for the suggestions. Is it possible to plot a mosaic plot using the data mentioned here. I have around 30 participants in each group. If so, how do you add the fourth variable? – Christina Aug 02 '22 at 04:08
@Christina, yes you can just simply add the fourth variable in the function like ohters! – Quinten Aug 02 '22 at 07:25

score 0 · Answer 2 · edited Jul 31 '22 at 03:39

You want to plot the information of four different variables.

One way could be to use a 3D plot (e.g., using points instead of bars) and add the fourth dimension by coloring the data points. For instance: y-axis = rt, x-axis = time, z-axis = emotion, color = group.

library(plotly)

df <- data.frame(
  group = c(rep("A", 6), rep ("B", 6)),
  emotion = rep(c("happy", "happy", "sad", "sad", "neutral", "neutral"), 2),
  time = rep(c("t1", "t2"), 6),
  rt = c(47.09, 40.09, 33.99, 33.99, 104.97, 98.75, 54.65, 64.65, 53.99, 43.99, 24.97, 98.75)
)

plot_ly(df, x = ~time, y = ~rt, z = ~emotion, color = ~group)

Depending on the data (and your personal preferences), the plot could appear "overcrowded"/ not too easy to read. If this is the case (in my view, here it is), you can reduce one dimension from the suggestion above (e.g., make it a 2D plot) and create two plots (e.g., three 2D plots; left plot entitled "positive", middle plot entitled "negative", right plot entitled "neutral")

library(ggplot)

df <- data.frame(
  group = c(rep("A", 6), rep ("B", 6)),
  emotion = rep(c("happy", "happy", "sad", "sad", "neutral", "neutral"), 2),
  time = rep(c("t1", "t2"), 6),
  rt = c(47.09, 40.09, 33.99, 33.99, 104.97, 98.75, 54.65, 64.65, 53.99, 43.99, 24.97, 98.75)
)

df %>% 
  ggplot(aes(x = time, y = rt, group = group, color = group)) +
  geom_point(size = 3) +
  geom_line() +
  facet_wrap(~emotion)

Best plot to represent categorical data

2 Answers2