A quick and easy guide to modifying and customizing the look of your AMCE plot outputs from cjoint. The figure below shows the comparison between the default plots produced with cjoint relative to the customized plots using the code provided, below.
library(cjoint)
library(car)
library(weights)
library(tidyverse)
In this case, I have created a sample dataset that can be used and tweaked for the purpose of this tutorial. Note: I have specified the factor levels of each variable. This is key when plotting variables in a specific order and when determining the reference/base group.
samp_dat <-
tibble(
## Use `factor` in order to label the factor levels
Country = sample(
c(
"Bosnia",
"Russia",
"Lebanon",
"Iran",
"Sudan",
"Libya",
"India",
"Pakistan"
),
size = 80,
replace = T
)
%>% factor(
levels = c(
"Bosnia",
"Russia",
"Lebanon",
"Iran",
"Sudan",
"Libya",
"India",
"Pakistan"
)
),
Education = sample(
c(
"Elementary School",
"High School",
"College",
"Master's Degree"
),
size = 80,
replace = T
) %>% factor(
levels = c(
"Elementary School",
"High School",
"College",
"Master's Degree"
)
),
Gender = sample(
c(
"Male",
"Female"
),
size = 80,
replace = T
) %>% factor(
levels = c(
"Male",
"Female"
)
),
## You can use `these hashes` and `tidyverse` to create column names that are non-traditional (including spaces)
`English Fluency` = sample(
c(
"Intermediate",
"Advanced",
"Fluent"
),
size = 80,
replace = T
) %>% factor(
levels = c(
"Intermediate",
"Advanced",
"Fluent"
)
),
Religion = sample(
c("Christian",
"Jewish",
"Muslim"
),
size = 80,
replace = T
) %>% factor(
levels = c(
"Christian",
"Jewish",
"Muslim"
)
),
greencard = sample(
c(0, 1),
size = 80,
replace = T
),
respondent = rep(
c(1:4),
20
)
)
Here, I include some code that is useful when modifying or customizing the plot to fit your needs. In this example, I have removed the grid lines in the plot, the background is white, and the labels are changed such that the category/variable is in bold, the reference/base group is in italics, and the other groups are in plain text.
The code is commented to indicate which codes correspond with plot modifications.
In this code, I have selected colors from the Viridis Palette. This is helpful!
samp_dat %>%
## The analyses with `formula` argument necessary
amce(
formula = greencard ~ Education + Gender + `English Fluency` + Religion + Country,
respondent.id = "respondent"
) %>%
## Plot the AMCE as normal
plot(
#add colors from viridis pallet; 5 colors for 5 attributes
col = c("#DCE319", "#55C667", "#1F968B", "#39568C", "#440154"),
## X-Axis label
xlab = "Likelihood of giving Green Card",
## Changing the plot theme
plot.theme = theme(
## Making the text on Y-Axis horizontally justified
axis.text.y = element_text(hjust = (0)),
## Removing the X and Y-Axis tick lines
axis.ticks = element_blank(),
## Changing the text options
text = element_text(
## Changing text size
size = 16,
## Here, I make the category bold text, the baseline italics, and rest plain text
### You have to start from the bottom upward
face = c(
# Religion
## "Muslim" in plain text
"plain",
## "Jewish" in plain text
"plain",
## Baseline of "Christian" in italics
"italic",
## Category of "Religion" in bold
"bold",
# Gender
"plain",
"italic",
"bold",
# English Fluency
"plain",
"plain",
"italic",
"bold",
# Education
"plain",
"plain",
"plain",
"italic",
"bold",
# Country
"plain",
"plain",
"plain",
"plain",
"plain",
"plain",
"plain",
"italic",
"bold"
)
),
## No legend
legend.position = 'none',
## No major or minor grid lines
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
## No color in the panel background
panel.background = element_blank(),
axis.line = element_line(colour = "black")
)
)
We can do the same modifications when we are interested in conditional AMCEs. In this example, we are graphing the likelihood of being awarded a green card conditional on both religion and country of origin.
samp_dat %>%
amce(
formula = greencard ~ Education + Gender + `English Fluency` + Religion * Country,
respondent.id = "respondent",
## Adding in interaction
respondent.varying = "Religion"
) %>%
plot(
## add one color this time; color from viridis
col = c("#440154"),
plot.display = "interaction",
xlab = "Likelihood of giving Green Card",
plot.theme = theme(
axis.text.y = element_text(hjust = (0)),
axis.ticks = element_blank(),
text = element_text(
size = 16,
face = c(
# Country
"plain",
"plain",
"plain",
"plain",
"plain",
"plain",
"plain",
"italic",
"bold"
)
),
legend.position = 'none',
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.line = element_line(colour = "black")
)
)
