Visualization

In this page, I’ll share R commands for visualizing data. The dataset I use here is a subset of the 2018 CCES Common Content Dataset. You can download the full dataset from the CCES Dataverse.

Opening data with import()

I use the rio package, which can read many file formats (.dta, .csv, .xlsx) with a single function.

library(rio)
cces <- import("cces18_subset.dta")

Looking at variables

head(cces)     # Shows first 6 rows
View(cces)     # Opens data in a spreadsheet view
names(cces)    # Lists all variable names

You can rename a variable like this:

names(cces)[names(cces) == "faminc_new"] <- "family_income"

Making a new variable

Let’s create an age variable from birthyr.

class(cces$birthyr)          # Shows if it's numeric, character, etc.
summary(cces$birthyr)        # Shows min, max, mean, etc.
sum(is.na(cces$birthyr))     # Counts missing values (NA in R)

cces$age <- 2018 - cces$birthyr
summary(cces$age)

Histograms

Histograms are helpful to look at the distribution of a continuous variable. The base R function is hist().

hist(cces$age)                                     # Basic histogram
hist(cces$age, main = "Age", xlab = "Age")          # With better labels

# Percentages instead of counts
hist(cces$age, freq = FALSE, main = "Age", xlab = "Age", ylab = "Percent")

# Change the number of bars
hist(cces$age, breaks = 10, main = "Age", xlab = "Age")
hist(cces$age, breaks = 15, main = "Age", xlab = "Age")

You can also split the data into groups and compare histograms side by side.

males <- cces[cces$gender == 1, ]
females <- cces[cces$gender == 2, ]

hist(males$age, main = "Age - Males", xlab = "Age")
hist(females$age, main = "Age - Females", xlab = "Age")

Working with a coded variable

The faminc_new variable is coded, so let’s check what values it takes before summarizing it.

summary(cces$faminc_new)
table(cces$faminc_new)        # Count how many of each value
sum(is.na(cces$faminc_new))
unique(cces$faminc_new)       # Shows all unique values

Note that 97 means “Prefer not to say,” so we exclude it before summarizing.

valid_income <- cces$faminc_new[cces$faminc_new < 20]
summary(valid_income)
mean(valid_income, na.rm = TRUE)   # na.rm = TRUE ignores missing values

hist(valid_income, main = "Family Income", xlab = "Income Level")
table(valid_income)

Bar plots for categorical variables

For categorical variables like marital status, use a bar plot instead of a histogram.

marstat_counts <- table(cces$marstat)
marstat_counts

barplot(marstat_counts, main = "Marital Status", ylab = "Number of People")

# As percentages
marstat_percent <- prop.table(marstat_counts) * 100
barplot(marstat_percent, main = "Marital Status", ylab = "Percent")

Saving a plot

png("hist_age.png")   # Start saving
hist(cces$age)          # Make your plot
dev.off()                 # Finish saving