24 Visualize a Relationship

How to build scatterplots to show how two scores go together

Authors

Andrew Silhavy

Shane McCarty

Published

10.05.2026

Abstract

This chapter teaches researchers to visualize relationship research questions, which examine how two continuous scores go together. In Play 1, using real insurance data, researchers build a scatterplot one layer at a time: data, mapping, points, a line of best fit, grouping with color or facets, labels, a theme, and axis scales. In Play 2, using the Cohort 10 health status data, they put a third variable into the plot with color and shape (self-rated health against subjective social status, by minoritized status and sex) and see when a rating can be treated as a score. In The Lab, researchers make a scatterplot with the lab dataset and then learn what to do when survey scores stack on top of each other: jitter the points and say so in the figure caption. In Your Turn, researchers use a copy-ready template and a checklist to build the relationship plot for their own research question.

Keywords

ggplot2, scatterplot, line of best fit, jitter, relationship

Ref’s kickoff: find the Ref before you code

Open Project → Open .qmd → Run load-library chunk → Run All Chunks Above → code. If anything looks wrong, use Ref’s Quick Checklist.

Resources: ggplot2

24.1 How This Chapter Works

You will go through this chapter three times, with three different datasets.

Pass	Section	Dataset	What you do
1	The Play	`insurancedata` (example)	Watch the play. Read the code and the output, step by step.
2	The Lab	lab dataset	Run the play. Make two scatterplots yourself during the R Lab, then check your own work. Labs are not graded.
3	Your Turn	your team’s dataset	Call your own play. After you finish all of the R Labs, come back and make the plot for your own research question.

Go: Know your route first

This chapter is the Visualize step on The Analysis Map for a relationship question: two continuous scores. New to ggplot2 layers? Read Coding a Plot first.

24.2 📋 The Play

24.2.1 Example Data: `insurancedata`

The example dataset is from the “Medical Cost Personal Costs” database on Kaggle. For this chapter, we refer to it as insurancedata.

insurancedata <- read.csv("data/insurance.csv")

library(ggplot2)

24.2.2 Relationship Research

For relationship research, the variables are 📏 Continuous because they range along a continuum, such as a scale for beliefs from 1 (strongly disagree) to 6 (strongly agree). To examine the relationship between two continous variables, use the steps below to build a custom scatterplot.

STEP-BY-STEP TO BUILD A SCATTERPLOT

Step 1: Set up data

ggplot(data = insurancedata)

Step 2: Set up mapping

What is the relationship between BMI and insurance charges?

ggplot(data = insurancedata,
       mapping = aes(
         x = bmi,
         y = charges))

📊 Layer: Data + Mapping

Two layers are used:

Data: insurancedata dataset
Mapping: aes(x = bmi, y = charges)

Step 3: Add geometry (points)

Add points to create a scatterplot using geom_point():

ggplot(data = insurancedata,
       mapping = aes(
         x = bmi,
         y = charges)) + geom_point()

🎨 Layer: Geometries

Geometry: geom_point() displays data as points

Caution: Don’t start a line with +

Layers are joined with a + at the end of a line. If a line starts with +, R runs the plot without that layer and then gives you an error. Look at the code above: the + comes right before geom_point(), on the same line as the layer before it.

Step 4: Add a statistical layer (trend line)

Add a linear model line with geom_smooth():

ggplot(data = insurancedata,
       mapping = aes(
         x = bmi,
         y = charges)) + geom_point() +
  geom_smooth(method = "lm")

📈 Layer: Statistics

Statistics: geom_smooth() calculates and displays a trend line

Notice how it looks like we have two different populations? Let’s explore this further.

Step 5. Incorporate grouping (color)

If you have three variables (2 continouous, 1 categorical/grouping variable), then you can use facets. There are two main approaches to visualizing grouping variables:

Method 1: Using Color

Map the smoker variable to color:

ggplot(data = insurancedata,
       mapping = aes(
         x = bmi,
         y = charges,
         color = smoker)) +
  geom_point() +
  geom_smooth(method = "lm")

Go: Put a variable inside aes()

color = smoker is inside aes() because smoker is a variable in the dataset. If you want every point to be one color, put the color outside aes(), like this: geom_point(color = "#005a43").

🔵 Layer: Facet (extended)

Facet: Added color = smoker to group by smoking status

Method 2: Using Facets

Create separate panels for each group with facet_wrap():

ggplot(data = insurancedata,
       mapping = aes(
         x = bmi,
         y = charges)) +
  geom_point() +
  geom_smooth(method = "lm") +
  facet_wrap(~smoker)

📊 Layer: Facets

Facets: facet_wrap(~smoker) creates separate panels

Both methods reveal that smoking status significantly affects the relationship between BMI and insurance charges!

Step 6. Customize Appearance

Adding Titles and Labels

Make your plot more informative with descriptive titles:

ggplot(data = insurancedata,
       mapping = aes(
         x = bmi,
         y = charges)) +
  geom_point() +
  geom_smooth(method = "lm") +
  ggtitle("Insurance charges vs 
          BMI for smokers and non-smokers") +
  xlab("Body Mass Index (BMI)") +
  ylab("Insurance Charges")

Changing the Theme

Apply a professional theme with theme_bw():

ggplot(data = insurancedata, mapping = aes(x = bmi, y = charges, color = smoker)) +
  geom_point() +
  geom_smooth(method = "lm") +
  ggtitle("Insurance charges vs 
          BMI for smokers and non-smokers") +
  xlab("Body Mass Index (BMI)") +
  ylab("Insurance Charges") +
  theme_bw()

Figure 1. Insurance charges by BMI, with a linear trend line.

🎨 Layer: Theme

Theme: theme_bw() controls the overall visual appearance

Required for Team Poster: pick one theme for your team

Team members should pick one theme (from the complete themes) to use with all plots in your individual reports and the team poster!

Adjusting Axis Scales

Sometimes, you need to manually control the range of your axes for better visualization or to match other plots or to capture the entire range of the possible response options (e.g., 1-6, 1-10, 1 - 100).

Set specific limits with xlim() and ylim().

ggplot(data = insurancedata, 
       mapping = aes(x = bmi, y = charges, color = smoker)) +
  geom_point() +
  geom_smooth(method = "lm") +
  xlim(15, 55) +  # Set x-axis from 15 to 55
  ylim(0, 65000) +  # Set y-axis from 0 to 65000
  ggtitle("Insurance charges vs BMI for smokers and non-smokers") +
  xlab("Body Mass Index (BMI)") +
  ylab("Insurance Charges") +
  theme_bw()

📏 Layer: Scales

Scales: xlim() and ylim() control axis ranges

Caution: Data outside the limits will be removed

Using xlim() and ylim() removes any data points outside the specified range. This can affect trend lines!

Required: x and y axes cover the full scale

If your response options range from 1 to 6. Use ylim( ) to adjust the y-axis to range from 1 to 6 too.

Go: Save your plot with ggsave()

Every plot in your RD Report and Final Report must be saved with ggsave(), so you can add it to your team poster. Save the plot to an object, then add two lines:

ggsave("plots/plot1_wellbeing_support_scatter.png",
       plot = plot1_wellbeing_support_scatter, width = 8, height = 5, dpi = 300)

The file name is the figure number, the variables, and the plot type. The full explanation is in Write the Results.

24.2.3 Play 2: Health status and subjective social status (class dataset)

The insurance data showed the layers. This second play uses a dataset collected by FRI students, and it adds the two things your own relationship plots will need: a third variable that splits the trend line into groups, and a rating scale on the y axis treated as a continuous score.

About the dataset: health_status_data.csv (the Cohort 10 class dataset)

Every FRI Public Health team in Cohort 10 asked the same self-rated health question, Would you say your health in general is excellent, very good, good, fair, or poor?, along with the class variables of that year (age, sex, racialized identity, perceived income, and the MacArthur ladder of subjective social status). The five team datasets were merged into one file of 343 respondents so that health status could be examined across the whole class. SAMPLE_ID says which team’s study a person came from.

This is real data, already cleaned and de-identified: no response IDs, passwords, dates, or open-ended answers, and the five team samples differ in whom they recruited (students, families, farmers-market visitors), which is why the chapters filter by age. The variable names predate the class naming rules, so the Play renames them: HEALTH_STATUS becomes HEALTHSTATUS (1 = Poor … 5 = Excellent); SSS is subjective social status (1 to 10); SEX is biological sex (0 = male, 1 = female); and RACIALIZED_IDENTITY (1 = identified as white only, 0 = any other identity) becomes MINORITIZED_01 (1 = minoritized, 0 = not), named for the 1 the way a _01 variable should be. The other columns (RG_…, RI_…, PoorFairHealth, ExcellentHealth) are earlier recodes of the same questions and are not used here.

The question. Is subjective social status (where people place themselves on a 10-rung ladder) related to self-rated health, and does the relationship look the same for people who are minoritized and people who are not? Health status is a 1-to-5 rating. It is ordered and we treat it as a continuous score, the same way the composites in your own data are scores, so that it can be the y axis of a scatterplot, one variable in a correlation (Relate 2 Variables), and the outcome of a regression (Relate 3+ Variables).

library(dplyr)
healthstatusdata <- read.csv("data/health_status_data.csv")

healthstatusdata <- healthstatusdata |>
  rename(HEALTHSTATUS = HEALTH_STATUS) |>
  mutate(
    MINORITIZED_01 = if_else(RACIALIZED_IDENTITY == 1, 0, 1),          # 1 = minoritized (any identity other than white only)
    MINORITIZED    = factor(MINORITIZED_01, levels = c(0, 1), labels = c("Not minoritized", "Minoritized")),
    SEX            = factor(SEX, levels = c(0, 1), labels = c("Male", "Female")))

healthstatusdata_25 <- healthstatusdata |> filter(AGE >= 25)      # adults 25 and older: the teams' student samples skew 18 to 22

nrow(healthstatusdata); nrow(healthstatusdata_25)

[1] 343

[1] 122

healthstatusdata_25 |> count(SEX, MINORITIZED)

     SEX     MINORITIZED  n
1   Male Not minoritized 15
2   Male     Minoritized 10
3 Female Not minoritized 75
4 Female     Minoritized 22

#source: Cohort 10 class dataset (five FRI Public Health teams, merged)
#explanation: rename to the class conventions, make the two grouping factors, and keep respondents 25 and older

The filter is a choice, and it is reported: with everyone included, the 18-to-22-year-old students dominate the picture and their health ratings barely vary. From 25 up, the ladder and health status have room to move. Say in Methods when you restrict a plot this way, and why.

library(ggplot2)

plot2_health_sss_minoritized_scatter <- ggplot(healthstatusdata_25,
                          aes(x = SSS, y = HEALTHSTATUS)) +
  geom_jitter(aes(color = MINORITIZED, shape = SEX),
              width = 0.15, height = 0.12, size = 2, alpha = 0.7) +   # jitter a little: both variables are whole numbers
  geom_smooth(aes(color = MINORITIZED), method = "lm", se = FALSE, linewidth = 1.3) +
  geom_smooth(method = "lm", se = FALSE, color = "black", linetype = "dashed", linewidth = 1) +
  scale_color_manual(values = c("Not minoritized" = "#2a9d8f", "Minoritized" = "#9b5de5")) +
  scale_x_continuous(breaks = 1:10, limits = c(1, 10)) +
  scale_y_continuous(breaks = 1:5, limits = c(0.8, 5.2),
                     labels = c("Poor", "Fair", "Good", "Very Good", "Excellent")) +
  labs(title = "Self-rated health by subjective social status (age 25+)",
       x = "Subjective social status (1 = bottom rung, 10 = top rung)",
       y = "Health status", color = "Racialized identity", shape = "Sex") +
  theme_bw()

plot2_health_sss_minoritized_scatter

A scatterplot with subjective social status from 1 to 10 on the x axis and health status from Poor to Excellent on the y axis. Points are colored teal for not-minoritized and purple for minoritized respondents. The teal trend line rises from Good toward Very Good as status increases; the purple line is nearly flat; a dashed black line for everyone rises gently between them. — Figure 2. Self-rated health by subjective social status among respondents aged 25 and older (Cohort 10 class dataset, n = 122). Points are jittered; lines are linear trends for each group and for everyone (dashed).

ggsave("plots/plot2_health_sss_minoritized_scatter.png", plot = plot2_health_sss_minoritized_scatter,
       width = 8, height = 5, dpi = 300)

#source: ggplot2 documentation https://ggplot2.tidyverse.org/reference/geom_smooth.html
#explanation: color is the third variable, so geom_smooth() draws one line per group; a second geom_smooth() without a color mapping draws the overall line; the y axis shows the rating labels but the values stay numeric

Read the lines before the points. For respondents who are not minoritized, health rises with social status; for minoritized respondents the line is nearly flat, so a higher rung on the ladder is not matched by better self-rated health. The dashed overall line averages the two and hides that difference, which is the reason to map a third variable at all. Relate 3+ Variables tests whether the two slopes differ (an interaction); here the plot is the question.

Go: two ways to put the third variable in

color = MINORITIZED inside aes() gives one line per group on the same panel, which is best for comparing slopes. facet_wrap(~ MINORITIZED) gives one panel per group, which is best when the points would otherwise pile on top of each other. Try both; keep the one where the pattern is easiest to see, and say which groups are being compared in the caption.

Caution: a rating is a score only if you say so

Treating a 1-to-5 health rating as continuous is a standard choice, and a choice: it assumes the steps are roughly equal. Say it in your Methods (“self-rated health was treated as a continuous score from 1 to 5”), keep the y axis on the full 1-to-5 scale (Required), and never average a variable whose numbers are only labels (a _#CAT variable, Transforming Your Data).

24.3 🏈 The Lab

You watched the play above with the insurancedata dataset (Pass 1). Now run two plays yourself with the lab dataset, a survey about what people believe about mental health.

Ref the raccoon

Ref’s kickoff. This lab is not graded, so I am here to help you check your own work. Try each play before you open my check or the solution. If your answer does not match, that is not a penalty. It is how you find out what to fix.

Lab play	What is new
Screen Left	a scatterplot, just like the play above
Screen Right	points that stack on top of each other, and how to jitter them

Before you start

Open your RStudio Project and your lab .qmd file.
Load the clean lab data by running this chunk. It runs the cleaning script you built in the earlier labs.

source("lab_prep.R")   # creates mh_clean
nrow(mh_clean)         # how many people are in the clean dataset?

Ref’s check: how many rows?

nrow(mh_clean) should be 188. If you get 200, you are looking at the raw data: the duplicate response and the people who failed the attention checks have not been removed yet.

24.3.1 Lab Play 1: Screen Left (a scatterplot)

Research question: Is age (AGE) related to the belief that substance use causes mental illness (MIAQ_SUBSTANCE, 1 to 7)?

Your task: Go back to the scatterplot play and rebuild it for this question. Put AGE on the x axis and MIAQ_SUBSTANCE on the y axis, use geom_point(), add a line of best fit, label both axes in plain words, add a title, and use theme_bw().

Ref’s check: what should the plot look like?

Your plot should include 186 people. R will warn you that it removed 2 rows with missing values (two people typed an impossible age). That warning is expected.
The line of best fit should slope gently upward, and the dots should be spread widely around it. This is what a weak relationship looks like (r = 0.20). Most relationships in survey research look more like this than like a tight line.
You can see almost every person as a separate dot, because AGE can take many different values.

👀 See the solution

plot3_substance_age_scatter <- ggplot(
  data = mh_clean,
  mapping = aes(
    x = AGE,
    y = MIAQ_SUBSTANCE)) +
  geom_point() +
  geom_smooth(method = "lm") +
  ylim(1, 7) +
  ggtitle("Age and the Belief that Substance Use Causes Mental Illness") +
  xlab("Age (years)") +
  ylab("Substance use as a cause (1 = not important, 7 = very important)") +
  theme_bw()

print(plot3_substance_age_scatter)

A scatterplot with age on the x axis and the substance use belief score on the y axis. The points are widely spread and the line of best fit slopes gently upward. — Figure 3. Age and the belief that substance use causes mental illness in the lab dataset. The line is a linear line of best fit.

24.3.2 Lab Play 2: Screen Right (a scatterplot with stacked points)

Research question: Is mental health stigma (STIGMA, 1 to 5) related to mental health self-efficacy (EFFICACY, 1 to 6)?

Your task, part 1: Make the same kind of scatterplot as Lab Play 1, with STIGMA on the x axis and EFFICACY on the y axis. Use geom_point(). Then count the dots. Does it look like 188 people?

A scatterplot of stigma and self-efficacy where the points sit in a grid pattern and many points hide other points. — Figure 4. Mental health stigma and self-efficacy with geom_point(): people who have the same two scores are drawn on top of each other.

Foul: stacked points

EFFICACY is the average of three survey items on a 6-point scale, so it can only take a few values (1, 1.33, 1.67, 2, and so on). STIGMA averages eight items, so it has more possible values, but still far fewer than there are people. When two people have the same two scores, geom_point() draws one dot on top of the other, and your plot hides most of your sample. This happens with almost every survey scale, so check for it every time.

Your task, part 2: Fix it. Replace geom_point() with geom_jitter(). Jitter nudges each dot a tiny random distance so that stacked dots separate. Adding alpha = 0.5 makes the dots see-through, so darker areas show where more people are.

  geom_jitter(width = 0.08, height = 0.08, alpha = 0.5) +

Caution: Jitter a little, and say that you did

Jitter only moves the dots in the picture. It does not change your data or your line of best fit. Keep width and height small (less than half the distance between two possible scores), and tell your reader in the figure caption that the points are jittered.

Ref’s check: how many dots can you see?

There are 188 people in this plot, but with geom_point() you can only see 123 dots. The biggest stack has 5 people on a single dot.
After you switch to geom_jitter(), you should see small clouds of dots where the single dots used to be. Your dots will not be in exactly the same places as the solution, because jitter is random. That is fine.
The line of best fit should slope downward: people who report more stigma report less self-efficacy (r = -0.47). The line is the same with or without jitter.

👀 See the solution

plot5_efficacy_stigma_scatter <- ggplot(
  data = mh_clean,
  mapping = aes(
    x = STIGMA,
    y = EFFICACY)) +
  geom_jitter(width = 0.08, height = 0.08, alpha = 0.5) +
  geom_smooth(method = "lm") +
  xlim(1, 5) +
  ylim(1, 6) +
  ggtitle("More Stigma, Less Mental Health Self-Efficacy") +
  xlab("Mental health stigma (1 = low, 5 = high)") +
  ylab("Mental health self-efficacy (1 = low, 6 = high)") +
  theme_bw()

print(plot5_efficacy_stigma_scatter)

A scatterplot with stigma on the x axis and self-efficacy on the y axis. Jittered, see-through points form small clouds. The line of best fit slopes downward. — Figure 5. Mental health stigma and mental health self-efficacy in the lab dataset. Points are jittered to show overlapping responses; the line is a linear line of best fit.

Ref’s call: fouls to check if your plot does not match

Your scatterplot looks like a neat grid of dots → your points are stacked. Use geom_jitter() (Lab Play 2).
Your jittered plot looks like a shapeless cloud → your jitter is too big. Make width and height smaller.
object 'EFFICACY' not found → EFFICACY and STIGMA are composites created in lab_prep.R. Run source("lab_prep.R") first.
An error that mentions + → each ggplot line must end with +. A line that starts with + breaks the plot.

Ref the raccoon

Ref’s final whistle. That is the end of the lab. Save your .qmd, render it, and back up your .qmd and .R files to the R Labs folder in your ELN before you quit RStudio.

24.4 🏆 Your Turn

There is no Ref. It’s game time, your turn!

Nobody has analyzed your team’s data before, so there is no answer to check against. Copy the play that matches your research question, change every word that starts with SWAP, and use the checklist at the end of this section to check your own plot. It is the same checklist your peer mentor and Dr. Shane use.

Once you finish all of the R Labs, come back here with your team’s dataset and make the plot for your relationship research question.

24.4.1 The relationship play

plot1_SWAPNAME <- ggplot(                    # SWAP: plot number + variables + plot type
  data = cleandata,                          # alldata = your team's imported .cleandataset (Import Data Once)
  mapping = aes(
    x = SWAPPREDICTOR,                      # SWAP: your predictor
    y = SWAPOUTCOME)) +                     # SWAP: your outcome
  geom_point() +                            # points stacked? use geom_jitter(width = 0.08, height = 0.08, alpha = 0.5)
  geom_smooth(method = "lm") +
  ggtitle("SWAP: your title") +
  xlab("SWAP: a plain-words label and the scale range") +
  ylab("SWAP: a plain-words label and the scale range") +
  theme_bw()                                # use the same theme in every plot

print(plot1_SWAPNAME)
ggsave("plots/plot1_SWAPNAME.png",                # SWAP: the variables and plot type, e.g. plot1_wellbeing_treated_barplot
       plot = plot1_SWAPNAME,
       width = 8, height = 5, dpi = 300)

24.4.2 Checklist for your RD Report and Final Report

Open the plot checklist

Criteria	Ask yourself
Data and mapping	Is the cleaned dataset used, with the correct variables on x and y?
Geometry	Is this the best plot for the data (e.g., scatterplot for two continuous variables; barplot or boxplot for groups)?
Stacked points	If points sit on top of each other, are they jittered, and does the caption say so?
Groups	Were very small groups combined or excluded? Does the caption say what was combined or excluded, with the counts?
Stats	Is there a line of best fit on a scatterplot? Are there error bars on a barplot? (Error bars are required in the Final Report, not the RD Report.)
Axes and labels	Are both axes labeled in plain words (not variable names)? Does each axis cover the full range of the scale?
Facets	Would a facet (one panel per group) help answer your research question?
Appearance	Is there an accurate title? Is a non-default theme used (the same theme in every plot)? Does color add meaning?
Code formatting	Can you read the code without scrolling to the right? Does the plot print without errors or extra output?
Figure caption	Does the chunk have `fig-cap`, starting with “Figure 1.”, “Figure 2.”, and enough detail to stand alone?
Alt text	Does the chunk have `fig-alt` that describes what the plot shows?
ggsave( )	Is the plot saved with `ggsave()` into the `plots` folder of your RStudio Project, named `plot#_variables_type.png` (for example `plot1_wellbeing_treated_barplot.png`), with the object named the same?
Stands alone	Could a reader understand the figure without reading your text?
Referenced in text	Does your results paragraph point to it (e.g., “Figure 1 shows the association between …”)?

Ref’s final whistle: before you quit RStudio

Save → Render → Back up to ELN → Quit, Don’t Save workspace. Details: Ref’s Quick Checklist.

24.1 How This Chapter Works

24.2 📋 The Play

24.2.1 Example Data: insurancedata

24.2.2 Relationship Research

STEP-BY-STEP TO BUILD A SCATTERPLOT

Step 1: Set up data

Step 2: Set up mapping

Step 3: Add geometry (points)

Step 4: Add a statistical layer (trend line)

Step 5. Incorporate grouping (color)

Method 1: Using Color

Method 2: Using Facets

Step 6. Customize Appearance

Adding Titles and Labels

Changing the Theme

Adjusting Axis Scales

24.2.3 Play 2: Health status and subjective social status (class dataset)

24.3 🏈 The Lab

24.3.1 Lab Play 1: Screen Left (a scatterplot)

24.3.2 Lab Play 2: Screen Right (a scatterplot with stacked points)

24.4 🏆 Your Turn

24.4.1 The relationship play

24.4.2 Checklist for your RD Report and Final Report

24.2.1 Example Data: `insurancedata`