Please turn in your .qmd and .html file on Brightspace. Please use the same naming convention as previous labs: lastname_firstinit_lab_04.qmd and lastname_firstinit_lab_04.html.
The above link will take you to your editing file. Please do not remove anything from this editing file!! You will only add your code and work to this file.e.
2 Purpose
In Lab 3, you built a categorical version of your composite score(s), a clean overall summary table, and a grouped summary comparing your composite score across the groups of your other main variable. This lab turns those summaries into the actual visualizations you’ll use in your final project.
Your final project’s one-page information sheet is built around three findings:
One main variable on its own
The relationship between your two main variables
One main variable alongside one additional variable
For Findings 2 and 3, you’ll use the categorical version of your composite score (from Lab 3) rather than the raw numeric score. Category breakdowns like “33% described access as difficult” are easier for a general audience to take in at a glance than a mean and standard deviation.
This lab has you build one polished visualization for each of these. By the end of this lab, you should have 3 finished, publication-ready plots saved as image files, which you can use in your final project. Some of the ggplot2 code here is more involved than what you’ve used before, so this lab walks through each new piece in more detail than usual.
3 Lab activities
3.1 Restate your selected project
Recall the following table from Labs 2 and 3 with each project’s population, main variables, and composite score(s).
#
Project Question
Population
Population variables
Main variables
Suggested additional variables
Composite score
1
For individuals under the federal poverty line, how do food support systems help individuals’ food security?
People under 100% federal poverty line (FPL) and received free groceries or meals in the past 12 months
Composite score of ease of access to free groceries or meals (combined score from Q26D_)
2
For individuals who identify as LGBTQ, how does trust and/or fair treatment in healthcare effect self-reported health?
People who identify as LGBTQ and have a healthcare provider
LGBTQ, Q108
Q108B_2, Q108B_1, Q27
Q108, Q70F_4, Q70F_5, PPETHM, Q28O
Level of trust, courtesy, and respect around treatment (combined score from Q108B_1 and Q108B_2)
3
For individuals with school-aged children and below the federal poverty line, how do benefits received effect their mental health?
People who have at least one child (aged 5-18 years old) in the household and are under the 100% FPL
ANY_CHILDHH_5_18, FPL_LT50, FPL50_100
Q30_1-Q30_6, Q53_A-Q53_J, Q11, Q31_4
NUM_CHILD_HH, Q1, Q37, PPETHM, Q27
Total mental health (combined scores from Q30_), Total benefits (combined scores from Q53_, Q11, and Q31_4)
ImportantTask
Copy and paste the row from the above table that corresponds to your project into your .qmd file, same as you did in Lab 3.
3.2 Import your saved data from Lab 3
ImportantTask
From the .rda file you exported at the end of Lab 3, import wbns_06_cat and summary_table_overall into R using load() or import() from the rio package. Print out the number of rows and columns of wbns_06_cat to verify the import.
3.3 Plan your three findings
In the final project, we will present 2-3 findings. These findings will include a table or plot to demonstrate the data/relationship. Before writing any plotting code, decide what each of your findings will show and which variable(s) it needs. This is a planning step, there is no code required here!
Finding 1 (one main variable on its own): Which one of your main variables will you visualize by itself? This could be your composite score’s categorical version (from wbns_06_cat from Lab 3) or your other main variable (e.g., FOOD_INSEC, Q27, or your total_benefits category).
Finding 2 (the relationship between your two main variables): No decisions here! You will visualize the relationship between your composite score’s categorical version compared across the groups of your other main variable.
Finding 3, optional (one main variable + one additional variable): If you’re building this one, which additional variable from your Lab 2 table will you use, and which main variable will you pair it with? (Feel free to reference the table above for suggested additional variables.)
ImportantTask
For Finding 1 and 3, write down which variable(s) you will visualize.
3.4 Finding 1: Create the dataframe
We want to visualize one of your main variables on its own. The variable you choose to visualize should be one of the main variables from the table above. This will be a categorical version of your composite score (from Lab 3) or the other main variable (e.g., FOOD_INSEC, Q27).
Before plotting, we want to summarize the data into percentages to make our display a little more polished for the audience. We will summarize our main variable into a small dataframe of counts and percentages using count() and mutate():
finding1_data <- wbns_06_cat |>count(<your_main_variable>) |>mutate(pct = n /sum(n) *100)
1
Replace <your_main_variable> with the name of your main variable. count() counts how many respondents fall into each category of that variable, storing the result in a new column called n.
2
mutate(pct = n / sum(n) * 100) divides each category’s count by the total across all categories, turning n into a percentage. Since there’s no grouping variable yet, sum(n) here is just the overall total.
ImportantTask
Use count() and mutate() to build a small dataframe of counts and percentages for your Finding 1 variable. Save it as a new object called finding1_data.
3.5 Finding 1: Visualize one main variable
Using the finding1_data dataframe you just built, create a bar chart of your main variable using ggplot2. The code below gives a starting point for this plot. Note that we are using geom_col() instead of geom_bar() here because we already calculated the percentages ourselves in finding1_data. geom_col() plots the y-values you give it directly, while geom_bar() would try to do its own internal counting.
Check out the extra arguments in geom_col()! You can change the fill or the width of the bars.
2
geom_text() adds the percentage labels above each bar. vjust = -0.5 nudges the labels slightly above the top of the bar, and paste0(round(pct, 1), "%") rounds the percentage to one decimal place and adds a percent sign.
3
Your data still has the raw category values (e.g., "Yes" and "No"). scale_x_discrete(labels = ...) lets you swap in reader-friendly labels for the x-axis without changing your underlying data. The names on the left of each = ("No", "Yes") must exactly match the values in your data; the names on the right are what the reader will actually see.
4
We can add limits to the y-axis. When we plot percentages, it is often a good idea to plot the y-axis from 0 to 100. Check out the limits argument in scale_y_continuous() to set the y-axis limits.
5
labs() lets you add a title, x-axis label, and y-axis label. Use plain-language labels rather than variable names or question codes.
6
We can adjust the theme and font size!
7
Now we can print the plot!
ImportantTask
Build your Finding 1 visualization using ggplot2, adding and editing the code above with your own variable and category labels. Save the plot as a new object (e.g., finding1_plot).
3.6 Finding 2: Create the dataframe
Use the categorical version of your composite score here (e.g., ease_free_food_cat) rather than the raw numeric score. (In the final project, you may wish to use the number and a table from Lab 3 might be more appropriate to include.)
Unlike Finding 1, you now need percentages calculated within each group of your other main variable (e.g., what percentage of the “food insecure” group falls into each access category?). count() and mutate() alone can’t do that “within-group” calculation, so this time you’ll add group_by() before you calculate pct.
Replace <your_grouping_variable> and <your_categorical_composite> with your two main variables. count() counts how many respondents fall into every combination of the two variables (e.g., “food insecure AND difficult access,” “food insecure AND moderate access,” and so on), storing the result in a new column called n.
2
group_by(<your_grouping_variable>) tells the following mutate() to calculate sum(n)separately within each group.
3
Calculate the percent of each category within its group. The formula is the same as Finding 1.
ImportantTask
Use count(), group_by(), and mutate() to build a dataframe of within-group percentages for your two main variables. Save it as a new object called finding2_data.
3.7 Finding 2: Visualize the relationship between your two main variables
Using the finding2_data dataframe you just built, create a 100%-stacked bar chart with ggplot2. As in Finding 1, we use geom_col() instead of geom_bar() since you already calculated the percentages yourself. The code below gives a starting point for this plot.
Mapping fill = <your_categorical_composite> (in the ggplot() line above) tells geom_col() to draw one segment per category, stacked on top of each other by default. This creates the 100%-stacked bar. Check out the extra arguments in geom_col() too, like width.
2
geom_text() places its labels at the same x/y position as the data by default, which for stacked bars means the top of each segment. position = position_stack(vjust = 0.5) moves each label to the vertical center of its own segment instead, so it doesn’t look like it belongs to the segment above it. paste0(round(pct), "%") builds the label text from your calculated percentage.
3
Same as Finding 1: swap in reader-friendly labels for your x-axis categories without changing your underlying data. The names on the left must exactly match the raw category values in your data.
4
Same as Finding 1: add a plain-language title, x-axis label, and y-axis label. Keep your title descriptive, not causal.
5
Same as Finding 1:djust the theme and font size if you’d like.
6
Same as Finding 1: adjust the theme and font size if you’d like.
Tip
If your composite score doesn’t split cleanly into a small number of meaningful categories, a boxplot of the raw numeric score (geom_boxplot(aes(x = <your_grouping_variable>, y = <your_numeric_composite>))) is a reasonable alternative, just be consistent about which form (categorical or continuous) you use across Findings 2 and 3.
ImportantTask
Build your Finding 2 visualization using ggplot2, adding and editing the code above with your own variables and category labels. Save the plot as a new object (e.g., finding2_plot).
3.8 Finding 3: Create the dataframe
Pick one of the suggested additional variables from your Lab 2 table and pair it with the categorical version of one of your main variables.
Caution
Your additional variables haven’t been cleaned yet the way your main variables were in Lab 2, they may still contain “Don’t know,” “Refused,” or skip-pattern missingness. Take a quick look with count() or skim() before you build this dataframe, and handle any NA/“Don’t know”/“Refused” values the same (or improved) way you did for your main variables in Lab 2 (Section 1.6). Skipping this step can quietly distort your plot.
Once your additional variable is cleaned, build the dataframe the exact same way you did for Finding 2:
Clean your chosen additional variable, then use count(), group_by(), and mutate() to build a dataframe of within-group percentages. Save it as a new object called finding3_data.
3.9 Finding 3: Visualize one main variable + one additional variable
Same approach as Finding 2, again using the categorical version of your composite score so the story stays consistent across findings.
If you’re unsure what any piece of this code is doing, revisit the annotated explanations under Finding 2 above.
ImportantTask
Build your Finding 3 visualization using ggplot2, adding and editing the code above with your own variables and category labels. Save the plot as a new object (e.g., finding3_plot).
3.10 Export your visualizations and save your environment
Save each of your finished plots as a .png file. These files will be used in later in the final project! If you need to adjust the size and look of your plots, you will need to come back to this lab, modify them, and re-export them.
Save at dpi = 300 or higher so your plot looks crisp on a printed page.
Use a descriptive file name for each plot (e.g., finding1_..., finding2_...) so they’re easy to find later.
ImportantTask
Export each of your polished visualizations as a .png file into your plots folder. Include a screenshot showing the exported files in your plots folder.
4 Last word: Looking ahead to your final project
The plots you polished and exported in this lab are exactly what will go into the three finding boxes on your final project template. You may need to adjust these plots or saved images for your information sheet.