PART TWO: Data Manipulation and Visualization Basics using R

This skills workshop introduced R and the tidyverse to Master’s students at the University of Denver
R
tidyverse
ggplot
data-viz
domestic-politics
Author

Mickey Rafa

Published

May 15, 2018

This is a continuation of the “Data Manipulation and Visualization Basics using R” workshop, which was originally delivered in 2018. This post assumes that you know the content from Part 1.

Overview

This workshop session will help you familiarize with two of R’s most widely used packages, dplyr and ggplot2. These packages are part of what is called the tidyverse, which is a collection of packages that are designed to work together in R to simplify and data manipulation, discovery, visualization, analysis, and more.1

Each tile is a package that makes up the “tidyverse”

dplyr is designed to standardize the data management and manipulation process in R. dplyr has a small set of functions that help users navigate the most common challenges of data manipulation.2 Generally speaking, these include restructuring, filtering, grouping, and summarizing your data, as well as creating new variables for your analysis.

ggplot2 is a package for visualizing data. In this lesson, we will learn the building blocks for creating beautiful visualizations. We will learn how to produce histograms, scatter plots, and line plots, and we will visualize ungrouped and grouped data.

dplyr and ggplot2 are packages that supply a grammar for data manipulation and data visualization, respectively. This workshop will only scratch the surface of what is possible, but learning the fundamentals of these packages can empower you to do a ton of exciting things in R.

Part Two: Analysis of U.S. Supreme Court data

The U.S. Supreme Court

  1. Install the tidyverse and gapminder packages (install.packages())

  2. Load the tidyverse and gapminder packages (library() and require() effectively do the same thing, but library() provides more information that is occasionally useful)

library(tidyverse)      #for data management, manipulation, visualization, and exploratory data analysis
── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.4     ✔ readr     2.1.5
✔ forcats   1.0.0     ✔ stringr   1.5.1
✔ ggplot2   3.5.1     ✔ tibble    3.2.1
✔ lubridate 1.9.3     ✔ tidyr     1.3.1
✔ purrr     1.0.2     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
#for graphics styling
library(ggtext)         #for use of richtext in graphics
library(glue)           #to embed html in text
library(patchwork)      #for plot_annotation() function
  1. I am providing a dataset on all U.S. Supreme Court decisions since 1946, which are organized by case and justice vote. These data come from The Supreme Court Database at the Washington University in St. Louis Law School.3 I’d like for you to put together the skills that you’ve learned in this session to do some exploratory analysis of the data.
sup_court <- read.csv(url("https://github.com/mrafa3/korbel-skills-workshops/raw/master/supreme_court_1946_2018.csv"))
  1. Let’s look at the data together. Use the glimpse() function to peek at the sup_court object.
glimpse(sup_court)
Rows: 78,857
Columns: 10
$ caseId        <chr> "1946-001", "1946-001", "1946-001", "1946-001", "1946-00…
$ caseName      <chr> "HALLIBURTON OIL WELL CEMENTING CO. v. WALKER et al., DO…
$ term          <int> 1946, 1946, 1946, 1946, 1946, 1946, 1946, 1946, 1946, 19…
$ chief         <chr> "Vinson", "Vinson", "Vinson", "Vinson", "Vinson", "Vinso…
$ majVotes      <int> 8, 8, 8, 8, 8, 8, 8, 8, 8, 6, 6, 6, 6, 6, 6, 6, 6, 6, 5,…
$ minVotes      <int> 1, 1, 1, 1, 1, 1, 1, 1, 1, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4,…
$ justiceName   <chr> "HHBurton", "RHJackson", "WODouglas", "FFrankfurter", "S…
$ issueAreaName <chr> "Economic Activity", "Economic Activity", "Economic Acti…
$ directionName <chr> "conservative", "liberal", "liberal", "liberal", "libera…
$ majorityName  <chr> "dissent", "majority", "majority", "majority", "majority…

Explore the dataset on your own, and see what you can find! Try to ask a question of the dataset and come up with an answer. In case you want some examples to guide you, you might ask (the “Code” box is there if you are stuck, but try your best to do this on your own.)…

  • Which Chief Justice has presided over the most cases since 1946? Which has heard the fewest?
sup_court %>% 
  group_by(chief) %>% 
  summarise(count_cases = n()) %>% 
  arrange(desc(count_cases))
# A tibble: 5 × 2
  chief     count_cases
  <chr>           <int>
1 Burger          25076
2 Warren          19736
3 Rehnquist       18322
4 Roberts          8416
5 Vinson           7307
  • How many decisions has Justice Gorsuch made in his short tenure on the bench? How often has he cast a dissenting opinion? How often has he made decisions that would be considered conservative? What types of cases has he heard most often to this point? (hint: Gorsuch can be found using the justiceName variable with NMGorsuch.)
sup_court %>% 
  filter(justiceName == 'NMGorsuch') %>% 
  group_by(justiceName) %>% 
  summarise(count_cases = n())
# A tibble: 1 × 2
  justiceName count_cases
  <chr>             <int>
1 NMGorsuch            44
sup_court %>% 
  filter(justiceName == 'NMGorsuch') %>% 
  group_by(justiceName, majorityName) %>% 
  summarise(count_cases = n())
`summarise()` has grouped output by 'justiceName'. You can override using the
`.groups` argument.
# A tibble: 3 × 3
# Groups:   justiceName [1]
  justiceName majorityName count_cases
  <chr>       <chr>              <int>
1 NMGorsuch   dissent                3
2 NMGorsuch   majority              14
3 NMGorsuch   <NA>                  27
sup_court %>% 
  filter(justiceName == 'NMGorsuch') %>% 
  group_by(justiceName, directionName) %>% 
  summarise(count_cases = n())
`summarise()` has grouped output by 'justiceName'. You can override using the
`.groups` argument.
# A tibble: 3 × 3
# Groups:   justiceName [1]
  justiceName directionName count_cases
  <chr>       <chr>               <int>
1 NMGorsuch   conservative           14
2 NMGorsuch   liberal                 3
3 NMGorsuch   <NA>                   27
sup_court %>% 
  filter(justiceName == 'NMGorsuch') %>% 
  group_by(justiceName, issueAreaName) %>% 
  summarise(count_cases = n()) %>% 
  mutate(perc_cases = round(count_cases / sum(count_cases) * 100, 1)) %>% 
  arrange(desc(perc_cases))
`summarise()` has grouped output by 'justiceName'. You can override using the
`.groups` argument.
# A tibble: 7 × 4
# Groups:   justiceName [1]
  justiceName issueAreaName      count_cases perc_cases
  <chr>       <chr>                    <int>      <dbl>
1 NMGorsuch   Civil Rights                11       25  
2 NMGorsuch   Judicial Power              10       22.7
3 NMGorsuch   Economic Activity            7       15.9
4 NMGorsuch   Criminal Procedure           6       13.6
5 NMGorsuch   Due Process                  5       11.4
6 NMGorsuch   First Amendment              3        6.8
7 NMGorsuch   Federalism                   2        4.5
  • Justice Kennedy has a reputation as being the swing voter on the Supreme Court. How might you test this conventional wisdom with this dataset? (hint: you might consider using the majVotes and minVotes variables for this question)
sup_court %>% 
  filter(majVotes == 5,
         minVotes == 4) %>% 
  group_by(justiceName, majorityName) %>% 
  summarise(count_cases = n()) %>% 
  mutate(perc_cases = round(count_cases / sum(count_cases) * 100, 1)) %>% 
  arrange(desc(perc_cases))
`summarise()` has grouped output by 'justiceName'. You can override using the
`.groups` argument.
# A tibble: 74 × 4
# Groups:   justiceName [37]
   justiceName majorityName count_cases perc_cases
   <chr>       <chr>              <int>      <dbl>
 1 AJGoldberg  majority              30       78.9
 2 AMKennedy   majority             390       76  
 3 FMVinson    majority              81       71.1
 4 SFReed      majority              99       68.8
 5 LFPowell    majority             280       68.1
 6 NMGorsuch   majority               2       66.7
 7 SDOConnor   majority             352       66.2
 8 WEBurger    majority             260       65.5
 9 TCClark     majority             167       65.2
10 BRWhite     majority             431       64.5
# ℹ 64 more rows
  • Which justice has heard the most Supreme Court Cases since 1946? (hint: the justice is the primary unit of analysis)
sup_court %>% 
  group_by(justiceName) %>% 
  summarise(count_cases = n()) %>% 
  arrange(desc(count_cases))
# A tibble: 37 × 2
   justiceName count_cases
   <chr>             <int>
 1 WJBrennan          5325
 2 BRWhite            4943
 3 WHRehnquist        4529
 4 JPStevens          4260
 5 WODouglas          4001
 6 TMarshall          3879
 7 HABlackmun         3768
 8 PStewart           3590
 9 HLBlack            3302
10 SDOConnor          2910
# ℹ 27 more rows
  • What are the trends in the types of case heard by term? What is the trend in civil rights cases – compared with other issue areas – over time? (hint: BY term and issue area)
(sup_court_term_issue <- sup_court %>% 
  group_by(term, issueAreaName) %>% 
  summarise(count_cases = n()) %>% 
  arrange(term, desc(count_cases)))
`summarise()` has grouped output by 'term'. You can override using the
`.groups` argument.
# A tibble: 807 × 3
# Groups:   term [71]
    term issueAreaName        count_cases
   <int> <chr>                      <int>
 1  1946 Economic Activity            432
 2  1946 Judicial Power               189
 3  1946 Criminal Procedure           171
 4  1946 Unions                       126
 5  1946 Federal Taxation              90
 6  1946 First Amendment               72
 7  1946 Civil Rights                  63
 8  1946 Federalism                    54
 9  1946 Interstate Relations          36
10  1946 Due Process                   27
# ℹ 797 more rows
sup_court_term_issue %>% 
  filter(!is.na(issueAreaName)) %>% 
  ggplot(.) + 
  geom_line(aes(x=term,
                y=count_cases,
                color=issueAreaName))

To create a graphic, I’ll show how you can create a theme to add to your ggplot object.

my.theme <- theme(
  text = element_text(family = 'Arial'), 
  plot.title = element_textbox_simple(color="black", face="bold", size=18, hjust=0), 
  plot.subtitle = element_textbox_simple(color="black", size=12, hjust=0), 
  axis.text = element_text(size=15), 
  axis.ticks = element_line(color="gray90"),
  plot.caption = element_textbox_simple(color="black", size=12), 
  panel.background =  element_blank(), 
  panel.grid.major = element_line(color="gray90"),
  panel.border = element_blank(), 
  legend.position = 'none',
  strip.text = element_text(color="black", size=14, face='bold'),
  strip.background = element_blank())

And I’ll create objects for the title, subtitle, and caption text.

title_text <- glue("Count of <span style='color:darkgreen;font-face:bold'>civil rights</span> cases heard by the U.S. Supreme Court over time")
subtitle_text <- glue("<br>Count of cases")
caption_text <- glue("*The Supreme Court Database* at the Washington University in St. Louis Law School")
sup_court_term_issue %>% 
  filter(!is.na(issueAreaName)) %>% 
  mutate(CivilRights = ifelse(issueAreaName=='Civil Rights', 'Civil Rights', 'Other')) %>% 
  ggplot(.) + 
  geom_line(aes(x=term,
                y=count_cases,
                group=issueAreaName,
                color=CivilRights),
            linewidth=1.1) + 
  labs(x='',
       y='') + 
  scale_color_manual(values = c('Civil Rights' = 'darkgreen',
                                'Other' = 'lightgray')) +
  my.theme + 
  plot_annotation(
    title = title_text,
    subtitle = subtitle_text,
    caption = caption_text, 
    theme = theme(text = element_text(family = 'Arial'),
                  plot.title = element_textbox_simple(color="black", 
                                                      face="bold", size=20, hjust=0),
                  plot.subtitle = element_textbox_simple(color="black", 
                                                      size=15, hjust=0),
                  plot.caption = element_textbox_simple(color="black")))

  • Does the Court seem to have strong tendencies for votes based on the issue area? (Probably many ways that you could think about answering this one.)

  • Justice Scalia was known as a staunch conservative presence on the bench. Were there some issues that he leaned more centrist? (hint: group_by() and summarise() would come in handy here, yet again)

sup_court %>% 
  filter(justiceName == 'AScalia') %>% 
  group_by(justiceName, issueAreaName, directionName) %>% 
  summarise(count_cases = n()) %>% 
  mutate(perc_cases = round(count_cases / sum(count_cases) *100, 1)) %>% 
  filter(directionName == 'liberal') %>% 
  arrange(-perc_cases)
`summarise()` has grouped output by 'justiceName', 'issueAreaName'. You can
override using the `.groups` argument.
# A tibble: 12 × 5
# Groups:   justiceName, issueAreaName [12]
   justiceName issueAreaName      directionName count_cases perc_cases
   <chr>       <chr>              <chr>               <int>      <dbl>
 1 AScalia     Federal Taxation   liberal                47       69.1
 2 AScalia     Federalism         liberal                83       50.9
 3 AScalia     Economic Activity  liberal               234       45.5
 4 AScalia     Attorneys          liberal                19       39.6
 5 AScalia     Due Process        liberal                37       34.9
 6 AScalia     Judicial Power     liberal               126       32.7
 7 AScalia     Unions             liberal                22       30.6
 8 AScalia     First Amendment    liberal                55       30.4
 9 AScalia     Civil Rights       liberal               134       29.1
10 AScalia     Criminal Procedure liberal               207       27.8
11 AScalia     Privacy            liberal                13       22.4
12 AScalia     Miscellaneous      liberal                 2       13.3
  • To truly answer a question like the previous one, you would want to take into account the directionality of all voting justices by issue area. For instance, if you explored the previous question, you found that Scalia voted “liberal” on almost 70% of Federal Taxation cases. However, if the Court has a tendency to vote liberal on this issue, then Scalia might not be as “liberal” on this issue as it might seem. This question is a bit more complicated, but does Scalia lean liberal based on the Court’s tendencies in Federal Taxation Cases? (This question still uses dplyr – and if you’d like, ggplot2 – but with a few bells and whistles.)
(sup_court_fedtax <- sup_court %>% 
  filter(term > 1980, issueAreaName == 'Federal Taxation') %>% 
  group_by(justiceName, issueAreaName, directionName) %>% 
  summarise(count_cases = n(), .groups = "drop_last") %>% 
  mutate(perc_cases = count_cases / sum(count_cases)) %>% 
  filter(directionName == 'liberal') %>% 
  ungroup() %>% 
  left_join(x=.,
            y=baseline_vote,
            by='issueAreaName'))
# A tibble: 19 × 6
   justiceName issueAreaName    directionName count_cases perc_cases
   <chr>       <chr>            <chr>               <int>      <dbl>
 1 AMKennedy   Federal Taxation liberal                47      0.746
 2 AScalia     Federal Taxation liberal                47      0.691
 3 BRWhite     Federal Taxation liberal                47      0.922
 4 CThomas     Federal Taxation liberal                27      0.574
 5 DHSouter    Federal Taxation liberal                29      0.707
 6 EKagan      Federal Taxation liberal                 4      0.5  
 7 HABlackmun  Federal Taxation liberal                41      0.774
 8 JGRoberts   Federal Taxation liberal                10      0.833
 9 JPStevens   Federal Taxation liberal                48      0.6  
10 LFPowell    Federal Taxation liberal                16      0.64 
11 RBGinsburg  Federal Taxation liberal                29      0.784
12 SAAlito     Federal Taxation liberal                 9      0.75 
13 SDOConnor   Federal Taxation liberal                44      0.579
14 SGBreyer    Federal Taxation liberal                27      0.771
15 SSotomayor  Federal Taxation liberal                 7      0.875
16 TMarshall   Federal Taxation liberal                28      0.683
17 WEBurger    Federal Taxation liberal                16      0.8  
18 WHRehnquist Federal Taxation liberal                55      0.724
19 WJBrennan   Federal Taxation liberal                29      0.744
# ℹ 1 more variable: baseline_issue_vote <dbl>
library(scales)                 #for changing axis label to include '%'

Attaching package: 'scales'
The following object is masked from 'package:purrr':

    discard
The following object is masked from 'package:readr':

    col_factor
sup_court_fedtax %>% 
  filter(count_cases >=10) %>% 
  mutate(fedtax_lean = ifelse(perc_cases > baseline_issue_vote, 'Liberal', 'Conservative')) %>% 
  ggplot(.) + 
  geom_point(aes(x=perc_cases,
                 y=reorder(justiceName, perc_cases),
                 color=fedtax_lean), size=5) + 
  geom_vline(xintercept = sup_court_fedtax$baseline_issue_vote) + 
  annotate(geom='text', x=.48, y='JGRoberts', 
           label=paste('Baseline vote for Federal Taxation cases is ', 
                       round(sup_court_fedtax$baseline_issue_vote[1] * 100), '%', sep = ''),
           size=5) + 
  scale_x_continuous(labels = percent,
                     limits = c(0, 1)) + 
  ggtitle('The Court leans liberal on Federal Taxation cases',
          subtitle = 'Percent of Federal Taxation cases with liberal decision by Justice since 1980') + 
  labs(x='Percent of cases coded as a liberal decision\n',
       y='',
       color='',
       caption='Source: Washington University of St. Louis\nThe Supreme Court Database\nExcluded Justices with less than 10 Federal Taxation cases') + 
  theme_minimal() + 
  scale_color_manual(values = c('Liberal' = 'blue',
                                'Conservative' = 'red')) + 
theme(legend.position = 'top',
        text = element_text(size=15),
        plot.title = element_text(size=18, face='bold'),

        plot.subtitle = element_text(size=16),
        strip.text = element_text(size=15, face='bold'))

Footnotes

  1. https://www.tidyverse.org/packages/↩︎

  2. https://cran.r-project.org/web/packages/dplyr/vignettes/dplyr.html↩︎

  3. http://scdb.wustl.edu/data.php↩︎