Visualizing Supreme Court Justice Agreement, Part I

For a visualization project, I wanted to show how often the justices of the Supreme Court of the United States (SCOTUS) agree with each other. As someone who has limited knowledge of the Supreme Court outside of what I read from various mainstream news sources I thought this would be a good way to learn more about it.

Based on the idea that the Supreme Court was politicized, I expected to see high agreement between individual justices of similar ideologies (conservative/liberal) and low agreement between those of opposite ideologies. I also wanted to look a little deeper to see how often a justice agreed with two or more justices of the similar/opposite ideology. After looking at some of the sources I used to obtain data, I thought it might also be helpful to include the Martin-Quinn scores for each Justice. These scores quantify where a justice falls along a liberal-conservative continuum. Seeing these scores may help understand the agreement rates.

This project only focuses on the past five SCOTUS terms (2017–2021), which reflects data used from Ballotpedia (my initial data source), and is done in two parts. Part I of this project will focus on the Martin-Quinn scores and single justice agreements and Part II will focus on multiple justice agreement.

All visualizations were created using the ggplot package in R, and code for preparing the data and generating the visualizations along with all visualizations can be found at the following Github repository.

Martin-Quinn Scores

Visualizations for the Martin-Quinn scores were created using the justices.csv file from the Measures section of the Martin-Quinn Scores webpage. This data set contains scores for each Justice beginning in the 1937 term through the 2021 term. Their methodology for determining scores can be found here. As recommended by the website, the posterior mean scores were used for these visualizations.

The first visualization I created is based on one seen on each Justice’s Ballotpedia page. I decided to put the Justice names in the middle and arranged them from most liberal to most conservative and placed the scores for each Justice at the end of their bar. Similar the to the Ballotpedia visualizations, traditional blue/red colors for the liberal/conservative justices. I would’ve liked to use a gradient for colors, with the darker blue/red representing increasing liberal/conservative ideology, but without values for the limits on liberal/conservative scores it wouldn’t be possible to compare the visualizations across years. (For reference, the most extreme liberal score is -7.929 (Douglas, 1975) and the most extreme conservative score is 4.511 (Rehnquist, 1979)).

The visualization below is for the 2021 term.

In addition to the scores for the most recent term, I thought it would also be informative to see how Justices’ ideologies may have evolved across the past five terms, to see the variability in their scores. My first thought was a line graph, but I wanted to try something different before doing a line plot. To do this I did a scatter plot with the Justices and their Martin-Quinn Scored for the 2017–2022 terms, with the Justices arranged from most liberal to most conservative. Each point was also colored by term. Here is the plot:

The original plot used the default ggplot colors which made it confusing. I wanted to stay with the red/blue used previously, but since the color represents each term it didn’t make sense. Thus, I went with the grayscale, with the lightest representing 2017 and the darkest representing 2021.

To create the line chart showing the scores for each Justice, I used the red/blue colors as in previous visualizations. The first and last scores for each Justice in this time frame were added to help see the change over the time period. To label each line with the Justice name, I used the ggtextpath package.

(Since he only had a score for 2017, I excluded Justice Kennedy from this visualization).

It appears each Justice is consistent, with most Justices becoming slightly more liberal/conservative over the past five terms (away from the center). This visualization also shows the split within the ideologies. In 2017, the only cluster appears between Justices Kagan and Breyer (and Justices Roberts and Kennedy, who had a score of .402), while in 2021 there appear to be 4 clusters — Roberts/Gorsuch/Kavanaugh/Barrett, Alito/Thomas, Kagan/Breyer and Sotomayor. I also found it interesting that the majority of the conservative Justices are closer to the center than the liberal Justices.

If we go further back to the 2012 term:

It’s interesting to see how consistent some Justices are (e.g. Thomas), the changes in Justice Sotomayor’s scores, and how Justice Kennedy was the swing vote.

Single Justice Agreement

For the single justice agreement visualizations I scraped data from each Justice’s Ballotpedia page. The tables on these pages have the agreement rates for each particular Justice with the others beginning with the 2017 term. Heat maps of the agreement rates for each term were created for each Justice, faceted by ideology (no visualizations were created for Justice Kennedy since the 2017 term was his last, but he is included in the visualizations).

The same color palette (red for conservative and blue for liberal) was used for these visualizations. In the case, a gradient for the colors were used — redder indicating higher agreement with conservative Justices and bluer indicating higher agreement with liberal Justices. Below is the visualization for Justice Thomas:

It should not be a surprise in the split between his agreement rates with conservative and liberal Justices based on his most recent Martin-Quinn score (3.05).

This can be contrasted with Justice Sotomayor, who is the most liberal Justice according to her most recent Martin-Quinn score (-4.14).

With Justice Sotomayor we see much higher agreement with her ideological colleagues compared Justice Thomas’ agreement with his ideological colleagues.

To compare all the Justices, non-faceted heat maps were created and combined using the patchwork package with conservatives in the top and bottom rows and liberals in the middle row (may need to zoom in to see the values):

Here we can see the liberal Justices with high agreement rates with one another (82% — 96%), and the conservative Justices have a wider range of agreement rates with one another (65% — 100%). It should not be a surprise that the Justices with the most extreme Martin-Quinn scores (Sotomayor, Thomas, Alito) have the lowest agreement rates with Justices of other ideologies. Within ideologies, I expected to see Justices Kagan and Breyer to have higher agreement rates, similar to the rates of Justices Roberts and Kavanaugh (100% agreement in the 2021 term). Looking across the terms, the Justices of opposing ideologies agreed less in the 2021 term compared to the other terms.

Summing Up

Creating visualizations always involves design choices. For colors, initially I tried some different color palettes. While I liked the way the visualizations looked, after examining the visualizations I thought that the traditional red/blue for conservative/liberal better communicated the results. I also tried to do different types of visualizations, but felt that the line plots and heat maps with the red/blue colors were most effective. One other thought I had was including pictures/headshots of the Justices, but after mentally visualizing how it would look decided against it.

Overall, I was surprised at how often Justices of opposing ideologies do agree with one another. For the most part Justices tended to agree with one another on at least 50% of cases in most of the terms. These rates include the unanimous decisions, so it will be interesting to see the agreement rates using only the non-unanimous decisions (in Part II of this project).