How connected are the voting behaviors between a Supreme Court justices?


Visualization

INTRODUCTION

Since 1946, the Supreme Court database has been tracking all the legal details and nuances related to each case brought forth to the Supreme Court. Details, such as cases organized by docket or cases organized by issue / legal provision, are meticulously tracked through the database’s code book and each value is encoded to match the actual outcomes of each case’s decision.

For this report, I wanted to review a dataset on case-centered data and then analyze the cases organized by the Supreme Court Citation. My chosen dataset focused on consolidated cases (those with multiple dockets) or cases with multiple issues or legal provisions and their outcomes per justice. This datasets shows various fields related to how the justice’s vote differed from the majority, the case’s name, and much more. I wanted to see a visual overview of how connected each justice’s voting behavior is in reference to their chief’s final vote per case. Additionally I wanted to see how connected other Supreme Court chiefs’ voting behaviors were between each other as well.

DISCUSSION

Since this dataset’s visualization analyzes social connectedness, Moreno’s sociograms were the first designs that popped into my mind. His visual representations of elementary school children’s relationships within each grade level’s class were simple in their visual communication; a reader can easily identify the edges within the class structure. Then the visualizations’ minimal aesthetic doesn’t distract the reader’s attention, rather it emphasizes each child’s position in the network using color, label sizing, and network structure. One leaves his design feeling more informed about his research and receptive to the social workings of each class.

Moreno’s Sociograms

As I used Moreno’s network design and aesthetic as inspirations, I was interested in exploring experimentations with color, such gradients and color palettes. My chosen dataset may relate to politics or patriotic tones, but I wanted to represent the data’s connections in a more interactive and playful manner, similar to the below visualizations. Each one plays with color without distracting the viewer’s attention from the underlying purpose of what it’s communicating.

MATERIALS

The dataset, Cases Organized by Supreme Court Citations, was used and derived from the Supreme Court database, managed by Washington University Law. This dataset shows the voting behaviors from 1946 to 2019 for each justice during that time period and much more. The file was a large csv with more than 9,000 records.


The visualizations were created through Gephi, an open-source and multiplatform software used to visualize various networks within a dataset. Then Googlesheets was used to manipulate and edit the dataset’s file, so it is more readable to import into Gephi.

METHODS

  1. Find a clean dataset in csv format and manipulate the data to be a readable format.

While the dataset was already a csv file, the file needed to be formatted and manipulated in order to make it readable for the specific visualization I was interested in exploring. It was a mostly clean dataset, but it had extraneous data categories and fields, such as date range and docket number. 

Original CSV File


I wasn’t interested in exploring those categories, so I had to delete and rearrange the columns to visualize the network between the justices and their voting behaviors. Additionally I had to reconfigure the dataset’s weight between each justice’s voting behavior; I decided to create a simple weighted scale (-1, 1, 2) depending on the justice’s vote with the majority per case. Previously the scale was simple, but it had null values for some justices’ votes and I needed to indicate their in-between state without interrupting my visualization.

Edited Dataset File

2. Experiment with network designs and statistical values to highlight weighted degrees in the network.

3. Explore gradient color palettes, label sizing, and font colors.

RESULTS

After some trial and error, I finally was able to create a final design that experimented with the right amount of color, network design, and label sizing. While this network was undirected, I wanted to explore network designs that highlighted the interconnectedness between the voting behaviors of justices. Compared to ForceAtlas2, Fruchterman Reingold highlighted the fluidity of this network and also clearly distinguished the weight of each justice’s voting connection to their chief.

Then on top of that, I wanted to highlight the statistical findings, such as the range of weighted degrees, modularity class, and harmonic closeness centrality. All of these statistical variables highlight the ways of measuring connectedness between the justices, but they visually imply the significance of certain relationships when it comes to voting.

REFLECTION

While it was initially a hassle to visualize my exploration, Gephi was such a useful tool for illustrating network data and it was accessible after a couple trial sessions. In regards to my final visualizations, I would definitely dive further into highlighting other statistical significances and/or relationships within the network, such voting patterns on case topics and the strength of edges between justices of the same political leanings. This information would be valuable when analyzing the initial insights further and would provide better support about the trends I noticed in my final visualizations.