Sorry SCOTUS, if you don’t want to be perceived as partisan hacks, stop making decisions like partisan hacks — Part II

TL;DR

The partisanship and polarization of the US Supreme Court has been going on for years and is not only a recent phenomenon. However, it has been steadily increasing since the 1950’s and is now the worst it’s ever been.

Context

In my previous post on this topic, I found that SCOTUS justices’ decisions in the 2020 term were clustered in partisan groups. This reinforced findings of an earlier analysis from the 2011 term.

However, what if these particular years were outliers? How would such an analysis look across more years?

Data

I used the data from the Supreme Court Database to analyze data going back to 1946.

For each year, similar to the previous analysis, I calculated the percent agreement between justices on decisions with 6 or fewer concurring. Sample results for each decade from 1950 to 2020 are shown below.

Percent agreement between justices on decisions with 6 or fewer concurring, for 1950 to 2020. Red indicates a high level of agreement and blue indicates a low level of agreement.

We can clearly see that the level of clustering and polarization have increased over the years. Namely, in the early years, you don’t see the justices clustered in islands of red. As the years go by, the justices start to cluster and by 1980 we can see two main clusters of red (high agreement rate) emerging, with low agreement rates with justices outside one’s cluster (blue areas).

To quantify how polarization evolves over time, for each year we cluster the judges into two clusters and calculate the following two metrics:

• in-cluster agreement rate: the average of agreement rates for judges within each cluster.
• out-cluster agreement rate: the average of agreement rates for judges with justices in the other cluster.

The expectation is that if there is no polarization, in-cluster and out-cluster agreement rates will be close to each other. And if there is high polarization, in-cluster agreement rates will be much higher that out-cluster agreement rates.

The figure below shows these metrics from 1950 to 2020.

in-cluster and out-cluster agreement rates for each year from 1950 to 2020.

The above figure clearly shows:

• The in-cluster agreement rates are slowly increasing over time, from around 55% in 1950 to around 75% in 2020
• The out-cluster agreement rates are slowly decreasing over time, from around 45% in 1950 to around 25% in 2020
• The gap between how often a judge agreed with judges in his/her own cluster vs with judges in the opposite cluster has increased dramatically from around 10% in 1950 to around 50% in 2020.

The bottom line

SCOTUS justices are now more polarized and partisan than ever and are dramatically more likely to agree with people on their side vs people from the other side. Things were better in the middle of the 20th century and have been steadily getting worse since then. Partisanship is now at the worst it’s ever been.

Sources:

1. Harold J. Spaeth, Lee Epstein, et al. 2021 Supreme Court Database, Version 2021 Release 1. URL: http://Supremecourtdatabase.org