Skip to content
EMPIRICAL SCOTUS

It is not a 3-3-3 Supreme Court

Adam Feldman's Headshot
By
The Supreme Court of the United States is pictured in Washington, D.C.
(Adam Michael Szuscik via Unsplash)

Empirical SCOTUS is a recurring series by Adam Feldman that looks at Supreme Court data, primarily in the form of opinions and oral arguments, to provide insights into the justices’ decision making and what we can expect from the court in the future.

Please note that the views of outside contributors do not reflect the official opinions of SCOTUSblog or its staff.

People often talk about the Supreme Court as divided into groups, or “blocs,” such as liberals, conservatives, and sometimes “swing” justices who can tip the balance. But these labels don’t always capture the real story. Over the years, the way justices vote together has shifted in ways that often surprise even close Supreme Court-watchers.

Since Chief Justice John Roberts became the leader of the court in 2005, the justices have faced big debates, national crises, and personnel changes. During this time, many have wondered if there’s a new way the court is dividing itself – not just into two camps, but sometimes into three, with a “3-3-3” structure, posited by multiple court watchers. In this theory, you have three conservatives (Justices Clarence Thomas, Samuel Alito, and Neil Gorsuch), three liberals (Justices Sonia Sotomayor, Elena Kagan, and Ketanji Brown Jackson), and three so-called “institutionalist” justices (Roberts, along with Justices Brett Kavanaugh and Amy Coney Barrett) who sometimes form their own group in the middle.

This article uses new data and modern analysis to dig deeper into these coalition patterns across nearly 20 years. It tracks when the court sticks to classic liberal vs. conservative lines, when it splinters into different groups (including the 3-3-3 idea), and when justices act independently. The goal: to give a clearer, more realistic picture of how Supreme Court alliances actually work. Based on this data, it appears that the idea of a 3-3-3 court is largely untrue.

Methodology: how I analyzed Supreme Court coalitions

To track common voting blocs, I measured agreement in two ways. “Hard” agreement means justices not only vote the same way but also give the same legal reasoning – so their decisions and logic match. “Soft” agreement is more lenient: justices are counted as agreeing if they end up on the same side, even if their explanations differ, such as writing a separate concurring opinion.

After calculating these agreement rates for every pair of justices, I sorted justices into groups, based on who agreed most often. This technique starts by pairing the two justices who agree the most, then adds the next closest, and so on, eventually forming natural “blocs.”

To see how good these groupings are, I calculated so-called “silhouette scores” for each justice. Here’s how it works:

For every justice, I compare two things:

  • How similar is this justice’s voting to others in their own group? (the average “distance” or disagreement)
  • How similar is this justice’s voting to those in the nearest other group?

The silhouette score is calculated by taking the difference between these two averages and dividing by the larger one. If a justice votes much more like the members of their own group than any other, their score is close to 1. If they’re right on the border, it’s near 0. If they’re more similar to another group, the score is negative.

When most justices have high silhouette scores, it means the blocs are clear and strong; low or negative scores, on the other hand, mean group lines are fuzzy or that the coalition structure doesn’t fit well.

The early Roberts court (2005-2008): classic two-group splits

In the first five years under Roberts, I found that the Supreme Court usually broke into two predictable groups: conservatives (Roberts, Justices Antonin Scalia and Anthony M. Kennedy, along with Thomas and Alito) and liberals (Justices John Paul Stevens, David Souter, Ruth Bader Ginsburg, and Stephen Breyer). Conservatives agreed about 70% of the time (hard coding); liberals were even more unified, at 75%.

Silhouette scores between 0.60 and 0.70 confirmed that two blocs was the best way to describe the court.

Within each group, some justices served as anchors – Kennedy for conservatives, Stevens for liberals – almost always agreeing with their bloc. Others, like Thomas and Breyer, were more likely to split off, but these exceptions were rare. This does not mean that these justices were not ideologically aligned with their respective bloc members. In fact, it could mean they are more ideological than other conservative or liberal justices, voting on the extremes.

Although civil rights cases caused more separate opinions or occasional cross-bloc votes, the splits didn’t last. More than 80% of voting patterns from 2005 to 2009 could be explained by this simple two-group conservative-versus-liberal structure.

The first few terms with Obama’s appointees (2009-2011)

When Sotomayor and Kagan joined, the basic two-bloc pattern continued. The court’s conservatives (Roberts, Scalia, Kennedy, Thomas, Alito) stayed together for the most part, and the liberals (now Ginsburg, Breyer, Sotomayor, Kagan) remained effectively unified.

The data showed the conservatives agreed with each other about 78% of the time (soft coding), while the liberals reached up to 88%. Silhouette scores confirmed that dividing the court into two groups still fit the voting data best.

Breyer and Kagan were the most consistent “anchors” for the liberals, and Kennedy continued as the anchor for conservatives. Thomas was again the most frequent outlier on the right (perhaps because of his more extreme views), with Ginsburg sometimes diverging on the left.

The addition of the two new liberal justices did not disrupt the two-bloc structure as statistical checks consistently showed no evidence of a lasting third or “middle” group during this period.

Section 4: The middle Roberts court years (2012-2016)

During these years, the Supreme Court still mostly followed a two-bloc pattern: conservatives (Roberts, Scalia, Kennedy, Thomas, Alito) and liberals (Ginsburg, Breyer, Sotomayor, Kagan).

Agreement rates within the groups stayed high: conservatives averaged 80-86% (soft agreement) and liberals 88-93%. Silhouette scores and other statistical tests again confirmed that two groups fit the data best.

Roberts and Kennedy often anchored the conservatives, while Ginsburg was the steady anchor for the liberals. Thomas (on the right) and Breyer or Kagan (on the left) were the most likely to break from their group and vote in the middle, but not often enough to form a new bloc.

Civil rights cases – like the Fisher affirmative action decisions– produced some surprise votes and short-term coalitions, especially with Kennedy occasionally siding with liberals in these cases. Despite these anomalies, statistical analyses showed that more than 80% of voting still lined up with the standard two-group split.

Section 5: Trump’s appointments (2017-2020)

With the arrival of Gorsuch, Kavanaugh, and Barrett, the court’s conservative bloc expanded to six (Roberts, Thomas, Alito, Gorsuch, Kavanaugh, Barrett), while the liberal bloc was reduced to three (Sotomayor, Kagan, Breyer, and (until 2020) Ginsburg).

My analysis showed the two-group pattern remained strong. Conservatives agreed with each other about 79% (hard coding) to 85% (soft coding) of the time; liberals were even more united, up to 94% of the time. Silhouette scores were consistently high (0.60-0.68), again supporting the two-bloc model over any alternative.

Kavanaugh and Barrett became key anchors for the conservatives, and Sotomayor for the liberals. Thomas was the most likely conservative to diverge, while Kagan played that role for the liberals.

There were a few cases, especially regarding civil rights, in which conservatives split into smaller “mini-blocs,” but these shifts were temporary. Statistical tests showed no stable third group, and the overwhelming majority of votes fit the classic two-coalition structure.

Section 6: The Supreme Court in the 2021-2023 terms

In the most recent terms, the Supreme Court continued to be sharply divided into a 6-3 pattern: six conservatives (Roberts, Thomas, Alito, Gorsuch, Kavanaugh, Barrett) and three liberals (Sotomayor, Kagan, Jackson).

Within each group, justices agreed at very high rates – conservatives at 80-86% (depending on the measure), liberals even higher at 94-96%.

Kavanaugh was the main anchor for conservatives, rarely dissenting from his bloc, while Sotomayor anchored the liberals. Thomas was the most likely to split off among conservatives, and Jackson among liberals.

While civil rights cases sometimes caused the conservative bloc to fracture temporarily – such as in Students for Fair Admissions for soft agreement, where separate opinions and concurrences appeared – these mini-blocs were short-lived. There were two strong and separate groups with no lasting “middle” or swing bloc.

Section 7: The most recent term (2024)

In the 2024-25 term, the Supreme Court kept its sharply defined 6-3 split, with six conservatives (Roberts, Thomas, Alito, Gorsuch, Kavanaugh, Barrett) and three liberals (Sotomayor, Kagan, Jackson).

Within these groups, agreement remained very high: Conservatives averaged 79% (hard agreement) and 85% (soft), while liberals stayed at 93-94%. Silhouette analysis confirmed that dividing the court into two blocs still provided the clearest and most accurate picture.

Kavanaugh continued as the anchor for the conservatives, and Sotomayor filled that role for the liberals. Thomas was the most likely to disagree within the conservative group, and Jackson stood out as the liberal most willing to dissent or write separately.

Civil rights cases once again saw the greatest fractures – sometimes splitting conservatives into smaller subgroups  with three middle justices (Kavanaugh, Roberts, and Alito) separated from the more conservative justices – but these moments were isolated. (Two rare instances of this include 6-3 majorities with Alito, Gorsuch, and Thomas in dissent in Kennedy v. Braidwood Management and FCC v.  Consumers’ Research.)  Statistical methods showed that the overwhelming pattern remained two strong, stable coalitions with no evidence for a third group or lasting “middle” bloc.

In the end the data show that, despite changes in membership and major national debates, the Supreme Court has overwhelmingly operated as two stable, well-defined blocs for nearly 20 years. In other words, I found no strong evidence of a so-called “3-3-3” court, as suggested by some prominent analysts of the current court. Indeed, when temporary fractures or mini-coalitions appeared in certain cases, especially in those concerning civil rights, the overall data-driven pattern remains a sharply divided but cohesive two-bloc court.

A longer-form version of this piece can be found at Legalytics.

Recommended Citation: Adam Feldman, It is not a 3-3-3 Supreme Court, SCOTUSblog (Aug. 5, 2025, 10:27 AM), https://www.scotusblog.com/2025/08/it-is-not-a-3-3-3-supreme-court/