Measuring the Justices: An Portrait of Ideology on the Current Supreme Court

The justices' ideological preferences are a contested topic. This article discusses the costs and benefits of measuring the justices' ideologies at all and presents a novel, composite ideology measure

Every term, observers describe the Supreme Court using ideological shorthand: a liberal bloc of three, a conservative supermajority of six, and a handful of justices who sometimes break from their expected coalition.

For example from Josh Gerstein on Politico:

“Most of the rhetorical clashes pitted the court’s conservative and liberal wings against each other in politically polarized cases. But not all of the spats fell squarely along ideological lines.”

Or more recently from Jason Willick at the Washington Post:

“Justice Neil M. Gorsuch, Trump’s first Supreme Court nominee, agreed with the liberals. But he took them to task for their pivot from the Biden years. ‘Dissenting in past major questions cases,’ Gorsuch wrote, the court’s liberal members ‘have argued that broad statutory language granting powers to executive officials should be read for all it is worth.’”

These labels are intuitive, but they are rarely grounded in systematic measurement. Political scientists have spent decades developing tools to estimate judicial ideology in more precise ways — tools that are imperfect, but far more transparent about what they measure and what they do not than casual categorization.

This article applies three of those tools to the current Roberts Court, drawing on data from OT2022 through OT2024. The aim is not to reduce judicial philosophy to a single number but rather to ask a narrower and more tractable empirical question: using established vote-based measures can we measure ideological preference without simply equating it with dissent frequency alone?

Two Structural Caveats

Any empirical effort to measure Supreme Court ideology must first confront two structural limitations that bear on how the results should be interpreted.

Case Selection. The Supreme Court’s docket is not a random sample of legal disputes. Each term’s argued merits cases — roughly 60 to 70 per year — are curated through the Rule of Four (at least four justices must vote to hear a case for the case to be granted to the oral argument docket), strategic petitioning, and the Court’s own institutional judgment. We observe how justices vote in the cases the Court chooses to decide. We do not observe the internal conference discussions that determine which petitions are granted. Ideology influences both the merits vote and the agenda-setting process that precedes it. This analysis captures only the former. It measures ideological behavior within the cases the Court decided, not the broader universe of cases it declined to hear.

Minority Inflation. The second challenge is structural rather than methodological. On a Court with a stable 6–3 conservative majority, liberal justices will dissent more often simply because they lose more often. If dissent frequency is treated as the primary definition of ideology, whichever bloc is in the minority will appear more extreme — not because of anything intrinsic to those justices’ views, but because of arithmetic. The same dynamic would apply in reverse on a hypothetical 6–3 liberal Court: Alito, Thomas, and Gorsuch would top the dissent charts, and a dissent-based measure would label them the most ideological justices almost by default.

A prior hurdle has been incorporating behavior from the shadow docket. Vote behavior from the shadow docket was added to construct the final composite measure.

To avoid that distortion, this analysis does not treat dissent counts as a proxy for ideological intensity but rather focuses on directional consistency, spatial distance from the Court’s center, and voting behavior in ideologically divided cases — three measures that treat left and right symmetrically.

The Three Measures

The analysis relies on three vote-based metrics. Each captures a different dimension of ideological preference, and none alone tells the full story.

1) Martin–Quinn Scores: Spatial Distance from the Court’s Median

Martin–Quinn (MQ) scores are the most widely accepted academic measure of Supreme Court ideology. Developed by political scientists Andrew Martin and Kevin Quinn, these scores estimate each justice’s ideological position by modeling voting coalitions in non-unanimous cases. Justices who consistently vote together are placed closer together on a one-dimensional left-right scale; those who rarely align are placed further apart. The model is updated each term as new votes accumulate, so scores shift over time as a justice’s record grows.

For this analysis, the relevant quantity is each justice’s distance from the Court’s median MQ score during OT2022–OT2024 — that is, how far a justice sits from the Court’s internal ideological midpoint. This is a symmetric measure: a justice sitting far to the left of the median is treated as equally extreme as one sitting the same distance to the right.

As the figure shows, Justice Sotomayor sits furthest from the Court’s median on this measure, at a distance of 4.648 MQ units. Justice Jackson is second at 3.266 units. Justice Thomas is the most distant conservative at 2.613 units, followed by Justice Kagan (2.301) and Justice Alito (2.039). Justice Gorsuch sits at a moderate 0.625 units from the median. Justices Roberts (0.131), Kavanaugh (0.088), and Barrett (0.058) are clustered tightly around the Court’s center.

One feature of the MQ results is worth noting directly: the three liberal justices — Sotomayor, Jackson, and Kagan — account for three of the five largest distances from median. The conservative pole, by contrast, is spread more broadly, with Thomas and Alito registering significant distances but Gorsuch, Roberts, Kavanaugh, and Barrett sitting far closer to center. This asymmetry is a product of the current Court’s coalition structure and reflects the minority inflation dynamic described above: because the three liberal justices are consistently on the losing side of non-unanimous cases, the MQ model infers that their ideological positions must be substantially further from the winning coalition’s center.

2) SCDB Directional Extremity: Left-Right Consistency

The Supreme Court Database (SCDB), codes each case outcome as liberal or conservative according to issue-specific doctrinal criteria developed over decades by political scientists. The coding is calibrated to what liberal and conservative outcomes have meant in different areas of law. Criminal procedure cases, civil liberties claims, and economic regulation cases each have their own directional logic.

From those codings, one can compute each justice’s percentage of conservative votes over a given period. To convert that into a symmetric measure of extremity, this analysis calculates the absolute distance from a 50/50 liberal-conservative split. A justice who votes conservatively 80% of the time and a justice who votes conservatively 20% of the time are treated as equally distant from the midpoint — both are 30 percentage points from center.

This figure shows that Justice Alito leads the Court in SCDB directional extremity at 34.3 percentage points from center, followed by Justice Sotomayor (30.9) and Justice Jackson (29.8). Justices Thomas and Kagan are tied at 27.5 points. The middle of the Court — Kavanaugh (14.0), Gorsuch (12.9), Roberts (9.6), and Barrett (7.3) — is considerably closer to the 50/50 midpoint. Barrett’s score of 7.3 is particularly notable: she is the justice nearest to directional neutrality on the SCDB measure, despite being commonly grouped with the Court’s more pronounced conservatives.

The SCDB results diverge somewhat from the MQ picture. Alito, rather than Sotomayor, leads on SCDB directional extremity. The difference reflects what each measure captures: MQ models spatial position within the Court’s coalition structure and is sensitive to how often a justice forms small dissenting groups, while SCDB tracks the liberal-conservative direction of a justice’s votes across all cases. A justice who frequently dissents alone from both blocs — as Thomas— may appear more extreme on MQ than on SCDB.

3) Bloc-Split Alignment Extremity: Coalitional Behavior in Divided Cases

The third measure focuses specifically on behavior in ideologically divided cases. Using OT2022–OT2024 non-unanimous merits decisions, the justices cluster into two consistent voting blocs: a liberal bloc of Sotomayor, Kagan, and Jackson, and a conservative bloc of Roberts, Thomas, Alito, Gorsuch, Kavanaugh, and Barrett. In cases where those blocs take opposite positions, this measure asks how consistently each justice aligns with one side rather than the other.

The metric is again symmetric: a justice who aligns with the conservative bloc nearly 100% of the time in split cases scores the same distance from center as one who aligns with the liberal bloc nearly 0% of the time. The question is not which side a justice is on, but how consistently and rigidly that alignment holds in contested cases.

The figure shows Justice Alito with the highest bloc-split extremity score (0.479), followed by Sotomayor (0.458), Thomas (0.438), and Kavanaugh (0.417). Jackson and Barrett tie at 0.396. Kagan (0.354) and Roberts (0.333) cluster in the middle of the pack on this measure. Justice Gorsuch’s score of 0.250 is the lowest on the Court — a meaningful result that confirms his cross-coalition voting behavior on issues like the Fourth Amendment, digital privacy, and criminal procedure. He votes conservatively in the aggregate, but his alignment with the conservative bloc in ideologically split cases is more variable than any other justice’s.

The Composite: Convergence Across Measures

Each of the three primary measures — Martin–Quinn distance from the Court’s median, SCDB directional extremity, and bloc-split alignment extremity — is first converted into a rank. For each measure, the justice who registers as most extreme receives the highest rank; the justice who registers as least extreme receives the lowest. Rather than normalizing by raw scale values, which can exaggerate small differences at the margins, this rank-based approach ensures that no single metric dominates the result.

To incorporate shadow-docket behavior without reintroducing minority inflation, a fourth component is added as a light layer. Shadow-docket dissents from OT2022–OT2024 are ranked in the same way as the other measures, but weighted modestly. The final composite is therefore calculated as a weighted average of ranks:

• 30% SCDB directional extremity

• 30% MQ distance from the median

• 30% bloc-split alignment extremity

• 10% shadow-docket dissent rank

This weighting preserves the primacy of vote-based spatial and directional measures while allowing shadow-docket opposition to register as a supplementary indicator of ideological intensity. The addition of the shadow component functions as a secondary signal of oppositional consistency in a setting not often incorporated into ideologic measures. The justices dissented in the following amounts in shadow docket decisions from OT2022 through OT2024: Jackson 41, Gorsuch 36, Sotomayor 31, Thomas 26, Kagan 21, Alito 19, Kavanaugh 2, Barrett 1, Roberts 0.

The resulting figure presents the convergence ordering. Justice Sotomayor remains at the top of the composite, followed closely by Justice Alito. Justices Thomas and Jackson are close to one another with Jackson slightly ahead. Justice Kagan occupies the middle tier, followed by Justice Kavanaugh. Justice Gorsuch remains below that cluster but his directional conservatism is visible even though his coalition behavior is less rigid than that of Justices Alito and Thomas. Chief Justice Roberts and Justice Barrett anchor the lower end of the ranking, reflecting comparatively lower directional extremity and greater cross-bloc movement.

The Thomas–Jackson proximity is particularly instructive. The two justices achieve similar composite positions through different patterns. Jackson scores higher on spatial distance from the Court’s median — a function of sitting in a three-justice minority on a 6–3 Court, which the MQ model registers as ideological distance. Thomas scores higher on directional extremity and bloc-split alignment, reflecting sustained conservative voting and consistent coalition loyalty. The shadow-docket component marginally reinforces Jackson’s oppositional profile without overwhelming the other dimensions.

The composite highlights convergence: the justices who repeatedly appear near the top across spatial, directional, coalition, and shadow-docket dimensions are those most consistently identified as ideologically intense. Reading the individual measures alongside the composite makes clear that ideological strength can manifest in different ways — and that no single metric fully captures it.

What the Findings Show

Across all measures, the Court divides into identifiable tiers. At the poles, Sotomayor and Alito register the highest levels of ideological consistency across all three dimensions. Thomas and Jackson form a second tier, followed by Kagan, Kavanaugh, and Gorsuch. Finally Roberts and Barrett both cluster close to the Court’s operational center on most measures.

The most important finding for purposes of the minority inflation problem is that once dissent frequency is set aside and replaced by directional, spatial, and coalitional measures, ideological intensity appears on both sides of the Court. The current Court’s conservative members are not uniformly closer to center than their liberal counterparts. On the SCDB directional measure, Alito registers higher extremity than Sotomayor. On the bloc-split measure, Alito again leads, with Thomas close behind. The composite shows Sotomayor slightly ahead of Alito, but the margin is modest. The symmetry across measures is a meaningful empirical finding: the two poles of the Court are roughly equivalent in ideological distance, even though one pole dissents far more often.

The picture would look considerably different on a hypothetical Court with a different ideological majority. A 6–3 liberal majority would likely produce MQ distances showing Alito, Thomas, and Gorsuch far from the new median — not because those justices changed, but because the measuring post would have moved. The liberal justices who currently register the highest MQ distances would, under that configuration, cluster closer to the majority and appear more moderate by spatial measures. Bloc-split and SCDB extremity would likely remain elevated at both ends, but the ordering might shift. That hypothetical underscores a key limitation of all three measures: they describe a Court relative to its own composition. Ideological extremity is always a relational concept, not an intrinsic property.

Concluding Thoughts

Several dimensions of judicial behavior resist quantification by these measures. Ideology may influence case selection at the cert stage, but that dimension is unobservable here. Voting behavior may reflect strategic opinion assignment, coalition management, or compromise rather than sincere ideological preference. SCDB coding captures doctrinal direction but cannot encode the full complexity of constitutional reasoning, and some cases are coded in ways that are counterintuitive relative to contemporary partisan valence.

Martin–Quinn scores are relative to a Court’s composition; a shift in membership shifts the median and changes every justice’s distance calculation. Even the bloc-split measure depends on how coalitions form in a given era — a different Court alignment would produce different bloc structures and different behavioral scores. These are reasons to interpret the measures carefully and to use multiple measures rather than relying on any single one.

These tools also do not measure moral philosophy, interpretive methodology, judicial craftsmanship, or institutional influence. They measure voting behavior along a single ideological dimension and coalition behavior in divided cases. That is a meaningful slice of judicial behavior — but not the whole of it.

Of course there are articles that criticize using politically laden terms like “conservative” and “liberal” to describe Supreme Court justices at all, but this article is not designed to get into the normative weeds of whether these terms are appropriate as baseline descriptors.

When spatial distance (MQ), directional voting consistency (SCDB), and coalition alignment (bloc-split extremity) are examined together, a coherent portrait emerges. The current Court has identifiable ideological poles on both ends, not a pronounced conservative wing and a neutral liberal minority. Sotomayor and Alito anchor the extremes, with Thomas and Jackson forming a closely grouped second tier. Roberts, Barrett, and — by some measures — Kavanaugh occupy the Court’s operational center.

The convergence across three independent measures strengthens the inference. Justices who score high across spatial, directional, and behavioral dimensions are not an artifact of any single modeling choice. And justices who score low across all three — Barrett, Roberts — are not moderate simply because they are conservative; they are measurably closer to the Court’s current center than their colleagues at either end of the spectrum.

What these measures cannot tell us is whether any of this would look the same under a different Court. The median is not fixed; it is a function of membership. The ideological extremity this analysis measures is extremity relative to the Roberts Court of OT2022–OT2024. A Court with a different center would produce a different map — and a different set of justices at the poles.