What do we know about how the Supreme Court votes?
Ideology and preferences are predictive
The current saga around Supreme Court nominations, stretching back to Merrick Garland’s nomination, the subsequent breakdown in the process, and recent nomination of Brett Kavaunagh has highlighted the need for reforms through adding term limits.
With all this drama, it is good to step back and examine what we know about Supreme Court ideology and how it votes.
Modeling justice ideology is a heavily researched topic in the political science literature. The current methodology uses what are known as Martin-Quinn scores developed by political scientists Andrew Martin and Kevin Quinn.
Their paper runs through the technique in explicit detail, but at a high level the way it works is to model judicial ideology on a single dimension of conservative to liberal in a time varying model.
They predict whether a justice will vote to affirm or reverse a particular case as a function of case specific effects (the type of issue at question, etc.) and the justice’s ideology at that point in time. This means they allow a justice’s ideology to vary over time in a way that takes into account the justice’s ideology in the previous term.
From their model judicial ideology is estimated for each justice in each term they serve based on a combination of their votes in that term and in all previous terms the justice served. A positive Martin-Quinn score denotes a conservative justice and a negative Martin-Quinn score denotes a liberal justice, with zero being neutral.
Combining their data with the vote history and case details of the Supreme Court from 1937 to today can teach us a lot about the court, how it actually votes, and how judicial ideology affects outcomes.
Judicial Drift
If estimates of judicial ideology are allowed to change over time, do they actually and does judicial ideology tend to drift in a particular direction?
It turns that justice ideology does drift over time and in a more liberal direction. The graph below plots the Martin-Quinn score for each justice in each term they served by term order.
We can see there appears to be a systematic liberal drift i.e. in a negative direction.
We can explicitly quantify this leftward drift using a regression model to predict change in a justices ideology as a function of term order i.e. number of years the justice has been on the bench.
There is a significant and negative relationship in how a justice’s ideology evolves over time although the effect size is not large.
Except on the margin, it appears conservative justices will stay conservative and liberal justices will remain liberal, regardless of how long they serve.
Court Ideology Over Time
If judicial ideology can change over time (albeit in small increments), then how do we think about the ideological leaning of the court in a particular term?
To answer this we need to understand how the court decides on issues. In any standard hearing of a case, there are nine justices who vote to affirm or reverse the lower court’s ruling on the given case. The outcome of the court will be an affirmation if a majority of justices affirm and a reversal if a majority of justices vote as such.
In a conceptual model of voting, a judge’s decision can be thought to be a function of just her ideology to a first order approximation. In that case we can say that liberal judges will vote for the liberal side and conservative judges will vote for the conservative side.
Under this framework the median justice is the one who will decide the outcome of any given case and whether it is affirmed or reversed because judges more liberal than the median justice will vote liberal and judges more conservative than the median justice will vote conservative. The intuition is the same as that of swing voters deciding elections.
Hence, we can summarize the court’s ideology over time using the Martin-Quinn score of the median justice.
Other related metrics of interest are the difference between the most conservative and liberal justices over time to understand the level of polarization in the court i.e. how wide is the ideological divide amongst justices in a term and level of ideological purity in the court as shown by the distribution of scores.
We show all of these metrics in the plots below.
For most of the Supreme Court’s history, it has trended conservative, except for the 1960s where we saw the bulk of civil rights legislation enacted and more recently where we saw rulings of the Supreme Court on things like gay marriage.
In terms of polarization, unlike congress for example, the course has not become more extreme over time as the range of Martin-Quinn scores on the court has stayed fairly constant since the 1980s; however, when we look at the range of conservative vs. liberal scores, we do see a tendency since the late 1980s for more extreme conservative justices compared to liberal justices. This is largely due to Clarence Thomas who is currently the most conservative justice on the court and has been consistently since his appointment in 1991.
What Predicts Voting Behavior
Let’s take a look at whether MQ Scores can predict whether an individual justice will vote Liberal or Conservative on a particular case and whether that tendency is affected by the popularity of the case. We will proxy for case popularity using whether it was mentioned in the NYTimes or not.
Below are regression results showing whether a justice will vote liberal or not on a particular case as a function of the justices MQ score or their lagged MQ score (from the previous year). We use lagged MQ score to test the robustness of this effect. We will also include whether the case was in the NYTimes and the interaction of MQ score and case salience.
This data comes from the Supreme Court Database assembled by Washing University in St. Louis. The table below shows the results.
The takeaway from these regressions is most easily seen in the last column. Justices with a positive MQ score in the previous year (so they are conservative) are almost 10% less likely to vote liberal on a particular issue, so ideology is predictive of case outcome.
This is not surprising as intuitively we know this, but sometimes our view of the court as an impartial body overshadows this reality. Judicial ideology has a significant effect on whether a justice will vote a particular way or not, so it is likely too that ideology will color the way they perceive the evidence and merits of different cases.
Beyond pure ideology being predictive, there also appears to be a difference in voting behavior on popular cases such as those features in the NYTimes. For those salient cases liberal justices (MQ score < 0) are 13% more likely than for a regular case to vote for the liberal outcome. Similarly, conservatives are significantly more likely to vote for the conservative outcome in these salient cases.
Whether this behavior to due to justices wanting to signal their liberal or conservative viewpoints on popular cases or that newspapers only tend to cover more controversial cases that cause justices to vote more ideologically is unclear. But we do see that justices tend to vote closer to their ideological leanings on certain cases in predictable ways.
As ideology is so important, we can wonder how a new justice might affect existing opinions on the court and whether that effect depends on the new justice’s ideology.
To examine this we can use historical data on voting patterns in a particular term similar to above, regression whether an individual justice will vote liberal or conservative on their baseline ideology and a host of other controls, adding in a term for whether a new justice joined that term and the new justice’s subsequent ideology.
When a new justice is added, on average all justices appear to become more likely to vote liberal, even controlling for their baseline ideology, tendency to become more liberal over time, and the types of cases being debated. This effect however disappears when the new justice that is being added is conservative, meaning that adding a conservative justice makes all existing justices relatively less likely to vote liberal on a particular case.
An individual justice’s voting pattern is affected by the types of justices they are surrounded by and when new ones are added, they affect existing ideology in predictable ways.
All the above is to say that we should remember that the court is made up of people at the end of the day who suffer from the same psychological biases that regular voters suffer from. Justices will selectively look at available information, affected by things like confirmation bias, and naturally be more susceptible to particular types of arguments.
Accepting this does not impugn the court’s independence or professionalism of the individual justices. Rather, knowing the limits of their ability to be impartial will lead to higher trust in the court and a realistic understanding of its limitations.
Data Sources:
