Justice Scalia's death has justifiably prompted a flurry of speculation about how, or even if, he will be replaced on the Supreme Court.
Underlying most analyses is some sense of how the Court fits into the ideological spectrum defined by Congress and the president. Usually this is easy for anyone who has been paying attention: Scalia was conservative, Justice Ginsburg is liberal, and so forth.
Getting more fine-grained than that is tricky, however. There are several ways to measure the ideology of Supreme Court justices. Martin-Quinn scores are one version of this, but there are other competing ways of doing it.
These varying approaches produce a surprisingly large range of ways to view the Roberts Court. Depending on the assumptions and models we use, we can make four plausible cases about the Court: It's very conservative, it's very liberal, it's moderately conservative, or it's both liberal and conservative at the same time.
Drilling down a bit on how these estimates work can help us appreciate how we should think about ideological evolution on the Court, which, in turn, can help us understand the politics of appointments and, more broadly, how the Court fits in the political system.
Case 1: The Court is very conservative
This view is not in vogue now, but only a couple of years ago it was common to note that the Court was not only conservative but historically so. The dashed red line in the figure below shows the Court median from Martin and Quinn scores from 2013. Looking at that, it's hard not to think the Court was indeed moving into Attila the Hun territory.
Case 2: The Court is very liberal
Martin and Quinn revised their scores in 2013, producing a very different image of the Court. The solid green line in the figure above shows the Martin and Quinn medians as calculated in 2014. Now the Court is not only much more liberal than had been calculated earlier, but it is quite liberal by historical standards. In fact, it is, according to that measure, more liberal than at any time since 1969.
And the Court could become much more liberal. If Obama successfully appoints a nominee to the left of Justice Breyer, these most recent Martin and Quinn scores suggest Breyer would become the most liberal median of the Court since 1937 (when the data set begins).
If you're like me, you weren't sold on the Attila the Hun story and are not sold on the Che Guevara story either. But that puts the onus on us to define what should constitute evidence of ideological movement on the Court.
The assumption underlying Martin and Quinn scores is that the agenda does not systematically move over time. That is, there are not systematic shifts in the conservatism or liberalism of cases. If they were scaling standardized tests, the assumption would be that the average difficulty of tests over time has not changed.
Such an assumption can be problematic. In recent years, there has been a high percentage of liberal opinions. Yet there is liberal, and there is liberal. Voting liberally on a state law for drawing and quartering jaywalkers is different from voting to free a murderer on a sketchy claim of incompetent counsel. The Martin and Quinn estimates do not bite off trying to pin down movement in the agenda over time, and this can mean that movements in the agenda may be hard to disentangle from changes in preferences.
This is almost certainly the reason that Martin and Quinn estimated the early 1970s Court (the Court responsible for finding abortion in the Constitution but not the death penalty) was very conservative.
Case 3: The Court is moderately conservative
This is my favored view, at least as of 2012. The key to my approach is using additional information to account for changes in the Court agenda over time. For example, while we can't directly observe if the agenda facing Justice Thomas in 1991 was more conservative or liberal than the agenda in 1973, we do happen to know what he thought in 1991 about some of the cases in 1973. In particular, he clearly expressed his view on Roe v. Wade (he didn't like it, as you would imagine).
The analogy is to test standardization. A test score based on one set of questions is hard to compare with another score based on a different set of questions. If we gave identical tests, we could compare scores, of course, but that is infeasible for standardized tests that cannot simply give identical tests year in and year out. Some questions can overlap, though, and this is enough to identify relative test performance, as the overlapping questions provide information about relative ability across tests.
For the Court case, we will identify certain "questions'' (cases) that have been "asked'' (voted on) over time. We will also be able to identify cases that are to the left or right of previous cases, information that also helps pin down agenda and preference change over time.
These bridging models leverage two sources of information that identify preference changes over time. One is the positions taken by justices on cases decided by earlier courts. It is relatively common for justices to state a clear position about an earlier decision. For example, in FEC v. Wisconsin Right to Life (2007) Scalia stated that Austin v. Michigan Chamber of Commerce (1990) was "wrongly decided." That helps us see where the 2007 Scalia was with regard to at least one case in 1990. The more observations like this we have, the better we can see how justices have changed (or not) over time.
The second source of bridging information is about the relative position of cases over time. It is not unusual for case law to evolve in an ideologically understandable way. For example, the Court held in Miller v. Alabama (2011) that juveniles convicted of murder cannot be subject to mandatory life sentences. The cutpoint of this case was clearly to the left of Graham v. Florida (2010), in which the Court ruled against mandatory life sentences for juveniles for crimes other than murder. To see this, consider the vote of a liberal on Graham. Does that liberal vote imply support for the liberal position on Miller? It does, as voting against mandatory life sentences for murder implies opposition to mandatory life sentences for lesser crimes.
Using a fairly standard "item response theory" measurement model to analyze this data produces the dark blue line in the figure below. In contrast to either of the Martin and Quinn estimates, these estimates imply the Court in 2011 was moderately conservative — basically near the level of conservatism it had been since 1990. Reassuringly, these estimates did not have the conservative spike in 1973 in the Martin and Quinn estimates.
Where is the Court today? I really don't know. (I'm working on it.) My intuition is that it probably hasn't moved much in terms of its overall median, as the liberal same-sex marriage decisions will be balanced by conservative decisions such as Shelby County v. Holder.
If Obama were able to successfully appoint someone to the left of Justice Kagan, the median would be -0.50, which, we can see from the figure, would make the Court quite liberal but not historically liberal (in contrast to the median implied by the most recent Martin and Quinn scores).
Case 4a: The Court is both liberal and conservative
Even if the Court is, on average, moderately conservative, that is not the whole story.
It is possible the Court is like a quantum particle, simultaneously existing in two states, a liberal one and a conservative one. Brandon Bartels and Chris Johnston advance this claim in an article arguing that the Court is liberal on some salient cases (which is what conservatives see) and conservative on other salient cases (which is what liberals see). This is how nice Chief Justice Roberts can piss off everyone.
To the extent that this is true, the inherent conflict between a Democratic president and a Republican Senate will only be exacerbated. Obama and supporters will think they need to move the Court to left simply to bring it in the mainstream, while Republicans think the exact opposite.
Case 4b: The Court is (sometimes) neither liberal nor conservative
While I think there are likely different subjective views on the Court, there is also something interesting and inherent about Court decision-making that should push us to recognize that the Court is sometimes neither liberal nor conservative. This argument has a few steps. The first is to note that the Court is not simply a little Congress. In Congress, if members are very conservative on voting rights, they tend to be very conservative on abortion, on foreign policy, and on welfare spending.
The Court, however, has several institutional features that complicate the relatively simple ideological story that works when describing Congress. The first is that life tenure gives more justices freedom to deviate from ideological norms. Members of Congress certainly can deviate from their expected ideological position at times, but they risk defeat in primaries and marginalization within their parties.
In addition, the ideological environment justices operate in is different. Yes, the role of government is an absolutely central question, but, unlike in Congress, other values are taken seriously as well, including the role of precedent, the legitimacy of the court to overrule elected officials, and more. Remembrances of Scalia typically note that he could sometimes be almost (gasp!) liberal in some areas of the law (even as he was breathtakingly conservative in others).
A second step in this argument is that the small size of the Court matters. While members of Congress can be idiosyncratic, surprising ideological choices simply wash out in the large chambers, as it is seldom the case that a single individual member of Congress exerts virtual control over the legislative outcome. On the Supreme Court, on the other hand, there are fewer institutional mechanisms to constrain justices and fewer decision-makers to wash out the idiosyncrasies that persist. This means the Court is more prone to be driven by idiosyncratic positions that don't necessarily map easily onto the main political ideological space.
The result is that the Roberts Court was dominated by Justice Kennedy, who is generally a moderate conservative but is liberal on some issues (such as same-sex marriage) and conservative on other issues (such as campaign finance).
So even though the Court is generally moderately conservative, it can produce a conservative blockbuster like Shelby County on one day and liberal blockbusters like Windsor (2013) and Hollingsworth (2013) on literally the next day.
The political fight over the Scalia seat will be primarily ideological. While some may argue that the Court is very liberal and could become epically liberal, I think the most reasonable view is that it is a center-right body, which Obama could possibly move to a center-left body. And whatever happens, the Supreme Court will retain some capacity to surprise us ideologically, as hard as it will be for partisans to see this.
Michael Bailey is the Colonel William J. Walsh professor of American government in the Department of Government and the McCourt School of Public Policy at Georgetown University.