The Supreme Court decided Evenwel v. Abbott this week – a case with vast implications for legislative districts. Some see the decision as snubbing Republicans by ruling that states and localities should use total rather than voting population to draw these districts. Other commentary views the ruling as sufficiently narrow to allow future litigation in the same area. The issue of how to properly draw legislative districts is not new to the Supreme Court. Especially since the Voting Rights Act of 1965, the Supreme Court has scrutinized redistricting and gerrymandering concerns many times over. In doing so the Court has defined specific cases as essential precedent concerning the constitutionality of redistricting plans and procedures.
Which cases are at the heart of the constitutional discussion of legislative districts? This post uses network analysis to analyze the relative importance of the Court’s past cases in defining this doctrine. I began by doing a case search with basic parameters looking for the phrase “legislative districts” between 1990 and the present. The search provided 21 Supreme Court cases. One of these cases only mentioned legislative districts once and do not inquire into this issue in the same manner as the other cases. The remaining twenty cases including Evenwel are deeply focused on the constitutionality of the redrawing of legislative districts. The twenty cases include: Evenwel v. Abbott, Shelby County v. Holder, Ariz. State Legis. v. Ariz. Indep. Redistricting Comm’n, Ala. Legis. Black Caucus v. Alabama, Bartlett v. Strickland, League of United Latin Am. Citizens v. Perry, Vieth v. Jubelirer, Georgia v. Ashcroft, Bartlett v. Stephenson, Sinkfield v. Kelley, DOC v. United States House of Representatives, N.J. v. New York, Lawyer v. Department of Justice, Bush v. Vera, Johnson v. De Grandy, Shaw v. Reno, Voinovich v. Quilter, Growe v. Emison, New York v. United States, and Dep’t of Commerce v. Mont.
My main interest was to see which cases are cited within these case, in order to examine the precedent that is central to the constitutional contours of this debate. To begin I generated a graph with these twenty cases that shows the cases these decisions cited to (the nodes for the cited cases aside from the twenty are not listed in this graph in order to keep the citing nodes visible).
When the cases are more closely clustered together in the graph, it shows that they are more central to the network. Based on the visual depiction alone cases like League of United Latin American Citizens v. Perry and Shaw v. Reno appear central, while cases like New Jersey v. New York and Bartlett v. Stephenson appear more on the outer edge of the network. Filling in the blank nodes the next graph labels every case that is cited to as well as the same citing cases from the figure above.
While this graph is much more difficult to interpret due to the number of cases it encompasses, it helps convey the complexity of this line of precedent. The citing nodes are still in the same place but the outer edge of this graph contains the cases that are cited to less frequently in this set. Similarly, the nodes closer to the central clusters in the graph are those that are cited more often by these cases. Based on this approach, the older cases have an advantage of having more cases citing to them compared with the more recent decisions.
With the network data I also calculated a few basic attributes of the individual cases. The first is the frequency with which the precedent is cited. The table below has all cases that are cited to by at least six of the twenty citing cases.
The most frequently cited case, US v. Detroit Timber & Lumber Co. (1906) is, as expected, one of the older cases in the group. It is not cited for redistricting purposes, but rather to clarify that the syllabus before the Supreme Court opinion is not part of the opinion. The next two cases, however, both precede the cases from 1990 and deal with redistricting. Reynolds v. Sims (1964) discussed the importance of districts with a similar amount of people and the Court ruled in Thornburg v. Gingles (1986) that North Carolina’s districts unconstitutionally prevented black voters from equal participation.
The next table accounts for inward and outward citations to locate the most central decisions in this group. The measure is known as eigenvector centrality.
The table shows the fifteen most central cases to the network. The centrality measures support the visual in the top figure. When accounting for the entire citation network, League of United Latin American Citizens (2006) and Bush v. Vera (1996) are the two most central cases followed by Vieth v. Jubelirer and Shaw v. Reno. The first non-citing case that comes up in this list is US v. Detroit Timber with Thornburg v. Gingles two below it. League of United Latin American Citizens is not one of the older cases but due to both citations to and from the case it is the most central. Bush v. Vera is a decade older than the Latin American Citizens decision and similarly cites to many of the most important precedents in this group. The Vera decision had more time to develop cases citing to it as well.
It will take years before we know if Evenwel will be central to this class of cases. Out of the most central cases in this group, however, Ginsburg’s majority decision in Evenwel only cites to Reynolds v. Sims. This by itself already places the case on the outer edge of the network graph. Right now its centrality is only .015 (entirely based on outward citations), yet this will change as it is cited in future cases.
Follow me on Twitter: @AdamSFeldman