Building a Taxonomy of Litigation: Clusters of Causes of Action in Federal Complaints

• Author: Zoran Obradovic,Christina L. Boyd,Kosta Ristovski,David A. Hoffman
• DOI: http://doi.org/10.1111/jels.12010
• Published date: 01 June 2013
• Date: 01 June 2013

Building a Taxonomy of Litigation:

Clusters of Causes of Action in

Federal Complaints

Christina L. Boyd, David A. Hoffman, Zoran Obradovic, and

Kosta Ristovski*

This project empirically explores civil litigation from its inception by examining the content

of civil complaints. We utilize spectral cluster analysis on a newly compiled federal district

court data set of causes of action in complaints to illustrate the relationship of legal claims

to one another, the broader composition of lawsuits in trial courts, and the breadth of

pleading in individual complaints. Our results shed light not only on the networks of legal

theories in civil litigation but also on how lawsuits are classified and the strategies that

plaintiffs and their attorneys employ when commencing litigation. This approach permits us

to lay the foundation for a more precise and useful taxonomy of federal litigation than has

been previously available, one that, after the Supreme Court’s recent decisions in Bell Atlantic

v. Twombly (2007) and Ashcroft v. Iqbal (2009), has also arguably never been more relevant

than it is today.

The idea of “a plain and short statement of the claim” has not caught on. Few complaints follow

the models in the Appendix of Forms. Plaintiffs’ lawyers, knowing that some judges read a

complaint as soon as it is filed in order to get a sense of the suit, hope by pleading facts to

“educate” (that is to say, influence) the judge with regard to the nature and probable merits of the

case, and also hope to set the stage for an advantageous settlement by showing the defendant what

a powerful case they intend to prove.

Judge Richard Posner

American Nurses Ass’n v. Illinois (1986)

*Address correspondence to Christina Boyd, University at Buffalo, SUNY, Department of Political Science, 520 Park

Hall (North Campus), Buffalo, NY 14260; email: cLboyd@buffalo.edu or David Hoffman, Beasley School of Law,

Temple University, 1719 N. Broad St., Philadelphia, PA 19122; email: David.Hoffman@temple.edu. Boyd is Assistant

Professor of Political Science at University at Buffalo, SUNY; Hoffman is James E. Beasley Professor of Law at Temple

University Beasley School of Law; Obradovic is Professor of Computer and Information Sciences and Director of the

Center for Data Analytics and Biomedical Informatics at Temple University; Ristovski is a Ph.D. candidate in the

Department of Computer and Information Sciences at Temple University.

We appreciate the support for our research from the University at Buffalo’s Baldy Center for Law & Social Policy

and Temple University’s Beasley School of Law. We also gratefully acknowledge the research assistance of Geoffrey

Bauer, Keith Blackley, Matthew Canan, Antima Chakraborty, Melissa Jabour, Nicholas Mozal, Jacqueline Sievert, and

Elizabeth Young and the preliminary analyses of Daniel Katz and Michael Bommarito. We received helpful comments

on this project from colleagues at workshops at Temple Law School, Duke Law School, North Carolina Law School,

Rutgers-Camden Law School, and the University of Georgia Department of Political Science, participants at the

Midwest Political Science Association and Stanford Conference on Empirical Legal Studies 2012 annual meetings,

attendees at a joint research workshop with the Federal Judicial Center and the Administrative Office of the U.S.

Courts, and Joe Cecil, Dawn Chutkow, Kevin Clermont, Scott Dodson, Marc Galanter, Lonny Hoffman, William

Hubbard, Greg Mandel, Morris Ratner, Paul Stancil, Rick Swedloff, Steve Subrin, and the editors and anonymous

reviewers at JELS. Authors are listed in alphabetical order.

bs_bs_banner

Journal of Empirical Legal Studies

Volume 10, Issue 2, 253–287, June 2013

253

Judge Posner’s opinion in American Nurses illustrates the dilemma of the complaint drafter.

Attorneys often want to tell a story in their pleading—to frame the litigation favorably for

an attentive judge or his or her clerk. But the stories that unlock the courthouse door

change. Once, judges appeared to prefer Hemingway’s concise prose, as “[t]he draftsmen

of the Civil Rules proceeded on the conviction, based on experience at common law and

under the codes, that pleadings are not of great importance in a lawsuit” (Wright et al.

2002). But it is now evident that William Gaddis is a better lodestar. After lower court

decisions in the 1980s, 1990s, and early 2000s, reams of scholarship, and the Supreme

Court’s eventual input in Bell Atlantic v. Twombly (2007) and Ashcroft v. Iqbal (2009),

well-counseled plaintiffs will create a detailed and plausible factual narrative, despite what

the Rules say. Indeed, plaintiffs’ complaints are said to be more important now than at any

time since the drafting of the Rules in the 1930s. But, given selection effects, pleading

strategy is as difficult as ever to study systematically.

In this article, we focus on a particular aspect of the pleading story: the channeling by

plaintiffs of their factual narrative into particularized legal claims, or causes of action. These

causes of action exemplify the cross-cutting tensions of pleading writ large. The rules of civil

procedure nominally permit liberal joinder of claims in one suit, and the failure to plead

a particular legal claim will often lead to preclusion in later cases. However, increased

judicial skepticism of private plaintiffs, and consequent doctrinal changes in pleading,

counsel against bringing causes of actions that the facts do not immediately suggest (Miller

2010). Thus, though the parties may still cast a wide net, it seems likely that the more

strategically wise choice is to be attentive to the relationship between causes of action, and

to attempt, to the extent possible, to frame a coherent nexus of causes of action in a

particular complaint.

To better understand this problem, we collected and culled a set of over 2,000 federal

complaints and coded the alleged causes of action in each. We then analyzed the relation-

ship between these complaints based on their underlying causes of action—over 7,400 of

them—using spectral clustering. Cluster analysis provides a means to objectively classify

large data sets and has been widely used for the sorts of taxonomic exercises that are critical

foundational work in many sciences. In this present study, cluster analysis allows us to

describe and summarize civil complaints, in isolation and in relationship to one another, in

ways that previous work simply could not do. Our analysis demonstrates that there are stable

relationships between the causes of action found in this set of complaints—indeed, we find

that causes of action cluster into eight typical patterns. These patterns permit us to develop

a more precise and therefore useful taxonomy of federal litigation than has been previously

available.

I. Complaints and Causes of Action

A. From Writ to Cause of Action

At the heart of doctrine lies the cause of action. In every U.S. jurisdiction, parties may join

together distinct theories that they believe justify legal relief. That is, they may bring

254 Boyd et al.

multiple causes of action; they may even join federal and state legal theories together in

federal court if they “form part of the same case or controversy under Article III of the

United States Constitution” (28 U.S.C. 1367(a)). But this modern cause of action practice

is a relatively recent procedural innovation.

In their original incarnation, the ancient system of writs coincided with distinctive

theories of legal relief. As Bracton wrote, “there may be as many forms of action as there are

causes of action” (Plucknett 1956:37). Each writ was issued in response to fact patterns that

reoccurred, and particular writs came to be used for common complaints. Over time, these

patterned writs were fixed—fact patterns had to be shaped to fit the available procedural

formula. Judges also greatly restricted the joinder—that is, the ability to bring together

distinct legal theories in one “case”—of distinct writs. The resulting system was arcane,

technical, and extremely expensive to access (Hepburn 1897).

New York’s famous Field Code sought to replace this obscure system and start afresh.

It employed the term “cause of action” to describe those groupings of facts that would result

in judicial intervention. The term originally therefore implied that the plaintiff had iden-

tified a set of circumstances for which there was a known remedy (Subrin 1987). Even so,

the Field Code limited joinder of these causes of action based on the substantive legal

nature of each (Hazard 1988). For example, New York permitted the joinder of just seven

general kinds of action in one complaint: contracts; injuries by force to person or property;

injuries without force to person or property; injuries to character; claims to recover real

property; claims to recover personal property; and claims against a trustee (N.Y. Laws, c.379

(1848)). Arguments over joinder bedeviled theorists, who viewed the intellectual incoher-

ence of the term “cause of action” as a precipitating cause (Gavit 1930).

Reflecting this hostility, Charles Clark, the reporter for and force behind the original

Federal Rules, believed that the cause of action was nothing more (or less) than “an

aggregate of operative facts, a series of acts of events, which gives rise to one more legal

relations of right-duty enforceable in the courts” (Clark 1924). Over time, this realistic

conception of the cause of action came to dominate, providing the architecture for the

innovative federal rules regime (Bone 1989; Sherwin 2008). The Rules famously avoid the

term “cause of action” entirely, instead focusing on a “claim for relief,” and the type of

factual notice that would apprise the defendant of the nature of the theories arrayed against

it. That is, as originally proposed, the Federal Rules do not require plaintiffs to plead causes

of action at all, and Rule 18, which governs joinder, enables bringing together theories of

relief without regard to the underlying doctrinal categories that had dominated practice.

Since most states’ procedural codes are modeled on the Federal Rules, one might have

imagined that the cause of action, like the writ, was extinct.

B. The Modern Practice and Theory of Multiple Claim Pleading

But nothing could be further from the truth. Most lawyers continue to plead independent

causes of action in both federal and state court. They do so for many reasons. Primarily, the

conservative nature of local legal culture de-motivates changes to traditional pleading

practices (Main 2001), and lawyers are told that increasing the number of causes per case

will lead to higher rates of recovery (Berger et al. 2005; Eisenberg 2007). That lesson begins

Building a Taxonomy of Litigation 255