Insight Center: Publications

Lessons from GM Bellwether Selection: Fairness Over Farce

Client Alert

Authors: Barry H. Boise and Colleen C. Kelly

Lessons from GM Bellwether Selection: Fairness Over Farce

By advancing their argument, plaintiffs exposed what parties have long known — that the ‘bellwether’ process has lost its meaning….This process results either in resolutions dictated by aberrant cases or, more likely, a waste of money and effort as the litigation resumes.

This article was published in the Automotive, Class Action, Product Liability, Transportation, and Trials sections of Law360 on February 19, 2016. © Copyright 2016, Portfolio Media, Inc., publisher of Law360. It is republished here with permission.

Although a group of the plaintiffs’ counsel lost the fight to topple the lead counsel in a significant multidistrict litigation, their strategy warrants scrutiny by bench and bar. Confusing “bellwether case” — an indicator of trends according to the dictionary and the caselaw — with “best case,” they accused co-lead counsel of betraying the plaintiffs in the nationwide General Motors Ignition Switch Litigation. Had they succeeded, the challengers would have undermined a core concept in mass litigation and devalued a tool that courts use to resolve expensive litigation in an efficient manner. This article analyzes the dangerous thinking behind the attempted coup and calls for a procedure to choose bellwether cases that will result in fairer and more cost-effective resolutions of costly litigation.

The GM Challenge

Seeking to unseat co-lead counsel, the lawyers who challenged the GM case selection process rejected the premise that bellwether cases should represent the remaining matters in the litigation. Instead, they argued that the “best” bellwether case is one that results in a plaintiff’s verdict that “causes the defendant to reconsider its litigation strategy and to place a higher settlement value on pending cases.”1 Thus, despite a court order that bellwether selections “must constitute a representative sampling of cases,”2 they argued that lead counsel bears a duty to hold a finger on the scales of fairness.

The court denied the plaintiffs’ motion, indicating that a full opinion would follow. But, by advancing their argument, plaintiffs exposed what parties have long known — that the “bellwether” process has lost its meaning. In many cases, bellwether selection is not about picking representative cases to allow for a fair assessment of the mass of cases. Rather, when the parties run the case selection process, plaintiffs choose outlier cases in an attempt to inflate the settlement value of their remaining inventory. This process results either in resolutions dictated by aberrant cases or, more likely, a waste of money and effort as the litigation resumes.

But bellwether selection can work better and in a way to reduce party bias. Other methods yield more representative cases and lead to more just resolutions. This article explores such methods and calls for a system of random selection that will produce representative cases and fairer resolutions.

The Purpose of Bellwether Trials

Bellwether trials are test cases used in MDLs to hasten the settlement of expensive mass litigation. Although MDLs coordinate and consolidate pretrial proceedings for many different cases and (theoretically) return them to their courts of origin for trial, most cases never leave the MDL court. To allow the parties to test their claims and defenses in front of a jury before dispersing cases for multiple trials across the country, the transferee courts often preside over bellwether trials. Ideally, these trials give the parties an opportunity to evaluate the strengths and weaknesses of their cases and assist in facilitating global settlement.3 While mass settlement has become the plaintiffs’ expected conclusion of the bellwether process — regardless of the merits of the underlying claims — bellwether trials also yield other valuable information. U.S. District Judge Jesse Furman in the GM litigation recognized as much when he denied the plaintiffs’ motion to remove the co-leads. In “focus[ing] myopically on the outcome of the first bellwether trial,” he stated, the plaintiffs ignored other ways that the trial had advanced the MDL, including the rulings on issues that would apply in future proceedings.4

To yield useful information and an eventual fair resolution, the cases selected for a bellwether trial must be representative of the rest of the litigation. According to the Manual for Complex Litigation, bellwether cases “should produce a sufficient number of representative verdicts and settlements to enable the parties and the court to determine the nature and strength of the claims, whether they can be fairly developed and litigated on a group basis, and what range of values the cases may have if resolution is attempted on a group basis.” When the selected bellwether cases are not representative of the remaining MDL cases, they skew the results and “waste[] substantial amounts of both time and money.”5

Bellwether Trial Selection Methods

In the seminal article on bellwether selection, U.S. District Judge Eldon Fallon described a process where courts shape the selection of representative bellwethers by first creating categories of major variables into which all MDL cases should be cataloged. In pharmaceutical cases, for example, these variables might include duration of use, type of injury and timing of injury in relationship to a label change. Candidates for bellwether selection are then drawn from each category, and individual discovery proceeds. The court or the parties then choose bellwethers from the remaining cases.6

Courts enjoy wide latitude in setting selection processes, but they typically use some variation on three methods: the parties pick, the court picks or the cases are selected randomly.

Party Selection

Often, the parties choose the cases for discovery and then for bellwether trials without court involvement. When courts participate in identifying cases, they often enlist party input, except in the instance of random selection.

Following the typical pattern, the GM Ignition Switch Litigation relied on the parties to choose cases both for the discovery pool and for the bellwether trials. The court set guidelines for the cases that would populate the “initial discovery pool,” a group of 18 personal injury or wrongful death cases — nine chosen by each side.7 The court limited the scope of the initial pool to cases involving six vehicle models in which airbags did not deploy.8 After case-specific fact discovery, each side identified five cases as bellwether trial candidates. Then the parties exercised two strikes on their opponent’s list, leaving six cases for expert discovery and trial.9

In his article, Judge Fallon endorsed this type of selection process because it “institutes fairness and attorney participation.” This method gives both parties a vested interest in presenting the best possible version of their case.10 The problem, however, with this bellwether selection process — employed in many litigations — is that it can lead to trials that yield information only about outlier cases, as opposed to the overall docket. And, because plaintiffs hold the power of dismissal, they can dismiss cases chosen by the defendant until the case up for trial favors them.

Once the court allows parties to select trial candidates, it has no means to ensure that those selections are representative of the larger litigation. To illustrate, Judge Furman instructed the GM litigants that “the cases selected as trial candidates must constitute a representative sampling of the cases in this proceeding” and that “the parties ... [were] not to select cases presenting unique or idiosyncratic facts or law that would render the results of these cases unenlightening.” But he lamented that the court “cannot police this request.”11

Random Case Selection

Random selection eliminates the mischief inherent in party selection. It more likely yields representative cases that do not unfairly favor one party over the other. Empirical data have shown that random choices result in cases that better represent the cases remaining on the docket.12

The Byetta litigation consolidated before Los Angeles Superior Court Judge William F. Highberger in California state court provides an instructive example of random case selection in the bellwether process. There, Judge Highberger selected a discovery pool of 25 cases at random from a list of those remaining in the litigation. He ensured that each of the major participating plaintiffs’ firms was proportionally represented in the pool of cases chosen. The parties then began discovery on the selected cases. As discovery progressed, the weaknesses of the overall inventory emerged as many of the randomly chosen cases were dismissed — either voluntarily or through summary judgment. The number of dismissals provided the court with insight on the strength of the overall inventory. At the end of discovery, the court — with party input — sequenced the six remaining cases for trial. However, the court added a safeguard against additional voluntary dismissals to manipulate the remaining pool: the court allowed the defense to elevate two cases to the top of the list each time a trial case was dismissed.

The Byetta bellwether selection process resulted in the fair selection of cases. Allowing a large pool of cases to undergo fact and expert discovery enabled the parties to weed out cases from the pool and from those left behind.

Despite the value of random selection, Judge Fallon’s article urged courts to reject the procedure because it detaches the attorneys from the process and thus impedes the goal of global resolution.13 Once removed from the selection process, counsel may not believe the bellwether cases are representative and may be unwilling to extrapolate the outcome to the rest of the litigation. Judge Highberger addressed this issue by allotting a certain percentage of discovery pool cases to each of the plaintiffs’ firms with large inventories. In this way, the major players had skin in the game and were motivated to present their strongest possible arguments. The court also selected a sufficient number of cases, which undermined any complaint that none of the 25 cases were representative.

Other potential problems with randomized bellwether selection can be overcome through the manner in which selection proceeds. The court and the parties can identify key variables and categorize the MDL docket accordingly. Cases can then be selected randomly from categories. In the Byetta litigation, the court sorted cases by alleged injury (pancreatitis) and the plaintiffs’ law firm and then randomly filled the discovery pool from those cases.

Science First: The Lipitor Selection Process

The court can further ensure bellwether efficiency by requiring that cases first meet certain indicia of specific causation, as U.S. District Judge Richard M. Gergel did in the Lipitor MDL. There, the court defined specific causation criteria, requiring counsel to demonstrate a “legally recognizable causative link to diabetes (the alleged injury)” before qualifying as potential bellwethers.14 Eligible cases had to comport with the relevant science — i.e., a clinical trial that found a doubling of the risk of new onset diabetes in patients using a certain dosage who also exhibited four risk factors.15 When plaintiffs failed to satisfy the criteria, counsel conceded that their claims could not survive summary judgment. Judge Gergel then eliminated them from bellwether consideration.16

By first defining, and then limiting, the universe of potential bellwethers to those with a cognizable scientific basis, the process weeds out those claims that eventually would not survive.


The pursuit of representative bellwether cases through party selection will not work because unrepresentative samples will not produce broad settlements. Many plaintiffs will determine, as did the GM litigants, that the “most important task” is to pick a strong case so that “the plaintiffs’ attorney [can] convince a defendant that if it takes a case to trial, it will get creamed.”17 The plaintiffs’ goal to shape the outcome of a litigation by trying outlier cases does not serve the goals of the bellwether process. Random case selection offers a vehicle for eliminating these outlier cases, which waste time and prejudice defendants, and allows the parties to craft a resolution based on a realistic view of all the underlying cases. Further, procedures are needed to safeguard against manipulation of the pool through voluntary dismissals. A science-first approach may winnow the field enough that bellwether selection from a large pool becomes unnecessary.

DISCLOSURE: The authors represent Defendant Eli Lilly and Company in ongoing Byetta products liability litigation pending in state and federal court.


1 Plaintiffs’ Motion to Remove the Co-Leads and Reconsider the Bellwether Trial Schedule at 10, In re General Motors Ignition Switch Litigation, No. 14-md-2543 (S.D.N.Y. Jan. 1, 2016).

2 Order No. 25 at ¶ 34, In re General Motors Ignition Switch Litigation, No. 14-md-2543 (S.D.N.Y. Nov. 19, 2014).

3 Manual for Complex Litigation (Fourth) § 20.132.

4 Order No. 95 at 2, In re General Motors Ignition Switch Litigation, No. 14-md-2543 (S.D.N.Y. Feb. 10, 2016).

5 Manual for Complex Litigation (Fourth) § 22.315.; Eldon E. Fallon, et al., The Problem of Multidistrict Litigation: Bellwether Trials in Multidistrict Litigation, 82 Tul. L. Rev. 2323, 2344 (2008).

6Fallon, 82 Tul. L. Rev. at 2326.

7 Order No. 25 at ¶¶ 8-9, In re General Motors Ignition Switch Litigation, No. 14-md-2543 (S.D.N.Y. Nov. 19, 2014).

8 Order No. 34 at 2, In re General Motors Ignition Switch Litigation, No. 14-md-2543 (S.D.N.Y. Feb. 6, 2015).

9 Order No. 25 at ¶¶44-46, 49, In re General Motors Ignition Switch Litigation, No. 14-md-2543 (S.D.N.Y. Nov. 19, 2014).

10 Fallon, 82 Tul. L. Rev. at 2362

11 Order No. 25 at ¶ 34, In re General Motors Ignition Switch Litigation, No. 14-md-2543 (S.D.N.Y. Nov. 19, 2014).

12 Loren H. Brown, Matthew A. Holian & Dov Rothman, Random Selection Is Best for Bellwether Trials, Law360 (Oct. 20, 2014).

13 Fallon, 82 Tul. L. Rev. at 2362.

14 Letter of B. Hahn to J. Gergel, In re Lipitor (Atorvastatin Calcium) Marketing, Sales Practices, and Products Liability Litigation, No. 2:14-mn-2502 (D.S.C. Jan. 19, 2016).

15 Case Management Order No. 61, at 2, In re Lipitor (Atorvastatin Calcium) Mktg., Sales Practices, and Prods. Liab. Litig., No. 2:14-mn-2502 (D.S.C. Jan. 12, 2016).

16 Case Management Order No. 66, In re Lipitor (Atorvastatin Calcium) Mktg., Sales Practices, and Prods. Liab. Litig., No. 2:14-mn-2502 (D.S.C. Jan. 25, 2016).

17 Plaintiffs’ Reply Brief in Response to Co-Lead Counsel’s Memorandum in Opposition to Lance Cooper’s Motion to Remove Co-Lead Counsel and for Reconsideration of the Order Approving the Qualified Settlement Fund at 3, In re Gen. Motors Ignition Switch Litig., No. 14-md-2543 (S.D.N.Y. Feb. 2, 2016).

The material in this publication was created as of the date set forth above and is based on laws, court decisions, administrative rulings and congressional materials that existed at that time, and should not be construed as legal advice or legal opinions on specific facts. The information in this publication is not intended to create, and the transmission and receipt of it does not constitute, a lawyer-client relationship.

Data protection laws have changed, so we have revised our Privacy Policy.