Volume 1 of 2

                           PUBLISHED

UNITED STATES COURT OF APPEALS
                 FOR THE FOURTH CIRCUIT


MARY B. GUNNELLS; WILLIAM A.             
GUNNELLS; TELE TECH CORPORATION;
CYNTHIA MULLINAX; W & W
TRUCK AND TRACTOR COMPANY,
INCORPORATED; JERRY FREEMAN; AAA
TRANSMISSIONS, INCORPORATED;
PAMELA B. ROBERTSON;
AVIONICS/SEAONICS INTERNATIONAL,
INCORPORATED; JAMES E. CASSLEMAN;
LARRY STRICKLAND; BRENDA
WORKMAN; HELEN CLARK; BOBBY F.
BUCHANAN; JOANNE BUCHANAN;
BRIAN HANSEN; JAMES E. WISE; JOHN
N. STANLEY; DANIEL L. DRIGGERS;
BRENDA VOWELS; DAVID J. VOWELS;
LIVINGSTON AUTO PARTS,
                                            No. 01-2419
INCORPORATED; PHILIP LIVINGSTON;
LUKES CYCLE SERVICES,
INCORPORATED; JEFF LUCARELLI; HH
MOORMAN & SONS; HAROLD
MOORMAN; OAKLAND CLUB; MARK B.
BUXTON; STYLON OF CHARLESTON;
MARY PARKER; TERRY HENSLEY;
EARL SINGLETARY; ORA SINGLETARY,
individually and as representatives
of all persons and entities similarly
situated,
                 Plaintiffs-Appellees,
                  v.
                                         
2              GUNNELLS v. HEALTHPLAN SERVICES, INC.


HEALTHPLAN SERVICES, INCORPORATED,      
individually and as successor in
interest to Third Party Claims
Management, Incorporated; THIRD
PARTY CLAIMS MANAGEMENT,
INCORPORATED,
               Defendants-Appellants,
                and
FIDELITY GROUP INC.; INTERNATIONAL
WORKERS GUILD, INCORPORATED;
NATIONAL ASSOCIATION OF BUSINESS
OWNERS AND PROFESSIONALS; IWG
HEALTH AND WELFARE FUND;
FIDELITY CLAIMS MANAGEMENT,
INCORPORATED; EXECUTIVE
CONSULTANTS INSURANCE AGENCY;
RAYMOND R. OLIVI; JOHN BRANHAM;         
STEVEN BROUGHTON; JOHN COLLINS;
REX CULBERTSON; ROBERT DAVIES;
RICHARD DEXTAZE; KENNETH EVERY
ANDREWS; FRANKIE FORT; ROBERT
GERMAN; GARY GEITLER; MELISSA
GEITLER; MARTIN T. GEITLER;
MICHAEL D. GEITLER; GEORGE
GRADEK; HUBERT G. GREENE; PHILLIP
S. HODGSEN; EARL ISABELL; STEPHEN
LACY; FREDERICK MALONE; DAVID E.
MARTIN; MICHAEL MASSEY; HENRY
MERCHANT; TINA MESSINGER;
MICHAEL R. REQUA; BRUCE ROWAN;
PAMELA REGOPOULUS; BRUCE RYALS;
BARBARA A. SCOTT; HENRY K.
SKINNER, JR.; ELMER STONE; KENNETH
R. TALLMADGE; LEE TAYLOR;
                                        
               GUNNELLS v. HEALTHPLAN SERVICES, INC.   3


MICHAEL THOMAS; SCOTT TODD;             
ARNOLD H. VALENTINE; RICHARD M.
VANN; LEWIS H. WADE; DAVID M.
WHITE; HERBERT L. WILLIAMS; BEN
L. WELLS; JACK E. WILSON,
individually and in his
representative capacity; ANCHOR
INSURANCE AGENCY; BRANHAM AND
GEITLER ASSOCIATES; DIVERSIFIED
INSURANCE; LEGEND EQUITIES
CORPORATION; FINANCIAL BENEFITS,
INCORPORATED; THE TAYLOR AGENCY         
INCORPORATED; TRUETT AND
ASSOCIATES; COLONIAL LIFE &
ACCIDENT INSURANCE COMPANY, a
UNUM Company; DWAYNE A.
SAMUELS; EUGENE O. DUNCAN; DAVID
SPOONER; CHARLES L. BRADLEY;
TERRENCE E. RHUE; NOEL SHAW;
PAUL D. ASKEW; PHILLIP STASHIN;
KEN AVERIES,
                          Defendants.
                                        
4               GUNNELLS v. HEALTHPLAN SERVICES, INC.



MARY B. GUNNELLS; WILLIAM A.               
GUNNELLS; CYNTHIA MULLINAX;
JERRY FREEMAN; PAMELA B.
ROBERTSON; JAMES E. CASSLEMAN;
LARRY STRICKLAND; BRENDA
WORKMAN; HELEN CLARK; BOBBY F.
BUCHANAN; JOANNE BUCHANAN;
BRIAN HANSEN; JAMES E. WISE; JOHN
H. STANLEY; DANIEL L. DRIGGERS;
BRENDA VOWELS; DAVID J. VOWELS;
PHILIP LIVINGSTON; JEFF LUCARELLI;
HAROLD MOORMAN; MARK B.
BUXTON; MARY PARKER; TERRY
HENSLEY; EARL SINGLETARY; ORA
SINGLETARY, individually and as
representatives of all persons and
entities similarly situated; W & W
                                               No. 01-2420
TRUCK AND TRACTOR COMPANY,
INCORPORATED; LUKES CYCLE
SERVICES, INCORPORATED; AAA
TRANSMISSIONS, INCORPORATED;
LIVINGSTON AUTO PARTS,
INCORPORATED; HH MOORMAN &
SONS; TELE TECH COMPANY,
INCORPORATED; AVIONICS SEAONICS
INTERNATIONAL, INCORPORATED;
OAKLAND CLUB; STYLON OF
CHARLESTON, entities similarly
situated,
                   Plaintiffs-Appellees,
                   v.
                                           
              GUNNELLS v. HEALTHPLAN SERVICES, INC.   5


MICHAEL R. REQUA; ARNOLD H.            
VALENTINE; FINANCIAL BENEFITS,
INCORPORATED; LEWIS H. WADE;
BRUCE RYALS; GEORGE GRADEK; TINA
MESSINGER; KENNETH EVERY
ANDREWS; RAYMOND R. OLIVI;
EXECUTIVE CONSULTANTS INSURANCE
AGENCY; MICHAEL MASSEY; ANCHOR
INSURANCE AGENCY; JOHN COLLINS,
individually and in their
representative capacities,
              Defendants-Appellants,
                and
FRANKIE FORT; FIDELITY GROUP INC.;
INTERNATIONAL WORKERS GUILD,
INCORPORATED; NATIONAL
ASSOCIATION OF BUSINESS                
OWNERS AND PROFESSIONALS; IWG
HEALTH AND WELFARE FUND;
FIDELITY CLAIMS MANAGEMENT,
INCORPORATED; JOHN BRANHAM;
STEVEN BROUGHTON; REX
CULBERTSON; ROBERT DAVIES;
RICHARD DEXTAZE; ROBERT GERMAN;
GARY GEITLER; MARTIN T. GEITLER;
GARY GEITLER; MICHAEL D. GEITLER;
HUBERT G. GREENE; PHILLIP S.
HODGSEN; EARL ISABELL; STEPHEN
LACY; FREDERICK MALONE; DAVID E.
MARTIN; HENRY MERCHANT; BRUCE
ROWAN; PAMELA REGOPOULUS;
BARBARA A. SCOTT; HENRY K.
SKINNER, JR.; ELMER STONE; KENNETH
R. TALLMADGE; LEE TAYLOR;
                                       
6              GUNNELLS v. HEALTHPLAN SERVICES, INC.


MICHAEL THOMAS; SCOTT TODD;              
RICHARD M. VANN; DAVID M.
WHITE; HERBERT L. WILLIAMS; BEN
L. WELLS; JACK E. WILSON;
BRANHAM AND GEITLER ASSOCIATES;
DIVERSIFIED INSURANCE; LEGEND
EQUITIES CORPORATION; THE TAYLOR
AGENCY, INCORPORATED; TRUETT AND
ASSOCIATES; COLONIAL LIFE &
ACCIDENT INSURANCE COMPANY, a
UNUM Company; HEALTHPLAN
SERVICES, INCORPORATED, individually
and as successor in interest to Third    
Party Claims Management,
Incorporated; THIRD PARTY CLAIMS
MANAGEMENT, INCORPORATED;
DWAYNE A. SAMUELS; EUGENE O.
DUNCAN; DAVID SPOONER; CHARLES
L. BRADLEY; TERRENCE E. RHUE;
NOEL SHAW; PAUL D. ASKEW;
PHILLIP STASHIN; KEN AVERIES,
individually and in their
representative capacities,
                           Defendants.
                                         
           Appeals from the United States District Court
          for the District of South Carolina, at Charleston.
                Patrick Michael Duffy, District Judge.
                             (CA-98-2659)

                    Argued: September 23, 2002

                     Decided: October 30, 2003

      Before NIEMEYER, MOTZ, and KING, Circuit Judges.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   7
Affirmed in part, reversed in part, and remanded by published opin-
ion. Judge Motz wrote the opinion, in which Judge King concurred.
Judge Niemeyer wrote an opinion concurring in part and dissenting
in part.


                              COUNSEL

ARGUED: William R. Lewis, NELSON, MULLINS, RILEY &
SCARBOROUGH, L.L.P., Charleston, South Carolina; James Y.
Becker, HAYNSWORTH SINKLER BOYD, P.A., Columbia, South
Carolina, for Appellants. Christian Hancock Hartley, RICHARDSON,
PATRICK, WESTBROOK & BRICKMAN, L.L.C., Charleston,
South Carolina; Justin S. Kahn, KAHN LAW FIRM, Charleston,
South Carolina, for Appellees. ON BRIEF: Charles R. Norris, NEL-
SON, MULLINS, RILEY & SCARBOROUGH, L.L.P., Charleston,
South Carolina; John H. Tiller, Elizabeth J.V. Speidel, HAYNS-
WORTH SINKLER BOYD, P.A., Columbia, South Carolina; H.
Michael Bowers, YOUNG, CLEMENT, RIVERS & TISDALE,
Charleston, South Carolina; Thomas J. Keaveny, II, SIMONS &
KEAVENY, Charleston, South Carolina, for Appellants.


                               OPINION

DIANA GRIBBON MOTZ, Circuit Judge:

   This appeal presents the question of whether the district court
abused its discretion in conditionally certifying a class action in a suit
brought by purchasers and beneficiaries of a multi-employer health
care plan for claims growing out of the plan’s collapse. The district
court conditionally granted class certification against the plan’s
claims administrator, Healthplan Services Inc., as successor in interest
to Third Party Claims Management, Inc. (collectively, "TPCM") and
against individual and corporate insurance agents who marketed and
sold the plan. The court properly applied controlling legal principles
and made well-supported factual findings supporting its decision to
certify a class action against TPCM; thus, we see no abuse of discre-
tion in that decision. However, because the court rested its class certi-
8              GUNNELLS v. HEALTHPLAN SERVICES, INC.
fication against the individual agents on findings grounded in a
misapprehension of governing law, we must conclude that the court
did abuse its discretion in certifying those separate class actions.
Accordingly, we affirm in part, reverse in part, and remand for further
proceedings consistent with this opinion.

                                  I.

   Since this is an interlocutory appeal, filed even before the parties
have completed discovery, the facts surrounding the claims of the
purchasers and beneficiaries (collectively, "Plaintiffs") have not been
fully developed. Thus, the record to date reveals only the following
rough outline of the events leading up to the present lawsuit.

   Sometime in 1995, Fidelity Group, Inc., the National Association
of Business Owners and Professionals ("NABOP"), and the Interna-
tional Workers Guild, Inc. ("IWG"), created the IWG Health and
Welfare Fund (the "Fund"), which in turn offered a health care and
dental plan (the "Plan") to interested purchasers. From August 1996
until June 1997, when the South Carolina Department of Insurance
ordered marketing to cease, the agents marketed and sold the Plan.
Although it was purportedly marketed as an ERISA plan, the Plan
never complied with ERISA requirements.

   In April 1995, Fidelity hired TPCM to process claims under the
Plan as a third party administrator. TPCM failed to keep pace with the
claims received, creating a huge backlog of unprocessed claims. In
May 1997, Fidelity fired TPCM for incompetence and attempted to
transfer the claims management process to an in-house operation.
Assertedly, difficulties in obtaining data from TPCM and the massive
backlog it had created ultimately led to the Plan’s collapse.

  Approximately 1400 employees and their families contracted for
coverage under the Plan. As many as 2900 claims have gone unpaid,
amounting to millions of dollars in unpaid medical bills.

   In August 1998, Plaintiffs sued Fidelity Group, NABOP, IWG, and
the Fund (collectively, "the Fidelity Defendants") in state court in
South Carolina. The defendants removed the case to federal court.
                 GUNNELLS v. HEALTHPLAN SERVICES, INC.                    9
Later in the year, Plaintiffs amended their complaint to add TPCM
and the agents as defendants. In that complaint, Plaintiffs allege that
one or more of the defendants engaged in negligent undertaking,
fraud, negligent misrepresentation, breach of contract, civil conspir-
acy, and violated the South Carolina Unfair Trade Practices Act, S.C.
Code Ann. § 39-5-140 (1985), and Title IX of the Organized Crime
Control Act of 1970, 18 U.S.C.A. §§ 1961-68 (West 2000 & Supp.
2000) ("RICO").1 In short, Plaintiffs assert that the agents breached
contractual and fiduciary duties owed to the Plaintiffs and misrepre-
sented the Plan’s attributes through their marketing efforts, which
caused Plaintiffs to purchase the woefully deficient Plan. As to
TPCM, Plaintiffs allege that it mismanaged administration of the
Plan, created a huge backlog of unpaid claims, did not timely transfer
information, and made it impossible to forecast rate increases accu-
rately, resulting in the Plan’s collapse and failure to pay hundreds, if
not thousands, of health care claims. In their prayer for relief, Plain-
tiffs

      seek recovery for all legally allowable damages including
      refunds of premiums paid, payment of outstanding medi-
      cal/dental bills, reimbursement for medical bills paid by
      Plaintiffs where treatment occurred while Plaintiff[s were]
      covered by the Plan, compensation for injury to credit, com-
      pensation for injury to ability to obtain credit, time spent
      and costs incurred attending to the difficulties resulting from
      the failure of the Plan, loss of enjoyment of life attendant to
      the stress caused by the failure of the Plan.

Plaintiffs also seek punitive damages.
  1
   Also in 1998, the United States Department of Labor filed suit against
the Fidelity Group for actions relating to the Fund in the Eastern District
of New York. That court issued an order on January 8, 1999 staying all
pending actions against the Fund. Pursuant to this order, the district court
in South Carolina issued a stay of the proceedings in the present case.
After the New York court clarified that its stay reached only the Fidelity
Defendants, the district court in the case at hand lifted the stay with
respect to the "non-Fidelity Defendants" — TPCM and the agents. Ulti-
mately, on the advice of a court-appointed fiduciary, the New York court
ordered the Plan terminated effective January 31, 1999.
10               GUNNELLS v. HEALTHPLAN SERVICES, INC.
   After some initial discovery, Plaintiffs moved for certification of a
class, pursuant to Federal Rule of Civil Procedure 23(b)(3), of "all
entities and people who purchased the Fidelity Plan or who were pro-
vided with coverage under the Fidelity Plan in South Carolina at any
time." In addition, with respect to certain "agent-specific" claims,
Plaintiffs moved for certification of subclasses consisting of "all enti-
ties and persons who purchased the Fidelity Plan from or were pro-
vided the Fidelity Plan by that particular agent in South Carolina."

   After refusing to certify any claims under the South Carolina
Unfair Trade Practices Act, see S.C. Code Ann. § 39-5-140(a) (West
1985) (a person "may bring an action individually, but not in a repre-
sentative capacity, to recover actual damages"), the district court
granted Plaintiffs’ class certification motion in large part. Gunnells v.
Fidelity Group, Inc., No. 2:98-2639-23 (D.S.C. Sept. 28, 2001)
(located at J.A. 1244-1271). The court conditionally granted Plain-
tiffs’ motion for class certification with respect to their mismanage-
ment claim against TPCM. Moreover, although the court declined to
certify Plaintiffs’ civil conspiracy and RICO claims against the
agents, it did certify subclasses to permit Plaintiffs to pursue their four
theories of liability — negligent undertaking, fraud, negligent misrep-
resentation, and breach of contract — via separate class actions
against each of the twenty-three agents that assertedly sold insurance
to a named class representative, withholding "ruling on the certifica-
tion of the agent classes for which there [wa]s currently no represen-
tative." J.A. 1265 n.21.

   Pursuant to Federal Rule of Civil Procedure 23(f), both TPCM and
a number of agents2 (collectively, "the Agents") petitioned for permis-
sion to file an interlocutory appeal, which we granted.
  2
   Originally, fourteen agents appealed the class certification order.
While the appeal was pending, eight agents moved to withdraw their
appeal because they had reached a settlement with the class. We now
grant the eight agents’ motion to withdraw their appeal and remand to the
district court for consideration of the settlement proposal pursuant to
Federal Rule of Civil Procedure 23(e) and in light of this opinion. See
Fed. R. Civ. P. 23(e); see also Amchem Prods., Inc. v. Windsor, 

, 620 (1997) (articulating different standard for settlement-only class
certification). Our holding with respect to the district court’s certification
of subclasses against individual agents, therefore, pertains only to the six
agents whose appeals remain before us.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                  11
                                   II.

   Class actions must meet several criteria. First, the class must com-
ply with the four prerequisites established in Rule 23(a):
(1) numerosity of parties; (2) commonality of factual and legal issues;
(3) typicality of claims and defenses of class representatives; and
(4) adequacy of representation. Fed. R. Civ. P. 23(a). Second, the
class action must fall within one of the three categories enumerated
in Rule 23(b); here Plaintiffs seek to proceed under Rule 23(b)(3),
which requires that common issues predominate over individual ones
and that a class action be superior to other available methods of adju-
dication. Fed. R. Civ. P. 23(b)(3).

   If a lawsuit meets these requirements, certification as a class action
serves important public purposes. In addition to promoting judicial
economy and efficiency, class actions also "afford aggrieved persons
a remedy if it is not economically feasible to obtain relief through the
traditional framework of multiple individual damage actions." 5
James Wm. Moore et al., Moore’s Federal Practice § 23.02 (3d ed.
1999). Thus, federal courts should "give Rule 23 a liberal rather than
a restrictive construction, adopting a standard of flexibility in applica-
tion which will in the particular case ‘best serve the ends of justice
for the affected parties and . . . promote judicial efficiency.’" In re
A.H. Robins, 

, 740 (4th Cir. 1989).

   To be sure, Rule 23(b)(3) class actions must meet predominance
and superiority requirements not imposed on other kinds of class
actions. This is because these suits involve situations where "class-
action treatment is not as clearly called for." Fed. R. Civ. P. 23 advi-
sory committee’s note (1966 Amendment, subdivision (b)(3)). How-
ever, as the Supreme Court has noted, the predominance and
superiority requirements in Rule 23(b)(3) do not foreclose the possi-
bility of mass tort class actions, but merely ensure that class certifica-
tion in such cases "achieve economies of time, effort, and expense,
and promote . . . uniformity of decision as to persons similarly situ-
ated, without sacrificing procedural fairness or bringing about other
undesirable results." 

(quoting Adv. Comm.
Notes, 28 U.S.C. App. at 697). For these very reasons, we have
expressly "embraced the view that the mass tort action for damages
may be appropriate for class action, either partially or in whole." Cen-
12               GUNNELLS v. HEALTHPLAN SERVICES, INC.
tral Wesleyan Coll. v. W.R. Grace & Co., 

, 185 (4th Cir.
1992) (citation, internal quotation marks, ellipses, and alterations
omitted).

   Furthermore, "[d]istrict courts have wide discretion in deciding
whether or not to certify a class and their decisions may be reversed
only for abuse of discretion," 

marks omitted),
recognizing, of course, that this "discretion must be exercised within
the framework of Rule 23." Lienhart v. Dryvit Sys., Inc., 

, 146 (4th Cir. 2001) (quoting In re Am. Med. Sys., Inc., 

, 1079 (6th Cir. 1996)). Our review is particularly deferential in
a case like this — involving an interlocutory appeal of a conditional
class certification. See Central 

(noting that
"[t]he tentative, limited nature of the conditional certification . . .
counsels in favor of affirmance.").3

   With these principles in mind, we turn to the case at hand. Plain-
tiffs’ claims against TPCM and the Agents rest on different bases.
Accordingly, like the district court, we consider class certification
with respect to TPCM and the Agents separately.

                                    III.

   The district court conditionally certified the plaintiff class against
TPCM allowing the class to pursue a single claim: "that TPCM vio-
lated its duties, both contractual and at law, as third party administra-
tor of the Plan and such conduct was a cause of the Fidelity Plan’s
failure," and that "failure directly injured Plaintiffs and the absent
class members." J.A. 1249.
  3
    Thus, in paying heed to "the ‘conditional’ aspect of [the] certification"
here, we do not, as the dissent asserts, "default[ ] our responsibility" or
"employ[ ] a backward burden." See post at 83 n.8. Rather, we simply
follow Central Wesleyan’s well-established directive. The dissent not
only ignores that directive, it also contends (without explanation) that
other "portions" of Central Wesleyan may have been "superseded" by
Amchem. See post at 86. We disagree. In fact, in Central Wesleyan, the
court carefully recognized and sought to avoid the problems created by
the lower court in Amchem. See Central 

, 189
(noting that "the Third Circuit’s ongoing experiment in class certification
[in the Amchem litigation] has not been an unqualified success").
               GUNNELLS v. HEALTHPLAN SERVICES, INC.                13
                                  A.

   We note at the outset both that the certification order concerning
TPCM is strikingly similar to the certification order we upheld in
Central Wesleyan and that the district court carefully followed the
framework approved in Central 

. Here,
as in Central Wesleyan, "the district court conducted a straightfor-
ward analysis of the class certification requirements under Fed. R.
Civ. P. 23(a) and 23(b)(3)," determined that each requirement had
been met, and "supported each of its 23(a) and 23(b)(3) holdings with
detailed findings." 

186.

   First, the district court found that the "1400 employees plus their
families" covered by the Plan, with possibly 2900 unpaid claims,
"easily" satisfied Rule 23(a)(1)’s numerosity requirement. J.A. 1248.
Cf. Central 

(noting district court’s finding
"that some 480 potential class members would easily satisfy the
numerosity requirement"). The court found Rule 23(a)(2) commonal-
ity and Rule 23(a)(3) typicality because of the many "factual and legal
issues pertaining to TPCM’s conduct . . . common" to "all Plaintiffs
and class members," such as: "(1) facts surrounding how TPCM man-
aged its operations; (2) whether it created a backlog of claims; ([3])
whether it failed to enter claims into its system; ([4]) whether it
breached its contractual duties; [and (5)] whether TPCM’s alleged
negligence proximately caused Plaintiffs’ injuries by contributing to
the failure of the Plan." J.A. 1254-55. The district court recognized
that some of the Plaintiffs’ damages claims might require "individual
inquiry" but found that, if necessary, those issues could be bifurcated
for individual trials as they were in Central 

-
90. See J.A. 1254. With regard to Rule 23(a)(4) adequacy, the court
recognized that "class representatives" must "be part of the same class
and possess the same interest and suffer the same injury as the class
members" and found that in this case the named plaintiffs satisfied
these requirements, noting particularly that no "potential conflict
existed" among class members. 

cf. Broussard v. Meineke,

, 337-38 (4th Cir. 1998) (holding "manifest conflicts of
interest" among members of class precludes class certification).

   As in Central 

-84, the district court next
turned to the Rule 23(b)(3) requirement that "questions of law or fact
14              GUNNELLS v. HEALTHPLAN SERVICES, INC.
common to members of the class predominate over any questions
affecting only individual members, and that a class action [be] supe-
rior to other available methods to the fair and efficient adjudication
of the controversy," and "specifically addressed the four matters the
Rule lists as pertinent to these findings." First, the court found that the
class members had evidenced "no great deal of interest . . . in individ-
ual litigation," noting as support for this finding that although "1400
potential litigants exist," there was a "dearth of individual cases filed
until now." J.A. 1260. Second, the court held that in view of the few
pre-existing cases, a class action would be productive. 

Third, the court concluded that "the flexibility and control of this
action in one forum [the federal district court in South Carolina] pro-
vides the benefits which outweigh the method of individual litiga-
tion." 

   Finally, and again as in Central 

, the dis-
trict court acknowledged "particular concern" with the fourth factor:
"the difficulties likely to be encountered in the management of a class
action." Fed. R. Civ. P. 23(b)(3)(D). The court found that bifurcation
of a few issues (e.g., some damages issues) requiring individualized
determination would assist it in managing the class action, but recog-
nized that, even so, "given the number of possible class members in
this action . . . manageability could quickly dissipate." J.A. 1260.
Accordingly, the district court resolved only to grant conditional certi-
fication of the class and explained that it would not hesitate to decer-
tify the class if it should prove unmanageable. 

(We
note that in Central 

-89, we upheld a district
court’s conditional certification order in the face of far more formida-
ble management concerns, i.e., not only the possible need for individ-
ualized damage determinations, but also a "daunting number of
individual issues" necessary to "establish liability," including determi-
nation of "comparative fault" among numerous defendants,
"[d]ifferent time bar defenses" with respect to differently positioned
defendants, and the necessity of application of the laws of different
jurisdictions.)

   Thus, here, as in Central Wesleyan, the district court carefully con-
sidered the requirements of Rule 23(a) and (b)(3) and made specific
and detailed findings as to compliance with those requirements. In
doing so, the court followed the admonition in A.H. Robins, 880 F.2d
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                15
at 740, to "take full advantage of the provision in [Rule 23(c)(4)] per-
mitting class treatment of separate issues . . . to reduce the range of
disputed issues" in complex litigation, and most especially the teach-
ings of Central Wesleyan itself, and then exercised its discretion to
grant conditional class certification.

   As in Central Wesleyan, we must acknowledge that "[s]uch an
approach has its 

. First, it appears likely that
in the absence of class certification, very few claims would be
brought against TPCM, making "the adjudication of [the] matter
through a class action . . . superior to no adjudication of the matter
at all." See 5 Moore’s Federal Practice § 23.48[1] (1997). Thus, class
certification will provide access to the courts for those with claims
that would be uneconomical if brought in an individual action. As the
Supreme Court put the matter, "[t]he policy at the very core of the
class action mechanism is to overcome the problem that small recov-
eries do not provide the incentive for any individual to bring a solo
action prosecuting his or her rights." 

(cita-
tion omitted).

   Moreover, as in Central Wesleyan, class certification in this case
seems likely to reduce litigation costs through the "consolidation of
recurring common 

. Such recurring common
issues make up the heart of Plaintiffs’ case against TPCM, i.e.,
whether TPCM mismanaged the Fund, whether TPCM’s mismanage-
ment was a proximate cause of the Plan’s collapse, whether the Fidel-
ity Defendants also mismanaged the Fund, and whether the Fidelity
Defendants’ mismanagement was an intervening cause of the Plan’s
collapse sufficient to relieve TPCM of liability. Proving these issues
in individual trials would require enormous redundancy of effort,
including duplicative discovery, testimony by the same witnesses in
potentially hundreds of actions, and re-litigation of many similar, and
even identical, legal issues.

   Consolidation of these recurring common issues will also conserve
important judicial resources. The record in this case already reveals
that a trial on these common issues alone is likely to be lengthy,
requiring extensive exploration of the Plan’s collapse and TPCM’s
role. The prospect of a long trial, with the possibility of a large dam-
age award may also encourage settlement, which "should be a factor,
16              GUNNELLS v. HEALTHPLAN SERVICES, INC.
and an important factor, to be considered when determining certifica-
tion." 

(quoting A.H. 

).

   Furthermore, class certification "provides a single proceeding in
which to determine the merits of the plaintiffs’ claims, and therefore
protects the defendant from inconsistent adjudications." 5 Moore’s
Federal Practice § 23.02 (1999) (emphasis added). This protection
from inconsistent adjudications derives from the fact that the class
action is binding on all class members. See Fed. R. Civ. P.
23(c)(2)(B). By contrast, proceeding with individual claims makes the
defendant vulnerable to the asymmetry of collateral estoppel: If
TPCM lost on a claim to an individual plaintiff, subsequent plaintiffs
could use offensive collateral estoppel to prevent TPCM from litigat-
ing the issue. See Parklane Hosiery Co. v. Shore, 

, 331
(1979) (concluding that the use of offensive collateral estoppel should
not be precluded in federal courts). A victory by TPCM in an action
by an individual plaintiff, however, would have no binding effect on
future plaintiffs because the plaintiffs would not have been party to
the original suit. See Allen v. McCurry, 

, 95 (1980)
("[T]he concept of collateral estoppel cannot apply when the party
against whom the earlier decision is asserted did not have a ‘full and
fair opportunity’ to litigate that issue.") (citing Montana v. United
States, 

, 153 (1979)). Class certification thus promotes
consistency of results, giving defendants the benefit of finality and
repose.4

  Indeed, it appears at this juncture that the single claim certified
against TPCM — mismanagement of the Plan resulting in its ultimate
collapse similarly affecting a large group of people in a single state
— seems even better suited for class treatment than the diverse issues
against many defendants involving the laws of a number of states that
we certified in Central Wesleyan.
  4
    Thus, TPCM benefits more from class certification than the Plaintiffs
in some ways, since the effect of class certification on collateral estoppel
redounds to the benefit of defendants, not plaintiffs, whereas the benefi-
cial economies of class certification redound to all parties, as well as to
the courts. One suspects, therefore, that TPCM resists certification in an
attempt to keep Plaintiffs with relatively small claims out of court alto-
gether — precisely the problem the class action mechanism was designed
to address. See 

.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                    17
                                    B.

   Notwithstanding these substantial advantages to class certification,
TPCM, like the defendants in Central Wesleyan, "argue[s] that the
district court committed multiple errors in conditionally certifying the

. We disagree.

                                     1.

   TPCM’s principal argument, which it sounds again and again, is
that the probable necessity of individual proof of some of the claimed
damages by individual class members prevents Plaintiffs from meet-
ing Rule 23’s requirements. Initially, TPCM contends that individual-
ized damages determinations destroy commonality, typicality, and
predominance.5 But Rule 23 contains no suggestion that the necessity
for individual damage determinations destroys commonality, typical-
ity, or predominance, or otherwise forecloses class certification. In
fact, Rule 23 explicitly envisions class actions with such individual-
ized damage determinations. See Fed. R. Civ. P. 23 advisory commit-
tee’s note (1966 Amendment, subdivision (c)(4)) (noting that Rule 23
(c)(4) permits courts to certify a class with respect to particular issues
and contemplates possible class adjudication of liability issues with
"the members of the class . . . thereafter . . . required to come in indi-
vidually and prove the amounts of their respective claims."); see also
5 Moore’s Federal Practice § 23.23[2] (1997) ("[T]he necessity of
making an individualized determination of damages for each class
member generally does not defeat commonality.").
  5
   TPCM couches this argument as one involving "due process" but cites
no authority indicating that this concept applies in this context. In any
event, TPCM contends that "due process" requires the district court to
conduct an individual inquiry to determine the benefit amount to which
each class member was rightfully entitled and if a member was not
deprived of any benefit as a result of the Plan’s collapse, then TPCM
could not be said to be the proximate cause of that member’s injury. Sub-
stantively, this argument is identical to the contention, discussed in text,
that, TPCM must be allowed the opportunity to show that an individual
class member did not suffer any damages. As explained above, affording
TPCM this opportunity does not destroy Plaintiffs’ ability to satisfy the
requirements of Rule 23.
18              GUNNELLS v. HEALTHPLAN SERVICES, INC.
   Indeed, "[i]n actions for money damages under Rule 23(b)(3),
courts usually require individual proof of the amount of damages each
member incurred." 

23.46[2][a] (1997) (emphasis added).
When such individualized inquiries are necessary, if "common ques-
tions predominate over individual questions as to liability, courts gen-
erally find the predominance standard of Rule 23(b)(3) to be
satisfied." 

precisely the situation here — "common questions [do] pre-
dominate over individual questions as to liability." 

and the district court certified, class claims against TPCM
that rest on a single theory: that TPCM’s mismanagement of claims
contributed to the ultimate collapse of the Plan and so caused Plain-
tiffs’ damage. Whether TPCM breached its contract to administer and
timely pay claims under the Plan, whether Plaintiffs are third party
beneficiaries of that contract, whether TPCM otherwise owed a fidu-
ciary duty of care to Plaintiffs, whether it breached that duty, whether
it mismanaged claims at all, and whether its mismanagement contrib-
uted to the ultimate collapse of the Plan — are issues common to all
potential class members, and do not require any individualized
inquiry.

   Of these issues, the only one that TPCM even contends is not com-
mon to all class members is that of causation. TPCM argues that an
individualized inquiry is necessary to determine whether its misman-
agement proximately caused Plaintiffs’ injury, arguing that Plaintiffs
"provided absolutely no credible evidence that any TPCM action or
inaction proximately caused damages to them." TPCM’s argument
fails for several reasons.

   First, the sufficiency of the evidence as to proximate cause pre-
sented by Plaintiffs goes to the merits of Plaintiffs’ case — an issue
the Supreme Court has held courts may not consider in ruling on a
motion for class certification. See Eisen v. Carlisle & Jacquelin, 

, 177 (1974).

   Second, to the extent that TPCM argues that it was not the proxi-
mate cause of the Plan’s collapse, the argument actually provides fur-
ther support for proceeding with this case as a class action. As
discussed earlier, class treatment will avert the necessity for hundreds
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                 19
of individual trials in which core issues — such as TPCM’s assertion
that it was not a proximate cause of the Plan’s collapse — would be
re-litigated, requiring testimony by the same witnesses, duplicative
discovery efforts, etc. If TPCM succeeds in establishing that it was
not the proximate cause of the Plan’s collapse, it will enjoy the pre-
clusive effects of this decision against all class members, rather than
be disadvantaged by the asymmetric application of collateral estoppel.
Compare Fed. R. Civ. P. 23(c)(2)(B) (making class action binding on
all members) with Allen v. 

(noting that col-
lateral estoppel "cannot apply when the party against whom the ear-
lier decision is asserted did not have a ‘full and fair opportunity’ to
litigate that issue" (citations omitted)).

    Third, to the extent that TPCM’s causation argument is that indi-
vidual inquiry is necessary to establish whether the collapse of the
Plan caused Plaintiffs any damages, this is precisely the same argu-
ment made by almost all defendants in mass tort cases: determining
damages will require an individualized inquiry. Courts have routinely
rejected this argument, concluding, as we have in previous cases, that
the need for individualized proof of damages alone will not defeat
class certification. See Central 

; Hill v. W.
Elec. Co., Inc., 

, 387 (4th Cir. 1981) ("Bifurcation of . . .
class action proceedings for hearings on . . . damages is now com-
monplace."); Chisolm v. TranSouth Fin. Corp., 

, 566
(E.D. Va. 1999) (collecting cases). As one court explained, "Quantita-
tively, almost by definition there will always be more individual dam-
ages issues than common liability issues . . . . Qualitatively, however,
. . . liability issues" may "far exceed in complexity the more mundane
individual damages issues." In re Honda Motor Co., 

, 367 (D. Md. 1997); see also In re Amer. Med. Sys., Inc., 

, 1080 (6th Cir. 1996) (stating that commonality test is "qualita-
tive rather than quantitative" (citation omitted)). So it is here. The
possibility that individualized inquiry into Plaintiffs’ damages claims
will be required does not defeat the class action because common
issues nevertheless predominate.

   Moreover, the damage calculations in this case do not appear to be
particularly complex, unlike the calculations in those cases in which
courts have found that numerous complicated individual damage
inquiries predominated over common issues. Cf. Windham v. Am.
20               GUNNELLS v. HEALTHPLAN SERVICES, INC.
Brands, Inc., 

, 66-67, 72 (4th Cir. 1977) (affirming denial
of class certification because of overwhelming predominance of thou-
sands of individualized damage questions). In fact, Plaintiffs’ claims
for punitive damages do not require any individualized inquiry at all
because this damage calculation would be based solely on TPCM’s
conduct. In addition, it appears that a court-appointed fiduciary has
already calculated most of the Plaintiffs’ claims for medical bills
using the same computer program that TPCM and Fidelity used to
evaluate claims. Thus, in many cases there is likely to be little or no
dispute over the amount to which an individual class member is enti-
tled for unpaid medical claims. Similarly, if the court determines that
Plaintiffs are entitled to a refund of premiums paid, these amounts
could be quickly and easily determined.

   Plaintiffs’ other damages claims — those relating to injury to
credit, time lost, and loss of enjoyment of life — may require individ-
ualized inquiry. But even considering Plaintiffs’ claims against
TPCM as a whole, we cannot conclude that the district court abused
its discretion in finding that at this juncture common issues predomi-
nate over the individual issues that may be involved in resolving these
claims. We note that the district court was fully aware of the possibil-
ity that individual issues, although currently subordinate to the com-
mon liability and damage issues, might in time predominate and the
case prove unmanageable. The court expressly indicated its resolve to
deal with this if it became a problem, and for this reason agreed only
to certify the class action conditionally. Such care reflects assiduous
respect for Rule 23’s requirements, not an abuse of discretion.6

                                     2.

   TPCM’s remaining arguments have even less merit and require lit-
tle discussion. Specifically, TPCM argues that the named plaintiffs
constitute inadequate class representatives because they purportedly
  6
    TPCM reasserts its individualized damages argument when contend-
ing that the certified class lacks numerosity and superiority. The argu-
ment is no more persuasive in those contexts. Indeed, TPCM’s repeated
attempts to reprise the same argument testify to the fundamental weak-
ness of its contention that the conditional class certification order consti-
tuted an abuse of discretion.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                21
lack sufficient knowledge, and because alleged insurmountable con-
flicts exist within the class. TPCM also contends that the interest of
certain individual class members in pursuing individual litigation
defeats any argument that a class action provides a superior vehicle
for adjudication of the claims. These arguments also fail.

   The lack of knowledge contention is particularly meritless. It is
hornbook law, as the district court recognized, that "[i]n a complex
lawsuit, such as one in which the defendant’s liability can be estab-
lished only after a great deal of investigation and discovery by coun-
sel against a background of legal knowledge, the representative need
not have extensive knowledge of the facts of the case in order to be
an adequate representative." J.A. 1256 (quoting 32B Am. Jur. 2d Fed-
eral Courts § 1888 (1996) (footnotes omitted)); see also Baffa v. Don-
aldson, Lufkin & Jenrette Secs. Corp., 

, 61 (2d Cir. 2000)
("The Supreme Court in Suorowitz v. Hilton Hotels Corp., 

, 370-74 (1966) expressly disapproved of attacks on the adequacy
of a class representative based on the representative’s ignorance.").

   TPCM’s argument as to insurmountable class conflicts fares no
better. According to TPCM, the interests of the employer class mem-
bers conflict with those of the employee class members because
employees may wish to sue their employers for unpaid medical bills
resulting from the collapse of the Plan.

   To defeat the adequacy requirement of Rule 23, a conflict "must be
more than merely speculative or hypothetical." 5 Moore’s Federal
Practice § 23.25[4][b][ii] (2002). In this case, no conflict exists that
renders the named representatives inadequate. TPCM has not cited a
single case in support of its contention that the employees would have
viable claims against their employers for unpaid medical bills caused
by the failure of a non-ERISA plan, no employee has filed such a
claim, and South Carolina’s three-year statute of limitations now bars
such claims. See 7 S.C. Code Ann. § 15-3-530 (West Supp. 2002)
(applying three-year statute of limitations to actions for, among other
things, breach of contract and fraud).

   Furthermore, even if an employee could bring such a claim against
its employer, this would not require decertification. For a conflict of
interest to prevent plaintiffs from meeting the requirement of Rule
22              GUNNELLS v. HEALTHPLAN SERVICES, INC.
23(a), that conflict "must be fundamental. It must go to the heart of
the litigation." 6 Alba Conte & Herbert B. Newberg, Newberg on
Class Actions § 18:14 (4th ed. 2002). Here, the employers and
employees share common objectives and the same factual and legal
positions. They have the same interest in establishing the liability of
TPCM. Thus, any potential conflict does not go to the heart of their
roles as class representatives. See Uniondale Beer Co. v. Anheuser-
Busch, Inc., 

, 343 (E.D.N.Y. 1987) (finding that eco-
nomic conflict between competitor class members in antitrust action
did not render representatives inadequate because all class members
sought to prove that defendants conspired to fix prices).7

   In addition, the dissent suggests a conflict between those Plaintiffs
who were not injured by the collapse of the Plan, but have "direct"
claims against TPCM for a delay in payment, and those Plaintiffs with
"indirect" claims against TPCM that derive from the collapse of the
Plan. Post at 77-81. As the Plaintiffs have made abundantly clear on
appeal, however, they do not seek class certification for any direct
claims against TPCM. See, e.g., Brief of Appellees at 14 ("Contrary
to TPCM’s assertions, Plaintiffs do not claim that TPCM improperly
adjudicated particular claims. TPCM injured all Plan insureds because
it was the cause of the collapse of the Plan."). To be sure, this position
appears somewhat at odds with the Plaintiffs’ original complaint.
However, "[t]he federal rules effectively abolish the restrictive theory
of the pleadings doctrine," which held that "a complaint must proceed
upon some definite theory, and on that theory the plaintiff must suc-
ceed, or not succeed at all." 5 Charles Alan Wright & Arthur R. Mil-
  7
   The dissent insists that an additional conflict exists between the
employer class members and the employee class members based on the
different remedies sought by each group. See post at 84-85. However,
"[p]otential conflicts relating to relief issues which would arise only if
the plaintiffs succeed on common claims of liability on behalf of the
class will not bar a finding of adequacy." Hanrahan v. Britt, 

, 364 (E.D. Pa. 1997) (internal quotation marks and citation omitted);
see Bogosian v. Gulf Oil Corp., 

, 449 (3d Cir. 1977) (find-
ing that representation was adequate even though current gas station les-
sees and former lessees would ultimately seek different remedies); see
generally 1 Newberg on Class Actions § 3:25, at 422-23 & n.4 (collect-
ing cases). Therefore, any conflict in the forms of recovery sought by
these groups of plaintiffs does not warrant decertification.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   23
ler, Federal Practice & Procedure § 1219, at 188-89 (2d ed. 1990)
(internal quotation marks and citation omitted). Cf. post at 80 ("[W]e
should take the complaint as written."). This observation seems par-
ticularly apropos to actions proceeding under Rule 23, which provides
that "an action may be brought or maintained as a class action with
respect to particular issues." Fed. R. Civ. P. 23(c)(4)(A) (emphasis
added).

   Furthermore, even if the Plaintiffs wanted to proceed with direct
claims against TPCM, the only claim that the district court certified
against TPCM is based on the theory that TPCM caused the collapse
of the Plan. See J.A. 1249-50 ("Plaintiffs have alleged that TPCM’s
mismanagement of processing claims contributed to the failure of the
Fidelity Plan. . . . Plaintiff’s allegations, if proven, will vitiate any
need to demonstrate that a particular claim was submitted prior to the
expiration of the contract between TPCM and Fidelity."). Thus, Plain-
tiffs’ counsel did not so much make "a questionable decision . . . to
abandon," a theory of liability on appeal, as the dissent suggests, see
post at 79, as accept the scope of the claim that the district court certi-
fied for class treatment.

   Finally, the dissent errs in suggesting that the district court’s order
"precludes recovery for those who were in the circumstance where
the collapse of the Plan did not cause injury." 

dissent’s
argument seems premised on the false assumption that under the dis-
trict court order the Plaintiffs’ direct claims would be barred by the
rule against claim-splitting. But a class action, "of course, is one of
the recognized exceptions to the rule against claim-splitting." 18
Moore’s Federal Practice § 131.40[3][e][iii] (2002) (citing Restate-
ment (Second) of Judgments § 26(1)(c) (1982)). Second, Rule
23(c)(2) permits members of a class maintained under section (b)(3)
to opt out of the class, providing an option for those Plaintiffs who
wish to pursue claims against TPCM requiring more individualized
inquiry. Thus, rhetoric aside, Plaintiffs with only direct claims are not
being "[j]amm[ed]," "sacrificed" or "caught" in any class action
against their will. Post at 79.8
  8
   We find somewhat disingenuous the dissent’s suggestion that we have
"done a disservice to an identifiable group of plaintiffs," by exposing
24              GUNNELLS v. HEALTHPLAN SERVICES, INC.
   TPCM’s final argument is that class members’ asserted "interest in
individual litigation of the claims" and purported possession of "large
damage claims" renders their pursuit of individual actions superior to
certification of a class action. This entire argument rests on one fact
— that a single named plaintiff, Mary B. Gunnells, claims $70,000 in
unpaid medical expenses — and one contention — that "at least 17
actions [have been] filed in South Carolina involving the sale and
eventual failure of the Fidelity Plan." The district court carefully con-
sidered and acted well within its discretion in rejecting this argument.

   The court found that while Ms. Gunnells claims $70,000 in unpaid
medical expenses, "not all of the Plaintiffs’ alleged damages are so
high and the damages could run the gamut to much lower figures."
J.A. 1260. Nothing in the record refutes this finding and we can take
judicial notice of the fact that many health care claims will involve
dollar amounts that are much lower. Moreover, we note that the size
of class members’ recovery is hardly determinative of superiority. See
5 Moore’s Federal Practice § 23.48[2][a] (1997) (noting that when "a
class action will provide the most fair and efficient adjudication of a
case, such an action may be superior even though class members have
sufficient means or incentive to proceed individually"). As to the sev-
enteen already-commenced actions, the district court concluded that
their existence did not make individual actions superior in a case
involving over 1400 potential litigants. Again, nothing in the record
in any way contradicts this common-sense finding. TPCM itself notes

them to arguments that their claims are now untimely. Post at 80 n.7.
First, Plaintiffs with direct claims against TPCM would be equally sub-
ject to these arguments under the dissent’s proposed resolution of the
case. Second, while the dissent is, of course, correct that TPCM could
raise a statute of limitations defense against Plaintiffs asserting direct
claims, it is far from clear that TPCM would be successful. Cf. Crown,
Cork & Seal Co., Inc. v. Parker, 

, 353-54 (1983) (holding
that "‘the commencement of a class action suspends the applicable stat-
ute of limitations as to all asserted members of the class who would have
been parties had the suit been permitted to continue as a class action,’"
and that "[o]nce the statute of limitations has been tolled, it remains
tolled for all members of the putative class until class certification is
denied" (quoting Am. Pipe & Constr. Co. v. Utah, 

, 554
(1974))).
                 GUNNELLS v. HEALTHPLAN SERVICES, INC.                    25
that it "has not been named as a Defendant in a single one of those
[seventeen] actions," thus raising questions as to whether the exis-
tence of those actions is even relevant to class certification against
TPCM. Accordingly, we can hardly conclude that in rendering these
findings the district court abused its discretion.9

                                     C.

   In sum, the district court followed the course we charted in Central
Wesleyan. Its careful analysis of Rule 23’s requirements and its
detailed factual findings supporting a decision to grant conditional
certification against TPCM reveal no abuse of discretion. Of course,
we recognize, as the district court did, that this case may ultimately
present manageability problems. Nonetheless, we cannot say at this
interlocutory stage that the decision to conditionally certify a class
action against TPCM constitutes an abuse of discretion, particularly
in light of the special deference due trial courts on this issue. See Cen-
tral 

("Issues of class action manageability are
properly committed to the district court’s discretion, because that
court ‘generally has a greater familiarity and expertise’ with the ‘prac-
tical . . . and primarily . . . factual’ problems of administering a law-
suit ‘than does a court of appeals.’" (quoting 

)). If the district court’s current assessment turns out to be inaccu-
rate, and it becomes apparent that individual damage issues will pre-
dominate or render the case unmanageable, the court undoubtedly will
recognize its responsibility to decertify the class.
  9
    The dissent’s effort to analogize this case to Lienhart, and to assert
that we ignore "the complex question of comparative fault" arises from
the same misperception regarding separate Plaintiff classes that animates
the dissent’s argument with respect to conflicts among Plaintiff classes.
See post at 80, 86. Because there is only a single Plaintiff class with a
single cause of action — that TPCM’s mismanagement contributed to the
demise of the Plan — issues of comparative fault will require no individ-
ual inquiry apart from that required to calculate damages. The extent to
which TPCM’s conduct contributed to the collapse of the Plan, and the
comparative fault of intervening parties, such as Fidelity, will be the
same for every Plaintiff. In 

, by contrast, the
intervening actors and the intervening acts were different for every plain-
tiff, thereby creating the possibility that the defendant’s degree of liabil-
ity would be different with respect to every plaintiff.
26               GUNNELLS v. HEALTHPLAN SERVICES, INC.
   We note also that in conditionally certifying the class the court spe-
cifically stated that it intended to "heed[ ] the Fourth Circuit’s warn-
ing to keep a watchful vigil in a complicated case such as this one to
‘make certain that manageability and other types of problems do not
overwhelm the advantage of conditional certification.’" J.A. 1260
(quoting Central 

). Indeed, the court expressly
noted that "should such concerns render the class mechanism ineffec-
tive, the district court must be prepared to use its considerable discre-
tion to decertify the class." 

We are confident that the
court will do precisely this, if necessary.10

                                     IV.

   The district court also granted, in substantial part, Plaintiffs’
motion for class certification against the individual agents responsible
for marketing and selling the Plan. Although the court refused to cer-
tify a class as to Plaintiffs’ conspiracy and RICO claims, concluding
that individual issues predominated, it did conditionally certify Plain-
tiffs’ claims based on four other theories of liability: negligent under-
taking, fraud, negligent misrepresentation, and breach of contract. The
court organized its certification of these claims into subclasses — one
for each of the twenty-three agents who had marketed insurance to a
named plaintiff. Each subclass was permitted to pursue its four theo-
ries of liability via a separate class action.

   Once again, we accord the district court’s class certification deci-
sion substantial deference. Nevertheless, in this instance, we believe
  10
     TPCM makes two ancillary arguments that we need address only
briefly. First, it asserts that the district court improperly relied on affida-
vits of principals of the Fidelity Defendants, whom TPCM could not
examine because it was not a defendant in this case when the affiants
were deposed, and the New York court later stayed all proceedings
against the Fidelity Defendants. This argument rests on a faulty premise,
i.e., that TPCM was unable to depose the affiants. In fact, we do not
know whether TPCM could have deposed the affiants because it never
sought to do so. Second, TPCM asserts that the district court’s ruling
rested on erroneous facts. Even if it could show that these factual find-
ings were "clearly erroneous," it has utterly failed to demonstrate that
any of these errors affected its substantial rights. See Fed. R. Civ. P. 61.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   27
the court abused that discretion. The individual class actions certified
against the Agents simply cannot satisfy Rule 23’s predominance and
commonality requirements.11 This is so because Plaintiffs’ claims
against the individual Agents require proof of elements not necessary
to their single claim against TPCM; and a number of these additional
elements can only be established after considerable individual
inquiry. Although the district court apparently recognized this as a
general matter, it misunderstood or ignored controlling legal princi-
ples in holding that in this case such individual determinations would
not be necessary and so common issues would predominate. In doing
so, "the district court abuse[d] its discretion by . . . misapprehending
the law with respect to underlying issues." See Quince Orchard Val-
ley Citizens Ass’n v. Hodel, 

, 78 (4th Cir. 1989) (internal
quotation marks and citations omitted).

                                    A.

   The first reason why common issues do not predominate in Plain-
tiffs’ claims against the Agents is the need for individual inquiry into
the issue of reliance. Indisputably, negligent misrepresentation and
fraud require proof of reliance. See Robertson v. First Union Nat’l
Bank, 

, 313-14 (S.C. Ct. App. 2002) (recounting ele-
ments of negligent misrepresentation and fraud claims under South
Carolina law). To establish fraud, a plaintiff must prove, inter alia,
that the hearer relied on the truth of a fraudulent statement, and had
a right to rely thereon. 

negligent misrepresentation,
a plaintiff must prove, inter alia, that the plaintiff "justifiably relied
  11
    In addition, we note our reservations about the district court’s deter-
mination as to compliance with the numerosity requirement, see J.A.
1265 (noting that some Agent subclasses have as few as eight class mem-
bers), and the superiority requirement, compare J.A. 1271 (holding the
South Carolina Department of Insurance’s current administrative pro-
ceeding to revoke the Agents’ licenses has "no bearing . . . on this pro-
ceeding") with 5 Moore’s Federal Practice § 23.49[3] (1997) (noting
that "if a governmental unit has brought suit on the same issue, a court
may decide that the proposed class action is unnecessary and an inferior
method of adjudication"). We need not address these issues, however,
because all of Rule 23’s requirements must be met. Therefore, our con-
clusion that the class fails to meet the commonality and predominance
requirements is dispositive in this case.
28              GUNNELLS v. HEALTHPLAN SERVICES, INC.
on [a false] representation; and . . . the plaintiff suffered a pecuniary
loss as the proximate result of his reliance upon the representation."

Thus, both theories of liability require that the plaintiff
demonstrate (1) actual reliance and (2) that the plaintiff had a right to
rely on the misstatement (in the case of fraud) or that such reliance
was justifiable under the circumstances (in the case of negligent mis-
representation). Almost inevitably, establishing these elements
requires an individualized inquiry.

   As we held in Broussard, "the reliance element of . . . fraud and
negligent misrepresentation claims [is] not readily susceptible to
class-wide proof;" rather, "proof of reasonable reliance . . . depend[s]
upon a fact-intensive inquiry into what information each [plaintiff]
actually had." 

. Nor are we alone in so
holding. See, e.g., Andrews v. AT&T, 

, 1025 (11th Cir.
1996) (decertifying class in part because "the plaintiffs would . . .
have to show, on an individual basis, that they relied on the misrepre-
sentations, suffered injury as a result, and incurred a demonstrable
amount of damages"); Castano v. Am. Tobacco Co., 

, 745
(5th Cir. 1996) (concluding that "a fraud class action cannot be certi-
fied when individual reliance will be an issue").

   The district court nevertheless found that the required proof of reli-
ance in this case would not require an individual "fact-intensive
inquiry" sufficient to render class certification improper, but instead
found that actual reliance could be "presumed" and the right to rely
"assumed." Those findings are grounded on a misunderstanding of
controlling law.

                                   1.

   First, the district erred in suggesting that Basic Inc. v. Levinson,

(1988), supports a presumption of actual reliance in this
case. See J.A. 1268 and n.27 ("[A] common sense approach presum-
ing reliance exists may be acceptable in this circumstance.") (citing

). In fact, neither Basic, nor any other prece-
dent, supports a presumption of actual reliance in the case at hand.

  Basic involved a shareholders’ suit alleging that a company made
material misrepresentations about the status of merger talks. In oppos-
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   29
ing class certification, the company argued that common issues did
not predominate because each shareholder had to establish reliance on
the misrepresentation in making decisions to buy or sell the stock.
The Supreme Court rejected this argument, reasoning that "[b]ecause
most publicly available information is reflected in market price, an
investor’s reliance on any material misrepresentations . . . may be pre-
sumed for purposes of a Rule 10b-5 action." 

.
This presumption was based on the "fraud-on-the-market theory," i.e.,
even if an individual investor was not aware of the misrepresentation
(and therefore did not directly "rely" on it in the traditional sense), the
market price of the stock reflected the misrepresentation, thereby pro-
viding "the requisite causal connection between a defendant’s misrep-
resentation and a plaintiff’s injury" that is the touchstone of reliance.

In Basic, therefore, the capacity of the capital markets to
rapidly assimilate public information into stock prices provided the
basis for presuming a causal connection between a material misrepre-
sentation and a plaintiff’s injury, even when that plaintiff may have
been completely unaware of the misrepresentation (let alone "relied"
on it in the traditional sense). By contrast, Plaintiffs’ allegations of
fraud and misrepresentation offer no substitute for actual reliance to
establish such a causal connection. Rather, if an individual plaintiff
in this case were unaware of the alleged misrepresentations, but nev-
ertheless purchased the Plan, we see no basis for presuming that the
misrepresentations nevertheless were a proximate cause of her dam-
ages. Absent this presumption, individualized inquiry will be required
to show that Plaintiffs actually relied on the Agents’ alleged misrepre-
sentations.

                                    2.

   The district court also erred in concluding that individualized
inquiry would not be necessary to establish the second element of
reliance — that Plaintiffs had a right to rely on the Agents’ alleged
misrepresentations or that reliance was justified.

   The court held that "[u]nder South Carolina law, such reliance on
an insurance agent is reasonable," J.A. 1268 (citing Trotter v. State
Farm Mut. Auto Ins. Co., 

, 351 (S.C. Ct. App. 1988),
for the proposition that it is "reasonable for [an] insured to assume the
truthfulness of" an insurance agent’s representations). But in fact,
30              GUNNELLS v. HEALTHPLAN SERVICES, INC.
South Carolina law only holds that "it is reasonable for the insured to
assume the truthfulness of" his agent’s representations under limited
circumstances not present in this case — and, indeed, not present in
Trotter either. See 

(concluding that reli-
ance was not reasonable and distinguishing circumstances from those
of Giles v. Lanford & Gibson, Inc., 

(S.C. Ct. App.
1985)).

   Thus, in Giles, South Carolina’s intermediate appellate court held
that when the unintelligible language of a written agreement would
prevent an insured from discovering the falsity of an insurance
agent’s representation, it is reasonable for the insured to assume the
truthfulness of the insurance agent’s oral misrepresentation of the
written agreement. 

. Because the case at
hand, like Trotter, does not involve allegations of oral misrepresenta-
tions, or a written agreement so unintelligible that Plaintiffs could not
have reasonably discovered any falsity of representation by reading
the policy, the Giles "assumption" has no bearing here. As even the
Giles court noted, it has long been the law in South Carolina that:

     [w]hether or not reliance upon a representation in a particu-
     lar case is justifiable or excusable, what constitutes reason-
     able prudence and diligence with respect to such reliance,
     and what conduct constitutes a reckless or conscious failure
     to exercise such prudence, will depend upon the various cir-
     cumstances involved, such as the form and materialty (sic)
     of the representations, the respective intelligence, experi-
     ence, age, and mental and physical condition of the parties,
     and the relation and respective knowledge and means of
     knowledge of the parties.

(quoting Thomas v. Am. Workmen, 

, 888
(S.C. 1941) (emphasis added)). Thus, the district court misconstrued
South Carolina law when it concluded that Plaintiffs’ right to rely on
the Agents’ misrepresentations could be assumed without inquiring
into the individual circumstances of each plaintiff’s case.

   Therefore, we conclude that the district court abused its discretion
in finding that justifiable reliance could be assumed or presumed in
this case. Absent this assumption or presumption, individual issues
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                  31
clearly predominate. As the Supreme Court recognized in 

, "[r]equiring proof of individualized reliance from each
member of the proposed plaintiff class effectively would have pre-
vented [plaintiffs] from proceeding with a class action, since individ-
ual issues then would have overwhelmed the common ones." Accord

.12

                                   B.

   A second reason why common issues do not predominate in Plain-
tiffs’ claims against the Agents relates to the duty owed to Plaintiffs
by the Agents. Plaintiffs’ contentions of negligent undertaking are
based on the Agents’ alleged failure in their "duty to properly inform
Plaintiffs and other members of the class about the value, quality, and
extent of coverage" provided under the Plan. These contentions also
require individualized inquiry.

   Generally, under South Carolina law, an insurance agent does not
have any duty to advise an insured. See Pitts v. Jackson Nat’l Life
Insur. Co., 

, 510 (S.C. Ct. App. 2002) (citing 

). A duty may be imposed, however, "if the agent,
nevertheless, undertakes to advise the insured." Carolina Prod.
Maint., Inc. v. United States Fid. & Guar. Co., 

, 43
(S.C. 1992) (citing 

). Absent an express
undertaking to assume such a duty (and Plaintiffs do not contend that
the Agents expressly undertook such a duty here), a duty can be
impliedly created. 

determining if an implied duty has been
created, courts must consider several factors, including whether: "(1)
the agent received consideration beyond a mere payment of the pre-
mium, . . . (2) the insured made a clear request for advice, . . . or (3)
there is a course of dealing over an extended period of time which
would put an objectively reasonable insurance agent on notice that his
advice is being sought and relied on." Id. (citing 

). Obviously, consideration of these factors — and thus proof
  12
    Of course, proof of individual reliance need not overwhelm the com-
mon issues in every case. However, considering the centrality of the
Plaintiffs’ allegations involving reliance, and the other issues requiring
individual inquiry, common issues clearly do not predominate in this
case.
32              GUNNELLS v. HEALTHPLAN SERVICES, INC.
of any implied duty — requires an inquiry into the facts of each par-
ticular case.

   When certifying subclasses against the Agents, however, the dis-
trict court ignored these legal principles. On appeal, Plaintiffs cannot.
They must, and do, acknowledge them. They maintain, however, that
the principles articulated in Trotter and Carolina Production apply
only to insurance agents and that the defendant Agents are insurance
brokers, who, Plaintiffs assert, "do have a special duty to their cli-
ents."

   But Plaintiffs’ argument — that the Agents were really "insurance
brokers" — rests on a flawed premise. Under South Carolina law, an
"insurance broker" represents people seeking insurance, while an "in-
surance agent" represents the insurance company. See S.C. Code
§§ 38-1-20, 38-43-10 (2002); see also Allstate Ins. Co. v. Smoak, 

, 753-54 (S.C. 1971) (distinguishing between insurance
brokers and agents on basis of whether employed by person seeking
insurance or by insurance company to solicit and write insurance). In
this case, the district court specifically found that the Agents are
licensed "insurance agents" who marketed and sold the Plan. Plain-
tiffs point to nothing in the record to support their contention that the
Agents were actually "insurance brokers."13 Accordingly, we can
hardly ignore Trotter and Carolina Production and the individualized
proof inquiries those cases require.

                                   C.

   In addition to the individualized determinations inherent in Plain-
tiffs’ claims against the Agents, the affirmative defenses of compara-
tive negligence, assumption of risk, and set-off asserted by the Agents
pose significant obstacles to class certification. Similarly, even if
actual, justifiable reliance could be presumed, the Agents would still
be permitted to introduce evidence to rebut this presumption with
respect to individual plaintiffs. See Banca Cremi v. Alex. Brown &
Sons, Inc., 

, 1036 (4th Cir. 1997) (citing Carras v.
  13
    Plaintiffs’ broker versus agent distinction is surprising given their
heavy reliance on Trotter, a case involving an insurance agent, for the
argument discussed in Part IV.A.2.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                33
Burns, 

, 257 (4th Cir. 1975), for proposition that the
inference of reliance is rebuttable). The district court again ignored
the problems for class certification posed by these defenses.

    On appeal, Plaintiffs maintain that "numerous courts" have rejected
the notion that the assertion of affirmative defenses renders class cer-
tification inappropriate. However, they cite only In re Visa
Check/Mastermoney Antitrust Litig., 

, 138 (2d Cir.
2001), in which the Second Circuit merely stated that the presence of
affirmative defenses does not "automatically" render class certifica-
tion inappropriate. Rather, like other considerations, affirmative
defenses must be factored into the calculus of whether common issues
predominate. 

   Moreover, regardless of other courts’ interpretations of Rule 23, we
have flatly held that "when the defendants’ affirmative defenses . . .
may depend on facts peculiar to each plaintiff’s case, class certifica-
tion is erroneous." 

(citation and internal
quotation marks omitted). Although it is difficult to determine with
any precision, it appears that here the Agents’ affirmative defenses
are not without merit and would require individualized inquiry in at
least some cases. Therefore, they may, indeed, "depend on facts pecu-
liar to each plaintiff’s case." Accordingly, "class certification [wa]s
erroneous." 

                                  D.

   Given the need for individual inquiry into these issues, we must
hold that the class actions certified by the district court against the
Agents do not meet the commonality and predominance requirements
of Rule 23. Thus, the district court abused its discretion in condition-
ally certifying these class actions.

                                  V.

   Finally, we turn to a contention never raised by any of the parties:
our dissenting colleague’s "threshold" argument, post at 58, that even
if common issues do predominate in Plaintiffs’ cause of action, as a
whole, against TPCM — i.e., "that TPCM violated its duties, both
34              GUNNELLS v. HEALTHPLAN SERVICES, INC.
contractual and at law, as third party administrator of the Plan and
such conduct was a cause of the Fidelity Plan’s failure," and that "fail-
ure directly injured Plaintiffs and the absent class members," J.A.
1249 — the district court erred in confining its predominance inquiry
to that cause of action. The dissent steadfastly contends that the dis-
trict court should instead have evaluated whether common issues pre-
dominated in the entire "action taken as a whole." See post at 53
(emphasis in original); see also 

61 ("the case as a whole");
57, 58, 59, 60, 61 ("the action as a whole"); and 62 ("the entire
action"). According to the dissent, a district court must first "deter-
mine that" an entire law suit "as [a] whole" (i.e. not only the cause
of action against TPCM but also the very different causes of action
asserted against the Agents) satisfies the predominance and superior-
ity requirements imposed by 23(b)(3) and only if the entire lawsuit
does satisfy these requirements may a court "manage[ ] through
orders authorized by 23(c)." See 

The dissent’s argument
finds no support in the law — not in Rule 23 itself nor in any case
or treatise. Indeed, in addition to ignoring the plain language of Rule
23, and rendering a subsection of the rule superfluous, the dissent’s
argument is contrary to the Supreme Court’s interpretation of Rule
23, our own precedent and that of every other court (including eight
federal appellate courts), and every scholarly treatise that has
addressed the issue.

   First, the dissent’s rigidly sequential approach ignores the plain
language of the rule. Rule 23(c)(4) provides that:

     [w]hen appropriate (A) an action may be brought or main-
     tained as a class action with respect to particular issues, or
     (B) a class may be divided into subclasses and each subclass
     treated as a class, and the provisions of this rule shall then
     be construed and applied accordingly.

Fed. R. Civ. P. 23(c)(4) (emphasis added). Thus, Rule 23 specifically
dictates that "[w]hen appropriate" a class action may be "maintained"
as to "particular issues" and, after that is done, "the provisions of this
rule," such as the predominance requirement of (b)(3), "shall then . . .
be construed and applied." 

The dissent’s
approach — that a court first determine "that the action as [a] whole
satisfies the predominance and superiority requirements imposed by
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   35
23(b)(3)" and only then employ subsection 23(c)(4), post at 64, sim-
ply ignores Rule 23 (c)(4)’s express command.14 Of course, courts
have no discretion to ignore the plain language of a statute or federal
rule. See, e.g., United States Nat’l Bank of Oregon v. Indep. Ins.
Agents of Am., 

, 454 (1993) (noting that "plain meaning
must be enforced, of course").

   Moreover, the dissent’s approach would render subsection (c)(4) of
Rule 23 superfluous. The dissent would require a court considering
the manageability of a class action — a requirement for predominance
under Rule 23(b)(3)(D) — to pretend that subsection (c)(4) — a pro-
vision specifically included to make a class action more manageable
— does not exist until after the manageability determination is made.
Thus, under the dissent’s reading of Rule 23, a court could only use
subsection (c)(4) to manage cases that the court had already deter-
mined would be manageable without consideration of subsection
(c)(4). This reading leaves subsection (c)(4)(A) without any practical
application, thereby rendering it superfluous. See In re Tetracycline
Cases, 

, 726-27 (W.D. Mo. 1985) ("If the requirement
under Rule 23(c)(4)(A) was that . . . one or more issues ‘predominate’
in the usual Rule 23(b) sense, when compared with all the issues in
the case, there would obviously be no need or place for Rule
23(c)(4)(A)." (emphasis in original)). Such an interpretation of Rule
23 contravenes the well-established prohibition against reading a pro-
vision in a manner that would make any "clause, sentence, or word
. . . superfluous, void, or insignificant," TRW Inc. v. Andrews, 

, 31 (2001) (internal quotation marks and citations omitted).
  14
    The Advisory Committee on Civil Rules has recognized subsection
(c)(4)’s force despite its placement after (b)(3). Thus, in 1995 the Advi-
sory Committee considered amending (c)(4) to read, "an action may be
certified as a class action with respect to particular claims, defenses, or
issues . . ." because "the placement in subdivision (c)(4) of the provision
permitting class actions for particular issues has tended to obscure the
potential benefit of resolving certain claims and defenses on a class basis
while leaving other controversies for resolution in separate actions." See
Edward H. Cooper, Class Actions and the Rulemaking Process: Rule 23:
Challenges to the Rulemaking Process, 71 N.Y.U. L. Rev. 13, 56, 58
(1996) (quoting Proposed Amendments to Rules of Civil Procedure,
Class Actions (Feb. 1995 Draft)).
36              GUNNELLS v. HEALTHPLAN SERVICES, INC.
   In addition to ignoring Rule 23’s plain language and rendering a
part of it superfluous, the dissent’s rigid, sequential reading of the rule
conflicts with controlling precedent. The Supreme Court has refused
to read Rule 23 in the strictly sequential fashion advocated by the dis-
sent. To the contrary, in its most recent discussion of Rule 23(c)(4),
Ortiz v. Fibreboard Corp., 

, 856 (1999), the Court sug-
gested that a district judge could initially act pursuant to subsection
(c)(4) to create subclasses so as to satisfy the previously listed
requirements of subsection (a)(4). Indeed, the Ortiz Court pointed out
that this result "is obvious after Amchem," the case on which the dis-
sent so heavily relies for its theory. See 

"it is obvious
after Amchem that a class divided between holders of present and
future claims . . . requires division into homogeneous subclasses
under Rule 23(c)(4)(B), with separate representation to eliminate con-
flicting interests of counsel." (citing 

(emphasis added))). Although the dissent claims otherwise, if its
interpretation of Rule 23 is followed, Ortiz notwithstanding, a court
could never use subsection (c)(4) to divide plaintiffs into homoge-
neous subclasses "to eliminate conflicting interests of counsel," 

"entire action" "as a whole" would not satisfy the
adequacy of representation requirement of Rule 23(a)(4), and would
therefore never "pass through the gate[ ]" of Rule 23(a). Post at 58;
see also 

(noting that an action must first satisfy "the require-
ments of Rule 23(a)" before a court can employ (c)(4)).15 Despite its
lengthy attempt to analogize Amchem to the case at hand, in an effort
to find some support for its gatekeeper theory, see post at 59-61, the
dissent fails to take into account the Supreme Court’s own character-
ization and application of Amchem in Ortiz.

   Moreover, the dissent acknowledges, but then disregards the import
of the conclusion of the Amchem Court itself that "settlement is rele-
vant to a class certification," and that Rule 23(e) affects a court’s
evaluation of 

(emphasis added). The
Supreme Court explained in Amchem that when dealing with a
settlement-only class pursuant to Rule 23(e), "a district court need not
  15
    Of course, Ortiz involves subclasses under Rule 23(c)(4)(B), while
the case at hand involves certification of particular issues under
(c)(4)(A), but nothing in Rule 23 suggests that (c)(4)(A) does not interact
with the other subsections of Rule 23 in the same fashion as (c)(4)(B).
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                37
inquire whether the case, if tried, would present intractable manage-
ment problems," which would ordinarily be necessary to satisfy Rule
23(b)(3)’s predominance requirement. 

(emphasis added)
(citing Fed. R. Civ. P. 23(b)(3)(D)). The dissent thus ignores the les-
son of Amchem most relevant to the dissent’s "threshold" contention
in the case at hand, i.e., the Supreme Court’s recognition that the sub-
sections of Rule 23 are interactive, and not to be followed, as the dis-
sent suggests, in a strictly sequential fashion. In sum, in its two most
recent class action decisions, Ortiz and Amchem, the Supreme Court
eschewed the dissent’s suggested approach of reading each of Rule
23’s provisions sequentially.

    Additionally, precedent from our own court flatly rejects the dis-
sent’s sequential interpretation of Rule 23. In In re A.H. 

, we counseled that "courts should take full advantage of
the provision in subsection (c)(4) permitting class treatment of sepa-
rate issues in the case." We expressly recognized that "if [an] action
includes multiple claims, one or more of which might qualify as a cer-
tifiable class claim, the court may separate such claims from other
claims in the action and certify them under the provisions of subsec-
tion (c)(4)," provided that "each subclass must independently meet all
the requirements of (a) and at least one of the categories specified in
(b)." 

(emphasis added). Thus, contrary to the dissent’s pro-
tests in this case, we do not espouse a new rule. Rather we follow the
rule articulated in A.H. Robins — that subsection 23(c)(4) should be
used to separate "one or more" claims that are appropriate for class
treatment, provided that within that claim or claims (rather than
within the entire lawsuit as a whole), the predominance and all other
necessary requirements of subsections (a) and (b) of Rule 23 are met.

   Nor are we alone. Not a single court has followed the dissent’s
approach. Several courts and a number of distinguished commentators
have explicitly endorsed a broad issue-specific predominance analysis
of Rule 23. See, e.g., Valentino v. Carter-Wallace, Inc., 

,
1234 (9th Cir. 1996) ("Even if common questions do not predominate
over the individual questions so that class certification of the entire
action is warranted, Rule 23 authorizes the district court in appropri-
ate cases to isolate the common issues under Rule 23(c)(4)(A) and
proceed with class treatment of these particular issues." (citations
omitted)); Simon v. Phillip Morris, Inc., 

, 28-31
38              GUNNELLS v. HEALTHPLAN SERVICES, INC.
(E.D.N.Y. 2001) (Weinstein, J.) ("The framers of Rule 23(c)(4)(A)
considered class actions brought under Rule 23(b)(3) . . . particularly
well suited for certification of fewer than all issues. Their conclusion
follows from the fact that Rule 23(c)(4)(A) assists in satisfying Rule
23(b)(3)’s additional class certification requirements of predominance
and superiority." (citations omitted)); Emig v. Am. Tobacco Co., 

, 395 (D. Kan. 1998) (explaining that Rule 23(c)(4)(A)
permits "adjudication of any issues common to the class even though
the entire litigation may not satisfy the requirements of Rule 23"); 6
Newberg on Class Actions § 18:7 ("Even cases which might not sat-
isfy the predominance test when the case is viewed as a whole may
sometimes be certified as a class limited to selected issues that are
common, under the authority of Rule 23(c)(4)."); 7B Wright & Miller,
Federal Practice and Procedure § 1790 ("Subdivision (c)(4) is partic-
ularly helpful in enabling courts to restructure complex cases to meet
the other requirements for maintaining a class action. . . . The theory
of Rule 23(c)(4)(A) is that the advantages and economies of adjudi-
cating issues that are common to the entire class on a representative
basis should be secured even though other issues in the case may have
to be litigated separately by each class member.").

   All other courts have explicitly or implicitly endorsed an interpre-
tation of (c)(4) that considers whether Rule 23’s predominance
requirement is met by examining each cause of action independently
of one another, not the entire lawsuit, as the dissent would. See, e.g.,
Smilow v. Southwestern Bell Mobile Sys., Inc., 

, 38-43
(1st Cir. 2003) (reversing decertification of two out of three decerti-
fied claims and affirming decertification with respect to third claim);
Piazza v. EBSCO Industries, Inc., 

, 1349, 1351-53
(11th Cir. 2001) (finding "no basis for concluding that the district
court abused its discretion by certifying a class against the EBSCO
Defendants on" claim for breach of fiduciary duty for lost profits but
finding abuse of discretion for certification of stock sale and under-
valuation claims against EBSCO and reversing class certification with
respect to claims against another defendant, Price Waterhouse Coo-
per); Robinson v. Metro-North Commuter R.R. Co., 

, 154
(2d Cir. 2001) (evaluating plaintiffs’ disparate impact claim sepa-
rately from pattern-or-practice disparate treatment claim); Becherer v.
Merrill Lynch, 

, 430 (6th Cir. 1999) (en banc) (Moore,
J., concurring in the judgment) (noting that district court "properly
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   39
certified a class with respect to the contract claims" against one defen-
dant, without ruling on fraud and other claims against multiple defen-
dants); 

(majority opinion) (noting without disapproval that
"the district court reached the class certification question with respect
to the contract issues, but not the remaining claims"); Allison v. Citgo
Petroleum Corp., 

, 421-22 (5th Cir. 1998) (conducting
an independent predominance inquiry for each cause of action); Can-
non v. Cherry Hill Toyota, Inc., 

, 544 (D.N.J. 1999)
("This court . . . has previously rejected the notion that class certifica-
tion under Rule 23 is ‘an all-or-nothing proposition’ requiring class
certification of all causes of action asserted in a single pleading."
(citations omitted)); Stephenson v. Bell Atl. Corp., 

,
289 & n.4, 295 (D.N.J. 1997) (granting class certification for antitrust
claims, denying certification for claims of statutory and common law
fraud, breach of duty of good faith and fair dealing, unjust enrich-
ment, and money had and received, and noting that "[d]ifferential
treatment of claims is permitted by Rule 23(c)(4)" and that no cases
suggest "that class certification of all asserted causes of action is an
all-or-nothing proposition"); Hudson v. Capital Mgmt. Int’l, Inc., 

, 630-31 (N.D. Cal. 1983) (granting class certification for
statutory claims of fraud and negligent misrepresentation and denying
certification for common law claims despite recognizing that
"[i]ndividual questions of law and fact would vastly overshadow any
common questions if the court were to certify classes based on the
common law claims"); Lewis v. Capital Mortgage Invs., 

, 307 (D. Md. 1978) (granting class certification for securities law
claims but excluding common law fraud claim from class action certi-
fication); Hernandez v. Motor Vessel Skyward, 

, 561
(S.D. Fl. 1973) (granting class certification for negligent exposure
cause of action but not breach of contract, breach of implied warranty,
or inadequate medical treatment causes of action); see also Hannah
Stott-Bumsted, Note, Severance Packages: Judicial Use of Federal
Rule of Civil Procedure 23(c)(4)(A), 91 Geo. L.J. 219, 231 (2002)
(observing that "there is little dispute" that certification of a single
claim in a suit where plaintiffs assert multiple claims "is an acceptable
use of the district court’s discretion under Rule 23(c)(4)(A).")

   Indeed, the Third Circuit has twice reversed district courts for tak-
ing the dissent’s approach and failing to consider class certification
of individual claims. See Bogosian v. Gulf Oil Corp., 

,
40              GUNNELLS v. HEALTHPLAN SERVICES, INC.
453 (3d Cir. 1977) ("Even assuming that the court were correct in its
conclusion that the lease claim is not appropriate for class determina-
tion, it nevertheless should have considered certification of the trade-
mark claim under Rule 23(c)(4)(A)."); Geraghty v. United States
Parole Comm’n, 

, 252-53 (3d Cir. 1978) (rejecting dis-
trict court’s conclusion "that a class action is inappropriate" because
"not all of the grounds of action alleged in the complaint are applica-
ble to the [proposed] class" because that conclusion "does not prop-
erly acknowledge the powers and duties of the trial court under
section (c)(4) of Rule 23"), vacated on other grounds, 

,
402-03 (1980).

    Again, no court has adopted the dissent’s interpretation of Rule 23;
no court has required a lawsuit-specific predominance analysis. Thus,
the dissent’s attacks on our approach are as baffling as they are ill-
founded. The dissent’s contention that our approach "bypasse[s]" and
"subvert[s]" Rule 23(b)(3)’s predominance requirement, and "strips
Rule 23(b)(3) of its meaning," post at 56, 59 and 61, is simply false.
We have scrupulously analyzed whether Plaintiffs’ claims against
TPCM and the Agents satisfy the predominance requirement. See ante
at 13-14, 17-20, 27-33. After doing so, we have concluded that Plain-
tiffs’ cause of action against TPCM — "that TPCM violated its duties
. . . as third party administrator of the Plan," which caused "the Plan’s
failure" and so "directly injured Plaintiffs" J.A. 1249 — taken as a
whole, satisfies all of Rule 23’s requirements, including predomi-
nance, ante at 13-14, 17-20.16 That the dissent reaches a different con-
clusion as to whether common issues predominate with respect to
Plaintiffs’ claim against TPCM does not mean that we "strip[ ] Rule
23(b)(3) of its meaning;" it just means that the dissent disagrees with
our assessment. Indeed, the continuing vitality of Rule 23(b)(3)’s pre-
dominance requirement dictated our reversal with respect to Plain-
tiffs’ claims against the Agents — a decision with which the dissent
concurs.
  16
    Thus, contrary to the dissent’s contention, post at 60, there is no rea-
son to "wonder exactly what" the common issues in the certified cause
of action against TPCM "predominate over;" they predominate over the
individual issues involved in that cause of action.
                 GUNNELLS v. HEALTHPLAN SERVICES, INC.                     41
   The dissent offers an equally perplexing response to the wealth of
authority cited above, including cases from the First, Second, Third,
Fourth, Fifth, Sixth, Ninth, and Eleventh Circuits, all of which have
rejected the dissent’s contention that a class can be certified only if
the common issues in "the entire action" — i.e., every cause of action
against every defendant — predominate over individual issues in "the
entire action." The dissent does not discuss, let alone distinguish,
most of the cases we cite; it simply maintains that they do not exist.
See post at 63 (dissent noting it can "identify" no circuit precedent
supporting the majority’s holding).17

  17
     Pointing to the various theories of liability alleged in Plaintiffs’ com-
plaint against twenty-three agents and TPCM, the dissent surprisingly
asserts that each of these theories constitutes a cause of action and that
the certified claim against TPCM is simply a "common question." See
post at 62. Even if this were so, it is unclear how this contention aids the
dissent — for the predominance inquiry would still be conducted within
the confines of these "causes of action" not the lawsuit as whole, as the
dissent would. See, e.g., 

. In fact, the dissent’s
contention is a red herring. A cause of action is "defined by its factual
contours" and is "‘coterminous with the transaction regardless of the
number of substantive theories . . . that may be available to the plain-
tiff.’" CoreStates Bank, N.A. v. Huls America, Inc., 

, 194,
200 (3d Cir. 1999) (quoting Restatement (Second) of Judgments § 24
cmt. a). Thus "claims are part of the same cause of action when they
arise out of the same transaction . . . or the same core of operative facts."
Grausz v. Englander, 

, 473 (4th Cir. 2003) (internal quota-
tion marks omitted). Certainly, in some cases, each theory of liability
arises from a different set of operative facts and so constitutes part of a
different cause of action. But here, although Plaintiffs assert several theo-
ries of liability against TPCM, all of these theories arise from a single
set of "operative facts" and so are part of a single cause of action — that
TPCM mismanaged administration of the Plan resulting in the Plan’s
failure to timely pay out benefits. Moreover, Plaintiffs assert different
causes of action, arising from a very different set of "operative facts,"
against the Agents — that the Agents misrepresented the Plan’s features
leading the Plaintiffs to purchase a woefully deficient Plan. Notably, nei-
ther TPCM nor any other defendant challenges the district court’s clearly
proper treatment of Plaintiffs’ claim against TPCM as a single cause of
action separate and independent from their claims against the agents.
42              GUNNELLS v. HEALTHPLAN SERVICES, INC.
   Just as confounding, is the dissent’s contention that the Fifth Cir-
cuit has adopted the dissent’s approach and that we, in assertedly
refusing to follow the Fifth Circuit, create a circuit conflict. See post
at 53-54, 62-63 (citing Smith v. Texaco, Inc., 

(5th Cir.
2001), Allison v. Citgo Petroleum Corp., 

(5th Cir.
1998), and Castano v. Am. Tobacco Co., 

(5th Cir. 1996)).
In fact, the Fifth Circuit cases do not state, let alone hold, as the dis-
sent would, that "the entire action," or "the action as a whole" — i.e.,
all causes of action against all parties — must satisfy Rule 23 (b)(3)’s
predominance requirement before a court can employ Rule 23(c)(4)
to certify a class. Rather, the Fifth Circuit cases merely require that
"a cause of action, as a whole, must satisfy rule 23(b)(3)’s predomi-
nance requirement" before "rule 23(c)(4) is available." See 

(emphasis added); 

; 

n.21.18 To be sure, as the Second Circuit has noted,
  18
     The dissent fundamentally errs in suggesting that we engage in some
sort of "wordsmithery" in noting this important distinction between the
Fifth Circuit’s view and the dissent’s approach. See post at 63. In fact,
the Fifth Circuit itself has made plain that it means exactly what it has
said — a court is to determine that predominance is satisfied when
assessing each "cause of action, as a whole," not the entire lawsuit, as a
whole, as the dissent posits. For example, in 

,
409, African-American plaintiffs charged employment discrimination
and sought class certification of two causes of action: (1) a claim of sys-
temic disparate treatment by their employer resulting in a pattern or prac-
tice of intentional discrimination and (2) a claim of disparate impact
arising from a facially neutral employment policy. The Fifth Circuit ana-
lyzed each of these claims separately in determining whether (b)(3)’s
predominance requirement had been met. 

In looking at the
first cause of action — the disparate treatment/pattern or practice claim
— the court held that the "first stage of the pattern or practice claim"
could not be divorced from its "second stage" for the predominance anal-
ysis. 

Instead, the court conducted the predominance
inquiry by looking at the pattern or practice claim as a whole, and found
that common issues did not predominate with respect to that cause of
action. 

different parts of the disparate treatment/pattern
or practice cause of action could not be "severed," the court conducted
a completely independent predominance analysis for the second cause of
action — the disparate impact claim — to assess, as it said, "the possibil-
ity of certifying a class action only on the disparate impact claim." 

(emphasis added). This is precisely the approach we have fol-
lowed; it is the majority, not the dissent, that the Fifth Circuit precedent
supports.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                  43
there is a circuit conflict as to whether predominance must be shown
with respect to an entire cause of action, or merely with respect to a
specific issue, in order to invoke (c)(4). See Robinson v. Metro-North
Commuter R.R. Co., 

, 167 n.12 (2d Cir. 2001) (noting
the Fifth Circuit’s "strict application of the (b)(3) predominance
inquiry to the entire pattern or practice claim" before a court can
invoke (c)(4) and further noting that "an alternative understanding of
the interaction of (b)(3) and (c)(4) . . . has been advanced elsewhere"
including by the Ninth Circuit in 

. But we
have no need to enter that fray (and similarly the dissent’s rumina-
tions about "pinhole" issue certification have no relevance here)
because, as we have demonstrated within, in this case Plaintiffs’
cause of action as a whole against TPCM satisfies the predominance
requirements of Rule 23. Thus even if the view adopted by the Fifth
Circuit (that predominance must be established within a given cause
of action to invoke (c)(4)) rather than that of the Ninth Circuit (that
this is not necessarily required) constituted the law of this circuit, our
holding here would be in full accordance with the Fifth Circuit view.

   In fact, we are aware of no decision by any court that has adopted
the dissent’s view that in cases such as the one before us, involving
more than one cause of action, a court must evaluate predominance
in the context of the "action taken as a whole" or the "case as a
whole." Post at 53 and 54. On the contrary, as discussed above, it is
the dissent’s approach that would create a direct conflict with all other
courts, and abrogate longstanding precedent and practice in the area
of class action law. See ante at 35-40.

   Finally, the dissent errs in predicting a "procedural nightmare" on
remand. See post at 56. The class action certified by the district court
against TPCM involves a single, straightforward cause of action con-
sisting of three questions common to all Plaintiffs — whether "TPCM
violated its duties . . . as third party administrator of the Plan,"
whether "such conduct was a cause of the Fidelity Plan’s failure," and
whether "that failure directly injured Plaintiffs and the absent class
members." J.A. 1249. Neither of the first two questions require any
individual inquiry, and the third question may require nothing more
than a ministerial determination that can be made simply and quickly.
We are confident at this stage that proceeding with this cause of
action on a class-wide basis will "‘achieve economies of time, effort,
44               GUNNELLS v. HEALTHPLAN SERVICES, INC.
and expense, and promote . . . uniformity of decision as to persons
similarly situated, without sacrificing procedural fairness or bringing
about other undesirable results.’" See 

(cita-
tion omitted).19 Moreover, it is important to remember that we are
only approving a conditional certification of the case, and that the dis-
trict court has explicitly stated that it would not hesitate to decertify
the class if it became unmanageable. J.A. 1260-61. Thus, in the
unlikely event that the case becomes a "procedural morass," as the
dissent predicts, post at 56, we are confident that the district court will
fulfill its responsibility to decertify the class.

                                    VI.

   In conclusion, we note that circuit precedent provided clear guid-
ance in resolving the issues presented in this case. On one hand,
Plaintiffs’ class action against TPCM bears many similarities to the
class action we approved in Central Wesleyan, and promises many of
the advantages: the plaintiffs advance a single theory of liability,
requiring no individualized inquiry; class certification seems likely to
conserve judicial resources and conserve litigation costs by reducing
the need to re-litigate liability issues in multiple trials; in the absence
of class certification it appears that many of Plaintiffs’ claims against
TPCM would not be brought; and class certification protects the
defendant, TPCM, from inconsistent adjudications and one-way issue
preclusion. Moreover, as in Central Wesleyan, the district court care-
  19
     While the dissent bemoans the need for 1,400 separate individual full
trials on the non-certified claims, its approach would, in fact, add claims
— all of the claims against TPCM — to what it describes as a "minefield
for legal error." Post at 82. Denying class certification against TPCM
would require Plaintiffs to prove the same elements of the certified cause
of action against TPCM in 1,400 individual trials, with the possibility of
contradictory findings and one-way issue preclusion. In contrast, by
holding that the district court did not abuse its discretion in conditionally
certifying the one cause of action against TPCM, we have faithfully
adhered to the text and purpose of Rule 23 and chosen a course that will
"achieve economies of time, effort, and expense, and promote uniformity
of decision as to persons similarly situated." See Fed. R. Civ. P. 23 advi-
sory committee’s note (1966 Amendment, subdivision (b)(3)). In doing
so, we believe we have removed from the dissent’s "minefield" those
mines that could properly be removed.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   45
fully assessed all of Rule 23’s requirements for a class action and
issued detailed factual findings explaining its decision in condition-
ally certifying the class. Indeed, to the extent there are significant dif-
ferences between the class action certified against TPCM and the
class action certification approved in Central Wesleyan — such as the
need in Central Wesleyan to determine comparative fault, and the
application of the laws of different jurisdictions — these differences
weigh even more heavily in favor of class certification here. In sum,
Central Wesleyan virtually compels the conclusion that the district
court did not abuse its discretion in conditionally certifying Plaintiffs’
class against TPCM.

   On the other hand, Plaintiffs’ proposed class actions against the
Agents have many parallels to the class action we rejected in Brous-
sard: the claims against the Agents require individualized proof of
reliance and individualized proof of a specific undertaking to advise.
The Agents’ affirmative defenses would also require further individu-
alized inquiry. Thus, Broussard would seem to require the conclusion
that common issues do not predominate, and that class certification
is not appropriate. Indeed, the district court recognized Broussard’s
force and therefore attempted mightily to distinguish the case. The
court could do so, however, only by misconstruing or ignoring con-
trolling legal principles. When a court misapprehends or fails to apply
the law with respect to underlying issues, it abuses its discretion.
Accordingly, we hold that the district court abused its discretion in
granting conditional class certification against the Agents.

                     AFFIRMED IN PART, REVERSED IN PART,
                AND REMANDED FOR FURTHER PROCEEDINGS
GUNNELLS v. HEALTHPLAN SERVICES, INC.              47
                                        Volume 2 of 2
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   53



NIEMEYER, Circuit Judge, concurring in part and dissenting in part:

   I concur in that part of the judgment holding that the district court
abused its discretion in certifying individual "subclasses" against the
Agent defendants. I dissent from the remainder. I would hold that the
district court abused its discretion in certifying a single class for
claims against TPCM because the representative parties have not sat-
isfied the requirements of Federal Rules of Civil Procedure 23(a)(3),
23(a)(4), and 23(b)(3).

   The majority opinion in this case is able to approve a partial class-
action certification only by failing to account for the full range of the
plaintiffs’ claims and the extent to which most of their claims must
concededly be adjudicated on an individual basis, ignoring the erosive
effect that the individual adjudications of these additional claims has
on the predominance requirement of Rule 23(b)(3) that common
issues predominate over the individualized questions raised in the
action taken as a whole. For example, the majority never addresses
the practical and legal difficulties that will arise from the certification
of a class action against TPCM with respect to only one or two issues
while leaving for individual adjudication in 1,400 separate individual
trials the issues of damages and, in some cases, causation, as well as:

     1) all claims under the South Carolina Unfair Trade Prac-
        tices Act;

     2) all claims alleging civil conspiracy;

     3) all claims involving violations of RICO;

     4) all issues that qualify each plaintiff for a claim under
        the applicable Plan; and
54              GUNNELLS v. HEALTHPLAN SERVICES, INC.
     5) all claims against the agents who sold the Plan.

The need for these individualized separate trials is conceded.

   Despite the overwhelming predominance of these individualized
issues and claims over the common issue that the majority now certi-
fies for class treatment, the majority has adopted an inventive
approach to Rule 23 that allows certification of a class where the pre-
dominance requirement of Rule 23(b)(3) is admittedly unmet in the
context of the case as a whole. According to the majority, to require
the certified issue in this case to predominate over the individualized
issues in the action as a whole ignores Rule 23(c)(4)(A), which it
appears to view as a fourth avenue for class certification, on equal
footing with Rules 23(b)(1), 23(b)(2), and 23(b)(3). In doing so, the
majority glorifies Rule 23(c)(4)(A) — a housekeeping rule that autho-
rizes a court to certify for class treatment "particular issues" in a case
that otherwise satisfies Rule 23(a) and 23(b) — with the effect of
materially rewriting Rule 23 such that Rule 23(b)(3)’s requirements
no longer need be applied to "[a]n action," see Fed. R. Civ. P. 23(b),
but rather to any single issue, no matter how small.

   Not only does the majority’s approach expand Rule 23 beyond its
intended reach, but it also creates a direct conflict with the Fifth Cir-
cuit which has held:

     A district court cannot manufacture predominance through
     the nimble use of subdivision (c)(4). The proper interpreta-
     tion of the interaction between subdivisions (b)(3) and (c)(4)
     is that a cause of action, as a whole, must satisfy the pre-
     dominance requirement of (b)(3) in that (c)(4) is a house-
     keeping rule that allows courts to sever the common issues
     for a class trial.

Castano v. American Tobacco Co., 

, 745 n.21 (5th Cir.
1996); see also Allison v. Citgo Petroleum Corp., 

, 421-
22 (5th Cir. 1998).

  In addition, the majority opinion attempts to shoehorn this case,
even with its limited focus on a single issue, into the parameters of
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                  55
our distinguishable holding in Central Wesleyan College v. W.R.
Grace & Co., 

(4th Cir. 1993), in which we affirmed the
conditional certification of a class of asbestos litigants. In doing this,
the majority fails to consider the broad complexities raised by the
unaddressed issues in this litigation and fails to apply the later and
more specifically applicable controlling ruling of the Supreme Court
in Amchem Products, Inc. v. Windsor, 

(1997).

   Finally, the majority fails meaningfully to address overt conflicts
of interest existing among members of the conditionally certified
class. For example, it does not attempt to explain how the condition-
ally certified class may include, on the one hand, employers seeking
rescission of the insurance contract and the return of premiums paid,
and, on the other hand, their employees seeking enforcement of the
same insurance contract to obtain benefits under it. The majority also
discounts, on scant deposition testimony, the potentiality that
aggrieved employee-class members who were damaged by nonpay-
ment or delayed payment of claims may implead as defendants their
employers, who are also class members and who made the decision
to purchase the woefully inadequate plan. Even aside from the antag-
onistic posture of the employer and employee class members, the
injured employees’ claims themselves will have to be aggregated into
antagonistic subclasses, based on when their claims were mismanaged
and by whom, with entitlement to relief premised on different theo-
ries. By lumping all of these employees into one class and apparently
casting aside as informally "disavowed" a group of employees’ best
theory of recovery, the majority deindividualizes inherently individual
claims and lops off substantial limbs from the various class members’
claims in order to bless the conglomeration of a similar, albeit now
disabled, class of litigants.

   Had these problems been addressed by the majority opinion and the
distinguishable aspects of Central Wesleyan, as well as the controlling
analysis of Amchem, been explored, it would have become apparent
that plaintiffs cannot satisfy, under any applicable precedent, the
requirements of Federal Rule of Civil Procedure 23 (1) that their
claims be typical of the claims of the class so that they can act as rep-
resentative parties, able fairly and adequately to protect the interests
of the class, as required by Rule 23(a)(3) and (4); (2) that common
questions of law and fact predominate over questions affecting only
56              GUNNELLS v. HEALTHPLAN SERVICES, INC.
individual class members, as required by Rule 23(b)(3); and (3) that
the class action be superior to other available methods for the fair and
efficient adjudication of the numerous claims and controversies, as
required by 23(b)(3).

   As a result of the majority’s limited focus on the facts related to
a single issue in this case, it has left a difficult and complex proce-
dural structure created by the need to try numerous individual claims
for each class member that will result in an unnecessary, and ulti-
mately unhelpful, procedural nightmare. And on a broad judgment
level, one has to question the utility of enduring this procedural
morass for the purpose of answering the single question certified for
class treatment: whether TPCM’s mismanagement was a cause of the
Plan’s failure. Answering this question resolves no class member’s
claim and only invites the difficult questions of how to proceed once
the question is answered. Little, if any, time or effort can be saved by
answering this question in the abstract because full individual trials
on liability will still have to be conducted for each individual class
member.

  I would reverse the certification of the class against TPCM in this
case for the reasons that follow.

                                    I

    Because the majority holds that the district court’s conditional cer-
tification of a single issue against TPCM constituted a proper applica-
tion of Rule 23(c)(4), ante at 14-15, it is necessary to address at the
outset the fundamental issue of why the predominance requirement of
Rule 23(b)(3) cannot be bypassed by reliance on Rule 23(c)(4). In
short, I believe that the majority, in adopting an expansive interpreta-
tion of Rule 23(c)(4), has enlarged the reach of Rule 23 in a manner
not contemplated by the Rule’s drafters and not consistent with the
Supreme Court’s approach calling for an "undiluted" application of
Rule 23(b)(3)’s requirements in every case. 

;
see 

the Rule, as written, "sets the requirements
[courts] are bound to enforce" and counseling against "judicial inven-
tiveness" in application of the Rule). Thus, I would conclude that the
history of the Rule and its text, read in light of the Supreme Court’s
holding in Amchem, require that even in cases involving the certifica-
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                 57
tion of issue-only classes, the common issues must predominate over
questions affecting only individual members of the class in the con-
text of the action as a whole. The majority’s conclusion to the con-
trary seems to invite an unhesitating approach to granting class
certification where other procedural devices would suffice, and in
doing so, it opens a conflict with the Fifth Circuit. Thus, only after
I elaborate the basis of my conclusion that a predominance analysis
is mandated in every Rule 23(b)(3) class action do I then proceed with
the required analysis by looking at the various issues and claims pre-
sented in this case.

   Of the Rule 23 amendments adopted in 1966, the addition of Rule
23(b)(3) was "the most adventuresome innovation." 

(internal quotation marks and citation omitted). Because
Rule 23(b)(3) was an innovative device and indeed involved some
speculation about its impact, the Rule was careful to impose two spe-
cific and important requirements that must be found as conditions to
certification of any class under the Rule. The "predominance" require-
ment requires the district court to evaluate the "action" to determine
whether "questions of law or fact common to the members of the
class predominate over any questions affecting only individual mem-
bers," and the "superiority" requirement requires the court to analyze
whether the proposed class action would be "superior to other avail-
able methods for the fair and efficient adjudication of the contro-
versy" raised by the "action." Fed. R. Civ. P. 23(b)(3). If an action
does not satisfy those two "class-qualifying criteria" in addition to the
requirements of Rule 23(a), 

, there is no
basis for the court to certify a proposed class, even on a conditional
basis. The predominance and superiority requirements are thus
organic to every 23(b)(3) action. In Amchem, where the plaintiffs
sought to bypass the predominance requirement in certifying a class
for settlement, the Supreme Court could not have made that proposi-
tion any clearer:

    In addition to satisfying Rule 23(a)’s prerequisites, parties
    seeking class certification must show that the action is main-
    tainable under Rule 23(b)(1), (2), or (3).

                             *     *      *
58               GUNNELLS v. HEALTHPLAN SERVICES, INC.
      To qualify for certification under Rule 23(b)(3), a class must
      meet two requirements beyond the Rule 23(a) prerequisites:
      Common questions must "predominate over any questions
      affecting only individual members"; and class resolution
      must be "superior to other available methods for the fair and
      efficient adjudication of the controversy."

Indeed, the Court emphasized that Rule 23(b)(3)’s
requirements are "class-qualifying criteria" and that "[f]ederal courts
. . . lack authority to substitute for Rule 23’s certification criteria" any
other standard than the one adopted in the rule itself. 

   Even though every class action must pass through the gates of Rule
23(b)(3), a court is, of course, entitled to take into account the man-
ageability tools supplied by Rule 23(c)(4) in determining whether the
predominance and superiority requirements of Rule 23(b)(3) are satis-
fied with regard to the action as a whole. Thus, just as a court may
consider the availability of subclasses in determining whether an
action satisfies Rules 23(a) and 23(b)(3), see Ortiz v. Fireboard
Corp., 

, 856 (1999)(noting that, at the class certification
stage, the availability of subclasses to eliminate conflicting interests
could factor into whether the requirement of Rule 23(a)(4) is satis-
fied), so may it consider the availability of issue-only classes in mak-
ing the same determinations. In either case, for such subclasses or
issue-only classes to be useful management tools in a particular case,
they, too, must individually satisfy the requirements of Rule 23(a) and
23(b)(3). See Fed. R. Civ. P. 23(c)(4) (providing that, if a court uti-
lizes issue-only classes or subclasses, "the provisions of this rule shall
then be construed and applied accordingly"). In the case before us,
however, the threshold dispute I have with the majority is not whether
the single-issue class that has been certified here satisfies Rule 23(b)(3),1
but rather whether an issue-only class action may be maintained under
Rule 23(c)(4) when the common issue does not predominate over the
other issues in the litigation that affect only individual members.

   I conclude that when the text of Rule 23(b)(3) provides that "ques-
tions of law or fact common to the members of the class [must] pre-
  1
   For reasons that I explain infra, I would hold that the class certified
does not even satisfy the requirements of Rule 23(a) and 23(b)(3).
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                 59
dominate over any questions affecting only individual members" in
"[a]n action," "action" means the action as a whole. And I do not
believe that the possibility of bringing or maintaining an action "with
respect to particular issues" was ever meant to expand Rule 23(b)(3)
to allow class certification when the individual issues in the class
members’ case against a defendant tower over the comparably insig-
nificant common issue or issues. The majority disagrees with my
reading, which it disparages as "rigidly sequential," ante at 34, instead
adhering to a reading of Rule 23(c)(4) that allows a district court to
certify any common issue in the litigation, no matter how small. In
my view, the majority’s interpretation of Rule 23(c)(4) invites a
diluted application of Rule 23(b)(3) in the context of issue-only
classes, which I respectfully submit departs from the principle elabo-
rated in Amchem.

   The core question addressed by the Supreme Court in Amchem was
the interplay between Rule 23(b)(3) and 23(e), the provisions upon
which the parties sought certification of a settlement-only class and,
ultimately, approval of the negotiated settlement. The settling parties
viewed the predominance requirement as a requirement applicable
only to an action that was destined for trial, and therefore they argued
that cohesion of the action and fairness for settlement was better
determined by a hearing under Rule 23(e) when approving settlement.
Because the Supreme Court in Amchem analyzed the relationship
between the predominance requirement of Rule 23(b)(3) and the set-
tlement authority of Rule 23(e), its analysis is apposite to the analysis
of whether the predominance requirement of Rule 23(b)(3) can be
subverted by Rule 23(c)(4)(A).

   The district court in Amchem certified the settlement-only class
over objectors’ strenuous objections that the Rule 23(b)(3) class did
not meet the requirements of Rule 23(a) and (b)(3). The Third Circuit
vacated the class certification, holding that the settlement-only class
satisfied neither Rule 23(a) nor 23(b)(3) and that the cases should pro-
ceed either through consolidation under Rule 42(a) or as smaller class
actions under Rule 

. Affirming the Third Circuit,
the Supreme Court "emphasize[d]" that the requirements of Rule
23(a) and (b)(3) are "safeguards," not "impractical impediments" that
can be disregarded by a court that perceives a measure of fairness in
overlooking these standards in a particular case. 

While
60              GUNNELLS v. HEALTHPLAN SERVICES, INC.
holding that settlement was "relevant" to class certification and
acknowledging that the manageability of a trial was not relevant to a
class action where no trial was contemplated, the court noted that the
other specifications of the Rule, including the predominance require-
ment, "demand[ed] undiluted, even heightened attention." 

.
Requiring a settlement-only class to satisfy Rule 23(b)(3)’s require-
ments, the Court held, "serve[s] to inhibit appraisals of the chancel-
lor’s foot kind — class certifications dependent upon the court’s
gestalt judgment or overarching impression of the settlement’s fair-
ness." 

Court concluded that "it is not the mission
of Rule 23(e) to assure the class cohesion that legitimizes representa-
tion action in the first place." 

"If a common interest in a
fair compromise could satisfy the predominance requirement of Rule
23(b)(3)," the Court continued, "that vital prescription would be
stripped of any meaning in the settlement context." 

it is not the mission of Rule 23(e) to supply the cohesion
that legitimizes a settlement-only class action, neither is it the mission
of Rule 23(c)(4)(A) to supply the cohesion to legitimize an issue-only
class action. In both situations, the cohesion essential to legitimize a
23(b)(3) class action can be shown only when the action as a whole
satisfies the predominance requirement of Rule 23(b)(3). The princi-
ple of Amchem is that every Rule 23(b)(3) class action must satisfy
all of the provisions of Rule 23(a) and (b)(3), and the other provisions
of the Rule, including Rule 23(e), cannot be used to dilute the require-
ment that each proposed class must satisfy the predominance require-
ment to merit certification. In my view, the majority’s reading of Rule
23(c)(4) allows for a diluted application of Rule 23(b)(3) by removing
from the predominance calculus most of the individualized issues in
the case.

   Under the majority’s analysis, which looks at the common issue in
the case through a pinhole and ignores all other issues, Rule
23(b)(3)’s vital prescription is stripped of any meaning in an even
more dramatic way than it was in Amchem. The majority in this case
boldly holds that individualized claims and issues in this action are
not a factor in analyzing whether common issues predominate in the
certified claim against TPCM, leaving one to wonder exactly what the
isolated common issue must predominate over. For if Rule
23(c)(4)(A) allows a court to omit from its predominance analysis any
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   61
claims or issues affecting only individual members, it would seem
that the predominance of the selected issue is a foregone conclusion
since the common question of law or fact would always predominate
over the individual issues that are not a factor. And in the rare
instance where the majority might find a common issue not to pre-
dominate in a given case, it could simply narrow the pinhole until, in
its view, the selected issue predominates over the other issues it
chooses to see. Indeed, nothing in the majority’s opinion supplies any
lower limit on just how narrow a common issue may be before its
analysis must yield to common sense. By adopting a view of Rule
23(c)(4)(A) that strips Rule 23(b)(3) of its meaning, necessitating the
aid of additional principles found nowhere in the Rule itself, the
majority has "substitute[d] for Rule 23’s certification criteria a stan-
dard never adopted," a substitution that "[f]ederal courts . . . lack
authority" to make. 

.

   In an effort to enlist legal authority to interpret Rule 23 to authorize
a class action on a single issue, the majority, in Part V, misconstrues
how I have interpreted Rule 23 and fails to appreciate the conse-
quences of reading the Rule in a manner that lets Rule 23(c)(4) serve
as a gate house for determining whether a class action should be certi-
fied. Even as an action may include different "causes of action" or dif-
ferent theories of recovery, all causes of action or claims must be
related to form the action. See Fed. R. Civ. P. 18-21, 42. And Rule
23 provides that, to qualify "an action" as a class action, the prerequi-
sites of both subdivision (a) and subdivision (b) be satisfied. See Fed.
R. Civ. P. 23(b). Subdivision (c) then addresses the management of
the class action authorizing the certification of either issues or sub-
classes. But a class certification with respect to an issue cannot, with-
out bypassing Rule 23(b), be undertaken unless the issue is so
important to disposition of "the action" that it "predominates over any
questions affecting only individual members." Rule 23(b)(3). None of
the cases cited by the majority advocate such a bypassing that the
majority now advances as an entirely novel proposition.

   The majority criticizes my demand that a full predominance analy-
sis be conducted, as required by Rule 23 and Amchem, both insisting
that it can rely on a single issue to certify the class — notwithstanding
a conclusion that the predominance test would not be satisfied in the
context of the action as a whole — and stating that I have "ignore[d]
62              GUNNELLS v. HEALTHPLAN SERVICES, INC.
the plain language of [Rule 23(c)(4)]," ante at 34, and that my analy-
sis "render[s] a subsection of the rule superfluous," ante at 35.

   I disagree. This very case illustrates the importance of my concep-
tual disagreement with the majority and the inviability of the majori-
ty’s position. In this case, the class members have asserted seven
causes of action against TPCM: an unfair trade practices claim, a neg-
ligent undertaking claim, a fraud claim, a negligent misrepresentation
claim, a breach of contract claim, a civil conspiracy claim, and a
RICO claim. The majority, without identifying to which of these seven
causes of action the certified issue belongs, has certified for class
treatment the vague question whether TPCM was "a cause" of the
failure of the Plan. Naming this certified question a "cause of action,"
but not tying it to any specific claim in the complaint, the majority
makes the analysis of what this common question must predominate
over particularly abstruse. The majority’s analysis seemingly identi-
fies a common causation issue and then jumps immediately to Rule
23(c)(4)(A) to insulate the newly certified issue-only class from the
scrutiny of the predominance requirement in the context of the case
as a whole, even as the plaintiffs urged consideration of the entire
action and certification of the action as a whole. See Pls.’ Mot. for
Class Certification at 4 (seeking certification of a single class for all
seven causes of action against TPCM and certain other defendants).

   In failing to examine the various other issues presented in the entire
action to determine whether the required "cohesiveness" is present to
"legitimize[ ] representative action in the first place" — as mandated
by 23(b)(3), see 

— and by jumping to
23(c)(4), the majority not only bypasses one of the most essential and
most important checks imposed by Rule 23 on certifying Rule
23(b)(3) classes, it also opens a conflict among the circuits on this
issue. The Fifth Circuit has applied the Rule with the dogged require-
ment of satisfying predominance before considering other aspects of
the Rule in a 23(b)(3) class. See Smith v. Texaco, Inc., 

,
409 (5th Cir. 2001) (later withdrawn pursuant to settlement by the
parties, see 

(5th Cir. 2002)); 

-
22; Castano v. American Tobacco Co., 

, 745 n.21 (5th
Cir. 1996). In Smith, the court explained its position on the proper
relationship between the predominance requirement and Rule
23(c)(4):
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                 63
    The predominance inquiry involves a comparison of the
    issues common among the class members and the issues
    individual to them. This analysis remains unchanged
    whether a class is certified under one or more sections of
    rule 23(b). The inquiry’s constancy serves as an important
    limitation on the use of bifurcation by preventing a district
    court from manufacturing predominance through the "nim-
    ble use" of Rule 23(c)(4). Castano v. Am. Tobacco Co., 

, 745 n.21 (5th Cir. 1996).

    Therefore, the cause of action, as a whole, must satisfy rule
    23(b)(3)’s predominance requirement. 

require-
    ment is met, rule 23(c)(4) is available to sever the common
    issues for a class trial. To read the rule not as a housekeep-
    ing rule, but instead as allowing a court to pare issues
    repeatedly until predomination is achieved, would obliterate
    rule 23(b)(3)’s predominance requirement, resulting in auto-
    matic certification in every case in which any common issue
    exists, a result drafters of the rule could not have 

; see also 

. Cf. Valentino
v. Carter-Wallace, Inc., 

, 1234 (9th Cir. 1996) (reversing
a district court’s effort to rely on 23(c)(4) and bypass 23(b)(3) but
acknowledging that in "appropriate cases" the predominance require-
ment could be overlooked). We can identify no precedent from the
Supreme Court or from our fellow Circuits to support the majority’s
holding.

   The majority’s breathtaking assertion that its reading of Rule
23(c)(4) does not in any way contravene the Fifth Circuit’s cases is
simply not true. The Fifth Circuit, by its very language, only consults
the "housekeeping" provisions of Rule 23(c)(4) "once [the predomi-
nance] requirement is met." 

. This is exactly
the "rigidly sequential approach" that the majority criticizes in its
opinion, ante at 34, making its claim to consistency with the Fifth Cir-
cuit facetious, at best. The majority’s wordsmithery with respect to
the term "cause of action" does not disguise a plain conflict between
the Fifth Circuit’s holding and its holding. To be sure, the majority
identifies some district court opinions, commentaries, and a student
note that, citing one another, argue in favor of the majority’s interpre-
64              GUNNELLS v. HEALTHPLAN SERVICES, INC.
tation. But I respectfully disagree with the cited authorities, and in
doing so, I adhere to the view of the only other circuit court to address
the issue.

   I would hold therefore that under the proper analysis, the district
court must determine that the action as whole satisfies the predomi-
nance and superiority requirements imposed by 23(b)(3), and actions
satisfying these requirements may be managed through orders autho-
rized by 23(c). Thus, under Rule 23(c)(4), a court could create sub-
classes or even bifurcate claims by issues, grouping them for class
action treatment. But in this complicated case, the predominance and
superiority requirements cannot be satisfied even with the aid of the
manageability tools supplied by Rule 23(c), as I now proceed to dem-
onstrate in some detail.

                                   II

   The plaintiffs in this case are purchasers and beneficiaries of a
multi-employer healthcare plan (the "Plan") sponsored and marketed
in South Carolina by the Fidelity Group, Inc. On the Plan’s demise,
they commenced this class action under Federal Rule of Civil Proce-
dure 23(b)(3) against the Fidelity Group and related companies and
individuals, against 43 agents who marketed and sold the Plan, and
against the Plan’s first claims administrator, alleging breach of con-
tract, fraud, mismanagement, and related claims.

   The district court, by order dated September 28, 2001, certified 24
parallel classes — one against each of 23 agents and one against the
claims administrator — but left for individual trials numerous claims,
numerous issues within class-action claims, and all damage issues,
concluding that approximately 1,400 mini-trials will be required.

   We granted the defendants’ motion to review this order under Fed-
eral Rule of Civil Procedure 23(f).

   Although discovery has not been completed and the facts have not
yet been fully developed, it appears that the Fidelity Group intended
to create and market a multi-employer healthcare plan that would be
available to small businesses and individuals and that would be sub-
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                65
ject to regulation by the provisions of ERISA. To do this, it created
both the National Association of Business Owners and Professionals
("NABOP") to act on behalf of potential business customers and the
International Workers Guild, Inc., a union to act on behalf of the busi-
ness customers’ employees and other individuals who might elect
coverage under the Plan. NABOP and the International Workers
Guild thereafter "negotiated" the Plan to provide healthcare and dental
benefits to Plan participants and beneficiaries. The Fidelity Group,
through NABOP, then entered into a contract with Third Party Claims
Management, Inc. ("TPCM")2 to administer the Plan and process
claims. Under this contract, effective April 1, 1995, TPCM became
responsible for processing incoming claims, determining each claim-
ant’s eligibility for coverage, computing benefits payable, and paying
claims out of a bank account maintained by the Fidelity Group and
NABOP.

   Plaintiffs allege that over the next three years, the Plan’s adminis-
tration became deficient, causing a backlog in the processing and pay-
ing of claims, ultimately contributing to the Plan’s becoming
underfunded and to its demise. There is evidence in the record to sup-
port allegations that TPCM did a poor job of paying claims, some-
times losing claims and other times adjusting or "adjudicating" claims
that should not have been adjudicated. TPCM admits that it did not
have a sufficient number of trained claims examiners to stay current
with claims being filed, and, as a result, a backlog in claims process-
ing developed that grew to six to eight months. Plan members began
to complain widely of the delays in the processing of their claims.

   During the same period that TPCM was administering the Plan in
this case, it was also experiencing a claims backlog for at least ten
other clients. There is evidence that TPCM’s problems were caused
in part by the fact that, during the approximately 20 months it was
administering claims under the Plan, TPCM moved its claims admin-
istration operations twice, each time requiring new notifications to
healthcare providers of the need to forward claims to a new address.
  2
   At the time of contract, TPCM’s name was Millennium HealthCare,
Inc., and later TPCM was succeeded by Healthplan Services, Inc. For
convenience, I refer to any or all of these entities collectively as
"TPCM."
66              GUNNELLS v. HEALTHPLAN SERVICES, INC.
After TPCM failed to reduce the backlog and fulfill its continuing
assurances to the Fidelity Group and NABOP that it would correct the
deficiencies, NABOP terminated the contract with TPCM for cause,
effective May 1, 1997, and the Fidelity Group moved claims adminis-
tration "in-house," to be handled by Fidelity Claims Management,
Inc.

   Because there was a serious backlog and Fidelity Claims Manage-
ment, lacking both experience and expertise, apparently was not pre-
pared to step in as administrator, Plan members continued to suffer
delayed payment or nonpayment of their claims. After more than a
year of additional frustration by Plan members, the Secretary of Labor
commenced an action in 1998 against the Fidelity Group and related
companies and individuals in the Eastern District of New York to take
control of the Plan’s assets and to shut down the Plan’s operations.
The district court in New York issued a stay, enjoining the prosecu-
tion of suits against the Fidelity Group, the Plan, and related corpora-
tions and individuals, and it appointed an independent fiduciary to
manage the assets. On the fiduciary’s request that the Plan be termi-
nated, the New York district court ordered the Plan terminated, effec-
tive January 31, 1999.

   The plaintiffs commenced this action on September 11, 1998, and
the district court followed the New York court and likewise stayed
claims against the Fidelity defendants.3 But, by an order dated
December 27, 1999, the district court lifted the stay as to "all non-
Fidelity Defendants," allowing prosecution of the claims against
TPCM for claims mismanagement and against the selling agents for
misrepresenting the Plan. The selling agents are 43 individuals and
companies that sold the Plan on behalf of the Fidelity Group to busi-
nesses and individuals during the period from August 1995, when the
Plan commenced, to July 1997, when the agents were ordered by the
South Carolina Department of Insurance to cease marketing the Plan.

   In their complaint against TPCM and the 43 selling agents, the
plaintiffs alleged (1) violations of the South Carolina Unfair Trade
  3
   The Fidelity defendants include the Fidelity Group, NABOP, the
union that it formed, the Plan, Fidelity Claims Management, and
involved executives and employees.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                 67
Practices Act, S.C. Code Ann. § 39-5-10 et seq.; (2) negligence in
undertaking to administer, market, and represent the Plan; (3) fraud
in connection with representations made about the formation of the
Plan, its benefits, its administration, and its value; (4) negligent mis-
representation, which plaintiffs alleged induced them to purchase and
continue to pay for the Plan; (5) breach of contract; (6) civil conspir-
acy to form, market, and sell a "substantially worthless health care
and dental" plan; and (7) violations of RICO, 18 U.S.C. § 1961 et seq.
The gravamen of plaintiffs’ claims against TPCM is that TPCM,
under circumstances that vary from plaintiff to plaintiff, breached
duties to pay plaintiffs’ claims timely and to provide accurate infor-
mation concerning the status of pending claims. The plaintiffs allege
that these breaches not only caused them direct injury but also con-
tributed to the ultimate failure of the Plan, causing them injury on this
indirect basis. The gravamen of plaintiffs’ claims against the agents
is (1) that the agents sold the plaintiffs worthless insurance that did
not comply with South Carolina statutory requirements, (2) that the
agents were not licensed to sell the Plan, and (3) that they fraudu-
lently or negligently misrepresented the Plan and provided negligent
advice regarding the Plan.

   For relief, the employer-plaintiffs that provided the Plan benefits to
their employees seek return of the premiums paid for "worthless
insurance" as well as consequential damages. The employee-plaintiffs
enrolled in the Plan seek payment of the benefits to which they were
entitled and consequential damages, including damages for mental
anguish, loss of enjoyment of life, injury to their reputations and
credit ratings, and costs of rectifying delayed payments. All plaintiffs
seek punitive damages.

   The plaintiffs sought to represent a class of 1,400 persons, consist-
ing of all persons who purchased coverage under the Plan on or after
August 15, 1995, who have claims against TPCM and against the
agents as a group, and all employees and beneficiaries who had
claims during the three-year period it was operating. They also sought
to represent agent-specific subclasses with respect to particular negli-
gence and fraud claims against 23 particular agents. In explanation,
they alleged:

    Certain claims against the Defendant Agents, including civil
    conspiracy and RICO claims, can be treated on a class-wide
68              GUNNELLS v. HEALTHPLAN SERVICES, INC.
     basis. Other claims against the defendant agents may be
     more agent-specific and may exist only between those class
     members and the particular Defendant Agent who marketed,
     sold, or otherwise provided the coverage under the fund to
     plaintiffs.

As to the other 20 agents — for whom there were no corresponding
named plaintiffs — no subclasses were requested.

   Troubled somewhat by the problems of granting class treatment as
to all asserted claims, the district court granted in part and denied in
part plaintiffs’ motion for class certification. In doing so, the court left
a complex division between individualized claims and claims subject
to class-action treatment. Assessing the motion for class certification
against the requirements of Rule 23, the court first concluded that the
number of class members is "approximately 1,400," which it con-
cluded satisfied the numerosity requirement of Rule 23(a)(1). But on
the 23 proposed agent-specific "subclasses," the court acknowledged
that the numerosity issue was a close question. As to one agent, the
court noticed that only 8 employees were involved, and 2 of them
already had filed individual actions against the agent. Another agent
had enrolled only 13 to 14 employees. The court acknowledged that
"[w]hile these numbers do not definitely constitute a finding of
numerosity, Plaintiffs note that the agent Defendants failed to take
into account the significant number of dependents that were also cov-
ered by the Plan." It found that including dependents, there were 11
members of the class for the one agent and 21 for the other. The court
did not examine numerosity with respect to the other 21 agent classes
but it observed that they contained more members. It concluded,
"[w]hile these numbers may fall on the borderline of numerosity, the
court finds that numerosity is met as to these agents."

  The court next concluded that the proposed classes satisfied the
commonality and typicality requirements of Rule 23(a)(2) and (a)(3),
noting that the common issues included (1) whether TPCM misman-
aged plaintiffs’ claims, (2) whether TPCM’s alleged negligence prox-
imately caused the failure of the Plan, and (3) whether the agents
misrepresented the Plan to the plaintiffs. The court acknowledged,
however, that all claims for "mental anguish or emotional distress"
and damage to credit ratings and reputation would have to be decided
               GUNNELLS v. HEALTHPLAN SERVICES, INC.                69
individually. The court also acknowledged the potential need for indi-
vidualized mini-trials on proximate causation specifically relating to
the reason for nonpayment of a given claim. This was especially true,
the court noted, as to "non-adjudicated unpaid claims for medical
bills," i.e., those which had never been analyzed by a claims adjuster.

   In evaluating the adequacy-of-representation requirement set forth
in Rule 23(a)(4), the court rejected TPCM’s contention that plaintiffs
could not represent the class adequately because of conflicts of inter-
est among class members. TPCM had argued that there was a conflict
between the employer-plaintiffs who purchased insurance coverage
for their employees and the employee-plaintiffs who might have a
claim against their employers and who would in any event be seeking
different relief. The court ruled that the employee-plaintiffs’ claims
against their employers were not "viable," and, in any event, the
employee or employer class members "could opt out of the class."

   In considering the requirements of Rule 23(b)(3), the court first
concluded that common issues predominated over issues affecting
only some of the class members. The court reasoned that, because
most of the plaintiffs’ unpaid claims had already been "adjudicated,"
individualized inquiries relating to the proximate cause of plaintiffs’
unpaid claims would not be necessary. The court reiterated that causa-
tion issues relating to unadjudicated claims could be resolved in the
individualized mini-trials necessary also to resolve damages.

   As to the 23 agent-specific "subclasses," the court also acknowl-
edged that oral representations and individual reliance tended to sug-
gest individualized treatment of the claims against the agents.
Nonetheless, the court concluded that those defenses could be gener-
alized by presumptions and therefore treated on a class-wide basis:

    [A] common sense approach presuming reliance exists may
    be acceptable in this circumstance. Therefore, in certain
    cases, "reliance may be presumed for fraud-based common
    law claims when the alleged omissions and misrepresenta-
    tions are uniform and material and the class members acted
    in a manner consistent with reliance."

                            *     *      *
70              GUNNELLS v. HEALTHPLAN SERVICES, INC.
       Even if the court has erred in its analysis and reliance
     does threaten the predominance of common issues in this
     case, that threat can be eliminated by bifurcating the issues
     of reliance where individual inquiries would be proper.

   On the superiority requirement of Rule 23(b)(3), the court
acknowledged that "[m]anageability of this class action is of particu-
lar concern to the court," pointing particularly to the necessity of indi-
vidual "mini-trials." However, the court then concluded that

     The record presently before the court suggests that bifurca-
     tion of individualized damages issues and possibly proxi-
     mate cause issues regarding non-adjudicated claims would
     not present an unmanageable scenario for the court.

   Addressing separately the civil conspiracy and RICO counts, the
court concluded that individualized defenses predominated, and there-
fore it excluded those claims from class-action treatment. It said:

     [T]he court notes a significant problem regarding individu-
     alized issues inherent where numerous defendants (in fact a
     class of defendants) are accused of being part of a vertical
     conspiracy such as the one alleged in this case. These issues
     arise primarily due to individualized defenses which may be
     available to the Agent Defendants.

                             *      *     *

     Therefore, the court concludes that individual issues regard-
     ing Plaintiffs’ conspiracy claims predominate rendering
     class treatment of all Plaintiffs’ conspiracy claims against all
     Agent Defendants unadvisable.

  On the South Carolina Unfair Trade Practices Act count, the court
again required individual trials because the statute itself mandates
such treatment. See S.C. Code Ann. § 39-5-140(a).

   After making this overall division between claims and issues for
class-action treatment on the one hand and claims and issues for indi-
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                   71
vidual trials on the other, the court entered the following order certify-
ing this case as a class action:

     It is, therefore, ordered, for the foregoing reasons that Plain-
     tiffs’ Motion for Class Certification is conditionally granted
     as modified by this Order. Specifically, the proposed class
     of all Plaintiffs for claims against TPCM is conditionally
     certified (excluding claims under the South Carolina Unfair
     Trade Practices Act). However, class-wide treatment of
     Plaintiffs’ conspiracy claims against all Agent Defendants is
     denied. Instead, the court conditionally certifies Plaintiffs’
     proposed subclasses currently with representatives based on
     the Agent Defendants as individual classes for all of the
     Plaintiffs’ claims (excluding claims under the South Caro-
     lina Unfair Trade Practices Act).

   In sum, the court certified 24 distinct and parallel class actions,
having rejected any overarching class action against the agents, to try
a limited range of claims and issues. The court ruled that individual
trials would have to be conducted on:

     1) all claims under the South Carolina Unfair Trade Prac-
        tices Act;

     2) all claims alleging civil conspiracy;

     3) all claims involving violations of RICO;

     4) any reliance issues that could not, as a matter of law, be
        presumed;

     5) all damage issues;

     6) claims against the 20 agents with whom no named
        plaintiffs had dealings;

     7) all issues regarding the fact and nature of injury specifi-
        cally as to plaintiffs’ claims of mental anguish and
        emotional distress, injury to credit rating, and injury to
        reputation; and
72              GUNNELLS v. HEALTHPLAN SERVICES, INC.
     8) all questions relating to whether plaintiffs’ claims were
        "adjudicated" and to the cause of delayed payment or
        nonpayment for those not adjudicated.

  TPCM and the 23 agent-defendants against whom class actions
were certified filed a motion under Federal Rule of Civil Procedure
23(f) for permission to appeal the district court’s interlocutory order,
and by order dated November 30, 2001, we granted that motion.

                                   III

   It is appropriate to set forth the established principles applicable to
consideration of the issues. We have previously held that "[d]istrict
courts have ‘wide discretion in deciding whether or not to certify a
proposed class,’" Central Wesleyan Coll. v. W.R. Grace & Co., 

, 185 (4th Cir. 1993) (quoting In re A.H. Robins Co., 

, 728-29 (4th Cir. 1989)), but we have noted that this discretion
must be exercised "‘within the framework of Rule 23,’" Lienhart v.
Dryvit Sys., Inc., 

, 146 (4th Cir. 2001) (quoting In re Am.
Med. Sys., Inc., 

, 1079 (6th Cir. 1996)). Where, as here,
the plaintiffs have sought class certification pursuant to Federal Rule
of Civil Procedure 23(b)(3), the plaintiffs bear the burden of proving
that the "numerosity, commonality, typicality, representativeness, pre-
dominance, and superiority requirements of both Rule 23(a) and
(b)(3) are met." 

of this burden upon the parties
seeking class certification reflects the principle that a class action is
"an exception to the usual rule that litigation is conducted by and on
behalf of the individual named parties only." Califano v. Yamasaki,

, 700-01 (1979). Though the issue of manageability of
a proposed class action is "peculiarly within [the] discretion" of the
district court, Windham v. Am. Brands, Inc., 

, 65 (4th Cir.
1977), we will nonetheless vacate the certification of a class when
plaintiffs have not sufficiently shown the presence of all of the Rule
23(a) and (b)(3) requirements.

  Every class action must satisfy the four criteria set forth in Rule
23(a): numerosity, commonality, typicality, and adequacy of repre-
sentation. Fed. R. Civ. P. 23(a). Numerosity requires that the class be
so numerous that "joinder of all members is impracticable." Fed. R.
Civ. P. 23(a)(1). The "final three requirements of Rule 23(a) ‘tend to
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                  73
merge,’ with commonality and typicality ‘serv[ing] as guideposts for
determining whether . . . maintenance of a class action is economical
and whether the named plaintiff’s claim and the class claims are so
interrelated that the interests of the class members will be fairly and
adequately protected in their absence.’" Broussard v. Meineke Dis-
count Muffler Shops, Inc., 

, 337 (4th Cir. 1998) (quoting
Gen. Tel. Co. v. Falcon, 

, 158 n.13 (1982)). In addition,
a class action must satisfy the criteria of Rule 23(b)(1), (b)(2), or
(b)(3).

   Plaintiffs here have sought certification of a class under Rule
23(b)(3). Because Rule 23(b)(3) is designed for "situations in which
‘class-action treatment is not as clearly called for,’" plaintiffs must go
beyond satisfying the Rule 23(a) prerequisites and satisfy two addi-
tional requirements, "predominance" and "superiority." 

(quoting Fed. R. Civ. P. 23(b)(3) Advisory Committee
Notes). To satisfy the predominance requirement, plaintiffs must
show that the "questions of law or fact common to the members of
the class predominate over any questions affecting only individual
members," and to satisfy the superiority requirement, they must show
that "a class action is superior to other available methods for the fair
and efficient adjudication of the controversy." Fed. R. Civ. P.
23(b)(3). The predominance requirement is "far more demanding"
than Rule 23(a)’s commonality requirement and "tests whether pro-
posed classes are sufficiently cohesive to warrant adjudication by rep-
resentation." 

. Our recent opinions, decided
after Amchem, recognize this more demanding standard.

   In Broussard, we vacated a judgment against a defendant franchi-
sor and ruled that certification of the franchisee-plaintiff class was
error in an action where the franchisees had asserted claims for breach
of contract, breach of fiduciary duty, fraud, and negligent misrepre-
sentation, among 

. Despite our recognition that
the district court had exercised its "broad discretion" in deciding to
certify the class, we nonetheless decertified the class because the fran-
chisees had not shown that the commonality and typicality require-
ments of Rule 23(a) were met. 

The franchisees were "a
hodgepodge of factually as well as legally different plaintiffs": they
signed different contracts with the defendant, relied upon different
alleged oral misrepresentations, exhibited varying degrees of reliance
74              GUNNELLS v. HEALTHPLAN SERVICES, INC.
upon such misrepresentations, and presented factually different bases
for tolling relevant statutes of limitations. 

Moreover, the
plaintiffs in Broussard alleged damages that were "inherently individ-
ualized and thus not easily amenable to class treatment." 

We acknowledged that although the need for mini-trials on damages
is not necessarily fatal to class certification, plaintiffs’ claims for lost
profits were not a "natural candidate for classwide resolution" because
the governing state law required an individual case-by-case analysis
of such claims. 

   Most recently, in Lienhart v. Dryvit Systems, Inc., 

,
149 (4th Cir. 2001), we held that where "[t]he functional equivalent
of a full-blown trial on damages causation for each putative class
member would be required to determine to which individuals [the
defendant] is liable," the predominance requirement of Rule 23(b)(3)
is not met. This was an application of the principle we elaborated in
Windham v. Am. Brands, Inc., 

(4th Cir. 1977), that pre-
dominance may be destroyed when individualized issues regarding
damages would require a large number of separate mini-trials. See 

("[W]here the issue of damages and impact does not lend itself
to such a mechanical calculation, but requires separate mini-trials of
an overwhelming large number of individual claims, courts have
found that the staggering problems of logistics thus created make
(the) damage aspect of the case predominate, and render the case
unmanageable as a class action") (quotation marks and internal cita-
tions omitted).

  It is against this background that we must consider the class-action
certification.

                                    IV

   Presented with a seven-count complaint against TPCM and 43
agents, the majority has affirmed the conditional certification of a
class of 1,400 insureds and their families against TPCM under Rule
23(c)(4) to answer only "whether TPCM mismanaged the Fund" and
"whether TPCM’s mismanagement was a proximate cause of the
Plan’s collapse." Ante at 15 (emphasis added).4 In relying on Rule
  4
   The majority’s evolving definition of the certified issue now also
includes "whether the Fidelity Defendants"—who are not a party to the
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                 75
23(c)(4) and not the complete analysis of Rule 23(b)(3), the majority
fails to answer how these common questions predominate and how
this fractured class action advances judicial efficiency. The certified
questions do not resolve even a major portion of the issues that must
be decided individually, including up to 1,400 mini-trials on damages
and causation alongside 1,400 individual full trials on the civil con-
spiracy claims, RICO claims, and claims under the South Carolina
Unfair Trade Practices Act. The majority also concludes that the
requirements of Rule 23(a) have been satisfied despite substantial
conflicts of interest that are evident based upon the minimal discovery
already conducted.

   In the majority’s view, this case is a near carbon copy of Central
Wesleyan, a nationwide class action against asbestos manufacturers in
which we affirmed the conditional certification of a class. In my view,
however, the class certified here, unlike the one certified in Central
Wesleyan, is a hodgepodge of dissimilarly situated plaintiffs who
dealt with the defendants at different times and under different cir-
cumstances and who have antagonistic interests. The class here is
cobbled together to resolve very little before ultimately relegating the
hard questions to individual trials. Although the majority’s effort to
fashion a certifiable class under our precedents appears, on the sur-
face, to aid the aggrieved plaintiffs, I believe the majority deviates
substantially from the requirements of Rule 23, in the end denying
procedural fairness to all of the litigants. I begin by demonstrating the
plaintiffs’ most obvious failure — their failure to show that the com-
mon questions "predominate over any questions affecting only indi-
vidual members." Fed. R. Civ. P. 23(b)(3).

  The class as approved by the majority consists of employers,
employees, and individuals who purchased or used healthcare benefits
provided through a Plan sponsored and marketed by the Fidelity
Group and related companies over a period of three years. During the
course of those three years, claims under the Plan were administered
by TPCM during 1996 and part of 1997 and were then administered

class that was certified—"also mismanaged the Fund, and whether the
Fidelity Defendants’ mismanagement was an intervening cause of the
Plan’s collapse sufficient to relieve TPCM of liability." Ante at 15.
76              GUNNELLS v. HEALTHPLAN SERVICES, INC.
in-house by Fidelity during the remainder of 1997 until the Plan’s
demise in January 1999. Having suffered delayed payment of claims
or nonpayment allegedly due to claims mismanagement, with injuries
ranging from damaged credit ratings to emotional distress, plaintiffs
seek recovery from the administrators of the Plan, the Plan itself, and
the 43 agents who sold the Plan. It is the diversity of the factual and
legal bases underlying these claims that confounds the plaintiffs’ bold
effort to resolve them through the class action process. Indeed, the
majority recognizes that the individualized inquiries that pervade
plaintiffs’ claims against the individual agents preclude certification
of any classes against them. And both the majority and the district
court recognized that plaintiffs’ claims based on the South Carolina
Unfair Trade Practices Act, RICO, and civil conspiracy must proceed
individually. That leaves for potential class certification four of the
seven claims against TPCM, which sound in negligence and contract.
Yet the evidence that will need to be presented in the class action
against TPCM will need to be repeated to some extent in the individ-
ual trials against TPCM. For example, the complaint alleges that
TPCM, among other defendants, violated the South Carolina Unfair
Trade Practices Act by not providing timely coverage and not
explaining why payment was delayed. This claim, which must, by
statute, be tried individually by each class member, will of course
require proof that TPCM was responsible for making payments and
did not make such payments to that individual plaintiff in a timely
fashion, thereby causing that plaintiff injury. The substantial effort in
consolidating the "common question" in this case thus will yield
remarkably little in advancing the class members’ individual claims
toward a resolution.

    The majority is persuaded that, despite all of the differences among
the class members that require so many individual trials on the most
difficult issues, the common question in the class members’ claims of
whether TPCM’s alleged mismanagement of claims "contributed to"
or was "a cause" of the Plan’s failure "predominate[s] over any ques-
tions affecting only individual members" because Rule 23(c)(4)(A)
authorizes the certification of individual issues. But this does not ful-
fill the requirement that whole "action" be scrutinized to determine
whether individual claims predominate.

  Although I recognize that many — but not all — of the class mem-
bers’ claims require resolution of whether TPCM contributed to the
                 GUNNELLS v. HEALTHPLAN SERVICES, INC.                 77
collapse of the Plan, this is an underwhelming commonality in light
of the substantial individualized inquiries necessary to determine lia-
bility, not to mention damages. If the court proceeds under the class
structure as proposed and determines that TPCM’s claims misman-
agement was, in fact, a cause of the Plan’s failure, the liability inquiry
will only have just begun. The subsequent and more difficult ques-
tions, to be litigated individually by class members whose claims
were handled either by TPCM or Fidelity between 1996 and 1999,
would persist, namely, whether the failure of the Plan caused the class
member’s injury and the more complex question of comparative fault.

   Although the class as certified presumes a unified group of plain-
tiffs proceeding on a single legal theory of indirect liability, the com-
plaint for good reasons asserts differing theories of recovery requiring
the presentation of different, even conflicting, evidence, which under-
mines predominance. Specifically, the complaint alleges two theories
of liability against TPCM: (1) direct liability, as to those who, for
example, suffered injury to credit ratings solely because of a delayed
payment during the time TPCM managed claims, and (2) indirect lia-
bility, as to those whose claims went unpaid or were paid late by
Fidelity as a result of TPCM’s prior claims mismanagement. The
majority, recognizing that its "single theory" analysis "appears some-
what at odds" with the complaint and inconveniences its analogy to
Central Wesleyan, has apparently accepted an informal disavowal of
the direct theory of liability in appellees’ brief. But oral argument
clarified that appellees do not disavow anything in their complaint in
this regard. At oral argument, appellees’ counsel argued: "There is not
a single claim, and they can’t point to one, where we suggest that
Third Party Claims Management did anything with any specific claim
wrong." But counsel did not disavow direct liability claims; he simply
portrayed direct liability claims as something other than what they are.5
  5
   Questioning by the court at oral argument revealed that plaintiffs’
counsel was not disavowing anything in the complaint, but rather making
the demonstrably false argument that the complaint never alleged any
direct theory of liability against TPCM:
      Judge Motz: You have a direct argument when TPCM is still in
      there.
      Counsel: Your Honor, we’re not making that argument at all.
78               GUNNELLS v. HEALTHPLAN SERVICES, INC.
Counsel’s argument betrays the language of the complaint, which
clearly asserts direct claims against TPCM:

     Defendant [TPCM] breached their duty [by] . . . failing to
     timely pay claims submitted to them; . . . charging fees in
     excess of what was allowed by agreement; failing to inform
     plaintiffs and other members of the class of their lack of
     experience and mismanagement of claims processing; fail-
     ing to provide plaintiffs and other members of the class with
     accurate information concerning the status of claims submit-
     ted; in other words failing to exercise due care in the pro-
     cessing of claims submitted . . . ; in providing misleading,
     false and/or fraudulent information about why claims were
     not being paid. . . . As a direct result of Defendants’ negli-
     gent conduct, the Plaintiffs have been injured in numerous
     ways.

     Judge Motz: Okay, so you’re disavowing that argument.
     Counsel: We never made that argument.
   By embracing a disavowal that even plaintiffs’ counsel disclaimed he
was making, the majority leaves the plaintiffs with only direct liability
claims against TPCM out in the cold. As the majority points out, the
South Carolina statute of limitations appears to have run on actions for
breach of contract and fraud, ante at 21. That leaves these plaintiffs who
cannot — or prefer not to — proceed on an indirect liability theory to file
their own individual actions, and exposes them to arguments that their
claims are now untimely. Although equitable tolling might arguably res-
cue these unwanted plaintiffs from having their claims dismissed, the
need potentially to employ such an extraordinary measure only bespeaks
the peculiarity of the majority’s cutting these plaintiffs’ claims out of the
complaint to certify the surviving allegations. Perhaps recognizing the
severity of holding that the plaintiffs do not assert direct claims, the
majority now alternatively states that the plaintiffs "do not seek class cer-
tification for any direct claims against TPCM," but this conclusion, too,
flies in the face of the complaint and the motion for class certification,
both of which seek certification of all claims against TPCM. The direct
claims quite clearly inconvenience the majority’s analysis.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                 79
These allegations assert direct claims against TPCM based on indi-
vidual failures that encompass more than the alternative theory of
indirect liability flowing from the Plan’s failure.

   Because counsel did not disavow the complaint, but simply mis-
characterized it, I am unwilling to concede that these claims have
been removed from the case. It would be a questionable decision for
plaintiffs’ counsel to abandon the theory of liability that best suits an
identifiable segment of the certified class. For instance, a class mem-
ber whose claim was simply delayed, but ultimately paid, in 1996,
when TPCM was the sole administrator, would have only a direct
claim against TPCM for mismanagement, as the damage to that plain-
tiff was in no way related to the collapse of the Plan.6 Jamming these
unfortunate class members’ claims into the single-theory class certi-
fied here — which purports to proceed only on the theory of indirect
liability — not only requires a representative party to shoulder the
more attenuated theory of recovery for some of her fellow class mem-
bers but also actually precludes recovery for those who were in the
circumstance where the collapse of the Plan did not cause injury. And
even as to class members whose claims were ultimately unpaid, per-
sons who might have a direct claim against TPCM find themselves
sacrificed by counsel and caught in a class that requires them to show
not only that TPCM mismanaged claims but also that the mismanage-
ment led to the failure of the Plan and that the failure of the Plan
caused their injuries. This theory may suit some class members, par-
ticularly those whose claims went unpaid by Fidelity after the termi-
nation of TPCM’s contract, but it certainly is not fair to all class
members. The majority opinion now precludes any recovery by any
class member on a direct theory of liability against TPCM, regardless
of the merits of their case. Ante at 18, 22 ("[C]lass claims against
TPCM . . . rest on a single theory: that TPCM’s mismanagement of
claims contributed to the ultimate collapse of the Plan and so caused
Plaintiffs’ damage. . . . [Although] somewhat at odds with the Plain-
tiffs’ original complaint . . . Plaintiffs have made abundantly clear on
appeal . . . they do not seek class certification for any direct claims
against TPCM").
  6
   Indeed, counsel’s reluctance to pursue both direct and indirect claims
suggests their inability to represent absent class members whose only
claims are direct claims.
80              GUNNELLS v. HEALTHPLAN SERVICES, INC.
   By accepting the plaintiffs’ mischaracterization of the complaint —
which the majority treats as a disavowal of the direct theory of liabil-
ity — the majority also tacitly recognizes the conflict that inheres
when a single class proceeding on a single theory purports to repre-
sent plaintiffs with dissimilar theories of recovery. This court’s
response to a complex class action complaint alleging differing theo-
ries of recovery should not be to prune allegations and entire theories
of recovery until a class remotely certifiable emerges. Cf. 

n.17 ("[W]e should not condone a certification-at-all-costs
approach to this case for the simple purpose of forcing a settlement").
Rather, we should take the complaint as written, recognizing that the
conflicts and complexities obvious from a plain reading counsel that
this case be conducted by individual parties or by such small groups
of parties as may be appropriate.7 In my view, the existence of con-
flicting theories of recovery not only reveals the plaintiffs’ failure to
demonstrate that the common issues predominate but also reveals the
inadequacy of a single representative for these disparate theories.

   The majority also never addresses the complex question of compar-
ative fault. For example, a class member whose payment was delayed
but ultimately paid in 1996, when TPCM was the sole claims admin-
istrator, might prove that TPCM’s mismanagement rendered it 100%
liable for the class member’s injury, whether it be damage to credit
rating or emotional distress. In that case, the failure of the Plan would
  7
    The majority’s treatment of the plainly pleaded direct theory of liabil-
ity is mystifying. Perplexingly, the majority relies upon the liberal theory
of pleadings to justify its excision of a pleaded theory of liability from
the complaint, seemingly for the convenience of its certification analysis.
Ante at 22. Having excised that theory of liability from the complaint, the
majority then treats with indignation what it calls my "disingenuous"
suggestion that rewriting the complaint has done a disservice to an iden-
tifiable group of plaintiffs. Ante at 23 n.8. Making reference to excep-
tions to claim-splitting and the opt-out feature of Rule 23 to surround its
selective redrafting of the complaint in an air of legitimacy, the majority
does little to address why it is fair to any plaintiff to excise a timely
pleaded theory of recovery from the complaint, leaving any plaintiffs
wishing to assert the theory they thought they had alleged to plead it
anew, albeit now likely untimely. I would rather give the plaintiffs the
benefit of the fullness of their claims as timely plead, in accordance with
the liberal theory of pleadings.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                  81
be irrelevant to the cause of the delayed payment. Another class mem-
ber, whose payment was delayed or went unpaid in the summer of
1997, shortly after claims management went in-house at Fidelity,
might ultimately prove that TPCM was 42% or 70% liable, or even
0% liable if TPCM proved that some independent inadequacy of
Fidelity caused a particular claim to go unpaid. Yet another class
member, whose claim went unpaid in 1999, almost two years after
claims administration went in-house, might be able to prove even
lesser responsibility on TPCM’s part. The evidence necessary to
establish comparative fault — evidence regarding what TPCM did
wrong in a particular case versus what Fidelity did wrong in that case
— would require a second presentation during individual trials of
much of the evidence, albeit individualized, that the majority believes
the certified class structure allows to be presented only once. And, of
course, a given class member’s negligence in getting the proper
paperwork to the insurer could constitute a cause of delayed payment
or nonpayment, which negligence would bar recovery if that negli-
gence exceeded that of the administrator. See Davenport v. Cotton
Hope Plantation Horizontal Prop. Regime, 

, 571 (S.C.
Ct. App. 1997) ("[A] plaintiff in South Carolina may recover only if
his negligence does not exceed that of the defendant’s").

   Even apart from the direct-indirect claims problem and the individ-
ualized comparative liability questions, which include inquiries into
intervening causes of maladministration in particular cases, the indi-
vidual questions of entitlement to payment under the terms of the Plan
and the amount of any such entitlement also loom. This would surely
require, among other things, submission of proof relating to timely fil-
ing of a claim, proof of medical services rendered, and the like.
Indeed, some claimants have had their claims qualified for coverage
but not payment, and some have not been qualified for coverage.
Those whose claims have been adjudicated may challenge the bene-
fits determination, and TPCM has already indicated that it intends to
look at the accuracy of individual claims adjudications, if they will be
determinative of liability. Those whose claims have not yet been adju-
dicated would have to establish the facts that would qualify them for
coverage, which would include the nature of the illness or injury for
which the claim is made, deductibles and exclusions, doctor docu-
mentation, and similar matters. And, of course, as the district court
noted, in addition to all of this, "an individualized inquiry is necessary
82              GUNNELLS v. HEALTHPLAN SERVICES, INC.
for damages such as damage to one’s credit rating as well as emo-
tional distress and mental anguish."

   It is thus difficult to conceive what advantage is gained by certify-
ing a class premised on such a sliver of commonality. The individual-
ized evidence necessary to determine (1) comparative liability of
TPCM versus Fidelity based, in part, on the date the class member (or
medical provider) filed a claim, (2) the class member’s (or medical
provider’s) negligence, if any, in causing the claim to be delayed or
go unpaid, (3) any intervening causes of maladministration, (4) the
class member’s entitlement to payment based on the nature of the ser-
vices received and whether they are covered, (5) the amount of enti-
tlement, considering copayments due, deductibles, and other sources
of benefits, and (6) the existence and amount of other damages, such
as mental anguish, will surely require presentation of substantially all
of the evidence necessary to make the unhelpful determination that
TPCM was a cause of the Plan’s failure. And to the extent the Plan’s
failure can be shown to be a cause of a class member’s injury —
which will not always be the case, of course — there remain ques-
tions of whether there were other parties potentially responsible for
the Plan’s failure.

   Because the complaint alleges seven distinct causes of action and
the district court has determined that approximately one-half of these
claims will have to be resolved through individual trials, the resulting
complexity itself should counsel against class-action treatment. Nei-
ther the district court nor the majority have even begun to address
how the trials and mini-trials will proceed; how the law of the case
or res judicata will be applied; and how evidence will be received
when most of it will relate to various individual plaintiffs, individual
defendants, and individual claims. The combination of the various tri-
als — up to 1,400 mini-trials, individual full trials as to all defendants
on civil conspiracy, RICO violations, and South Carolina statutory
violations, and a class-action trial involving the differing claims
against TPCM — will produce a hodgepodge of factual findings and
legal rulings, perhaps conflicting, that will have to be sorted out and
applied either by the jury or by the court as the law of the case. This
will create a minefield for legal error.

   Given the overwhelming breadth of the issues reserved for separate
full trials and individual mini-trials, I conclude that the common
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                    83
issues in this case do not predominate over the individual ones.

("Given the greater number of questions
peculiar to the several categories of class members, and to individuals
within each category, and the significance of those uncommon ques-
tions, any overarching dispute about the health consequences of
asbestos exposure cannot satisfy the Rule 23(b)(3) predominance
standard").

   In addition, the substantial judicial effort required to manage proce-
durally a case in which the same evidence will be presented to the
court both individually and on a classwide basis is not justified by the
minimal contribution to the case that an answer to the certified ques-
tion provides. It is questionable whether, in fact, any time will be
saved by certifying the class against TPCM. In any event, the proce-
dural nightmare such a certification invites cannot be thought to come
close to satisfying the requirement of Rule 23(b)(3) that the class
structure be superior to other available methods of adjudicating the con-
troversy.8 See Fed. R. Civ. P. 23(b)(3).
  8
    The majority finds comfort in running to the "conditional" aspect of
certification, stating that "we are only approving a conditional certifica-
tion of the case, and . . . in the unlikely event the case becomes a ‘proce-
dural morass,’ . . . we are confident that the district court will . . .
decertify the class." Ante at 44. This view defaults our responsibility and
employs a backward burden in demonstrating the superiority of the class
action device. The majority apparently believes that no matter how com-
plex, a complaint comes to the court bearing an entitlement to certifica-
tion unless the court can prove why it should not be certified. This view
is reflected in the court’s willingness to excise inconvenient theories of
recovery from the complaint and its willingness to elevate Rule 23(c)(4)
to allow certification of any common issue, presumably subject to some
unspoken limiting principle. But the fact remains that a class action is an
"exception to the usual rule" that litigation is to be pursued by the indi-
vidual parties, 

(emphasis added), that the
party seeking certification bears the burden of proving satisfaction of the
class-qualifying criteria of Rule 23(a) and (b), and that the court is to
measure plaintiff’s ability to meet that burden against the action as
alleged in the complaint. The alternative I suggest, therefore, is that this
case proceed according to the usual rule that the claims be asserted by
the individual named parties, and I note the availability of our rules of
civil procedure to achieve economy even without a class action. See Ben-
84              GUNNELLS v. HEALTHPLAN SERVICES, INC.
   Thus, taking the class against TPCM as the district court has
defined it in its September 28, 2001 order, one can only conclude (1)
that the questions of law or fact common to members of the class do
not predominate over the questions affecting individual members and
(2) that the class-action process is not superior in this case. The
majority’s conclusion to proceed with one or two issues to the con-
trary, through class treatment under Rule 23(c)(4)(A) in my view,
enlarges the reach of Rule 23(b)(3) beyond its proper scope because
the constraints of predominance and superiority have not been
applied.

   I also believe that the class, as conditionally certified, does not sat-
isfy the requirement that the class representatives fairly and ade-
quately protect the interests of all of the class members. See Fed. R.
Civ. P. 23(a)(4). In my view, the majority ignores overt conflicts of
interest existing among members of the conditionally certified class.
Fundamentally, it does not mention or attempt to explain how the
conditionally certified class may include, on one hand, employers
seeking rescission of the insurance contract and return of the premi-
ums paid, and, on the other hand, employees seeking enforcement of
the insurance contract and benefits due under it. In addition to this
conflict, the employee class members have potential common-law
claims against their employers, who purchased the Plan, for negli-
gence or for misrepresenting the Plan.9 The incompatible relief sought
by the employers and the employees alone precludes adequate repre-
sentation of both groups by one representative, and this concern is
only heightened by the potential for lawsuits by employees against the

jamin Kaplan, Continuing Work of the Civil Committee: 1966 Amend-
ments of the Federal Rules of Civil Procedure (I), 81 Harv. L. Rev. 356,
391 (1967)("[T]he procedural alternatives are hardly confined to the
class action, on the one side, and the individual uncoordinated lawsuits,
on the other; there are often other possibilities ranging from use of a
model action to consolidation or coordination of the numerous individual
actions for all or selection purposes")(citation omitted).
   9
     It may well be, as the majority notes, that many of the claims the
employees may have against their employers are now barred by the stat-
ute of limitations, depending on what theory of when a third-party claim
accrues is applied.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                 85
employers. While the majority addressed only some of the conflicts
presented, it gave any notion that conflicts are untenable in class rep-
resentation the back of its hand.

   The district court also recognized some of these conflicts but
resolved them only with the observation that they could be cured by
the class members’ right to opt out of the class action. But this obser-
vation does not address a fundamental deficiency of a conflict among
class members nor the burden that the plaintiffs must carry in seeking
to represent the class as a whole, including those who, for whatever
reason, never opt out. Federal Rule of Civil Procedure 23(a)(4)
requires plaintiffs, as purported representatives of the class, to demon-
strate that they can fairly and adequately protect the claims of class
members. They obviously cannot fulfill this responsibility if some
class members’ claims are in conflict with others’ claims. See

(finding the plaintiffs could not fairly
and adequately represent a class where, within the class, there was a
"disparity between the currently injured and exposure-only categories
of plaintiffs" making asbestos claims).

   Given the relatively insubstantial common issues shared by only
some of the class members and the irreconcilable conflicts of interest
inhering in the certified class, I do not agree with the majority that
this case is "strikingly similar" to Central Wesleyan. Central Wes-
leyan was an asbestos removal case in which 16-23% of America’s
colleges and universities were potential class members and which
involved dozens of defendants and hundreds of asbestos products sold
over 

, 189. The issues conditionally certified
for class treatment were "primarily factual," 

in a context
where "the sheer volume of litigation" in the area impelled many
courts nationwide to conclude tentatively that class treatment of cer-
tain common factual issues in this very particular type of litigation
advanced judicial economy and fostered settlement, 

185. Conditionally certifying the primarily factual questions for class
treatment, the court held, would relieve the colleges and universities
of the need "to prove over and over when defendants knew or should
have known of asbestos’ hazards, or whether defendants engaged in
concerted efforts to conceal this knowledge, or even whether certain
of defendants’ products crumble and release dust under hand pres-
sure." 

86              GUNNELLS v. HEALTHPLAN SERVICES, INC.
   This case is unlike Central Wesleyan. This is not a massive nation-
wide litigation in which resolving certain primarily factual questions
on a classwide basis would substantially reduce repetitive factfinding.
The only question certified here is whether TPCM was "a cause" of
the Plan’s failure, which is a mixed question of fact and law, and
which, as noted, is only the first step in a causation inquiry that is
steeped in necessary individualized determinations requiring the pre-
sentation of often duplicative evidence. While the court in Central
Wesleyan countenanced individualized inquiries as to damages, it
noted that the potential for individualized inquiries relating to liability
could "pose management difficulties and reduce the judicial effi-
ciency sought to be achieved through certification," indicating that
decertification might be warranted. 

Here, the need for indi-
vidualized determinations as to causation is not hypothetical or even
unique to a few class members but rather is a characteristic of every
claim of every class member.

   The complexity of this case counsels that our decisions in Brous-
sard and Lienhart are far more apt comparisons than Central Wes-
leyan, portions of which may even have been superseded by the
Supreme Court’s analysis in Amchem. In both Broussard and Lien-
hart, we concluded that the district court abused its discretion in certi-
fying the class. In neither of those actions was there such an array of
individual actions that required individual adjudication as those pre-
sented in this case. Indeed, in Lienhart we stated that the requirement
of individual trials on damages causation alone would require a find-
ing that the predominance requirement of Rule 23(b)(3) was not 

. The majority glosses over the complexities in this
case, overlooking what clearly distinguishes this case from Central
Wesleyan.

   Rule 23(b)(3) was designed to effect efficiencies without sacrific-
ing procedural fairness. As the Advisory Committee Notes to the
1966 amendments to Rule 23 state:

     Subdivision (b)(3) encompasses those cases in which a class
     action would achieve economies of time, effort, and
     expense, and promote uniformity of decision as to persons
     similarly situated, without sacrificing procedural fairness or
     bringing about other undesirable results.
                GUNNELLS v. HEALTHPLAN SERVICES, INC.                 87
   The drafters of Rule 23 understandably counseled a cautious
approach with respect to Rule 23(b)(3) classes, recognizing that even
under their limited understanding as to how broadly it might apply,
the idea was adventuresome and was not "as clearly called for" as
class actions under (b)(1) or (b)(2). See 

.
This cautious and constricted approach to Rule 23(b)(3) envisioned
by its authors should not be relaxed now simply by the passage of
time and the additional experience that courts have had with the Rule.
Such comfort would tend to transmogrify what is already an adven-
turesome experiment into one applied for its own sake and not for the
sake of procedural efficiency and fairness. As the Court stated in
Ortiz v. Fibreboard Corp., 

, 861 (1999), "we are bound
to follow Rule 23 as we understood it upon its adoption, and . . . we
are not free to alter it except through the process prescribed by Con-
gress in the Rules Enabling Act." (Emphasis added). Because of its
already adventuresome contours, the original understanding of Rule
23(b)(3) class actions clearly contemplated that courts would conduct
a "close look at the predominance and superiority criteria." 

. Disapproving any metamorphosis of the rule, the
Court stated in Amchem that, while a 23(b)(3) class action was
thought to be the most adventuresome innovation of the 1966 amend-
ments, 

"[i]n the decades since the 1966 revision of Rule
23, class action practice has become even more ‘adventuresome,’" an
example of which was before the Court, 

    Although the district court’s efforts in patching together a certifi-
able class against TPCM may be well-intended, energetic, and indeed,
even creative, the class is nonetheless far too adventuresome simply
to resolve a few common issues, seriously risking the creation of pro-
cedural unfairness. Predict as we can how this case might be tried as
it is now structured, I see only a queue of problems that sacrifice pro-
cedural fairness and have the potential for creating legal error. With-
out forecasting how the numerous claims might be adjudicated to
secure the "just, speedy, and inexpensive determination" of each
claim, Fed. R. Civ. P. 1, I conclude that the plaintiffs’ effort to pro-
ceed against TPCM under Federal Rule of Civil Procedure 23(b)(3),
as certified by the district court in its order of September 28, 2001,
is neither "convenient" nor "desirable," 

, and
that such a structure amounts to an adventuresome embrace that is too
enthusiastic for Rule 23’s possibilities.
88            GUNNELLS v. HEALTHPLAN SERVICES, INC.
  Accordingly, I would vacate the entirety of the district court’s
order of September 28, 2001, and remand this case for further pro-
ceedings.