PRECEDENTIAL

         UNITED STATES COURT OF APPEALS
              FOR THE THIRD CIRCUIT
                   ___________

                        No. 16-2653


JAIME GONZALEZ; PATRICIA WRIGHT; KEVIN WEST;
 GERALD BOEHM; EDWARD MAAG; DIANE MAAG,
  on behalf of themselves and all others similarly situated,
                                                    Appellants

                              v.

  OWENS CORNING; OWENS CORNING SALES LLC
                 __________

      On Appeal from the United States District Court
         for the Western District of Pennsylvania
                 (D.C. No. 2-13-cv-01378)
       District Judge: Honorable Joy Flowers Conti
                       ___________

               Argued September 6, 2017
    Before: CHAGARES, JORDAN, and HARDIMAN,
                    Circuit Judges.

              (Opinion Filed: March 19, 2018)
Robert H. Klonoff [Argued]
Jordan D. Schnitzer Professor of Law
Lewis & Clark Law School
Earthrise Law Center
10015 Southwest Terwilliger Boulevard
Portland, OR 97219

Charles E. Schaffer
Levin Sedran & Berman
510 Walnut Street, Suite 500
Philadelphia, PA 19106

Shanon J. Carson
Lawrence Deutsch
Berger & Montague
1622 Locust Street
Philadelphia, PA 19103

Charles J. LaDuca
Cuneo Gilbert & LaDuca LLP
8120 Woodmont Avenue, Suite 810
Bethesda, MD 20814

Michael A. McShane
Audet & Partners
711 Van Ness Avenue, Suite 500
San Francisco, CA 94102

Robert K. Shelquist
Lockridge Grindal Nauen PLLP
100 Washington Avenue South, Suite 2200
Minneapolis, MN 55401
      Attorneys for Appellants




                               2
Carter G. Phillips [Argued]
Sidley Austin
1501 K Street, N.W.
Washington, D.C. 20005

Kara L. McCall
T. Robert Scarborough
Tacy F. Flint
Elizabeth M. Chiarello
Sidley Austin LLP
One South Dearborn Street
Chicago, IL 60603

Arthur H. Stroyd, Jr.
Del Sole Cavanaugh Stroyd
Three PPG Place
Suite 600
Pittsburgh, PA 15222
       Attorneys for Appellees

Allan P. Ides
Simona Grossi
Loyola Law School
919 Albany Street
Los Angeles, CA 90015

Michael J. Quirk
Williams Cuker & Berezofsky
1515 Market Street, Suite 1300
Philadelphia, PA 19102
       Attorneys for Appellants’ Amici Curiae




                                 3
                OPINION OF THE COURT
                     ____________

HARDIMAN, Circuit Judge.

       This appeal involves a putative class action brought by
consumers in four states who alleged that Appellees Owens
Corning and Owens Corning Sales, LLC (collectively, Owens
Corning) sold defective roof shingles and misrepresented the
shingles’ expected useful life. Appellants challenge an order
of the United States District Court for the Western District of
Pennsylvania denying class certification. We will affirm.

                               I

       In 2009, Appellants Patricia Wright and Kevin West
filed suit in the District Court under Rule 23 of the Federal
Rules of Civil Procedure on behalf of a proposed nationwide
class of individuals who “owned, own, or acquired” structures
on which certain Owens Corning-manufactured roofing
shingles “are or have been installed since 1986.” Gonzalez v.
Owens Corning, 

, 455 (W.D. Pa. 2016).
Wright and West claimed that Oakridge shingles—an Owens
Corning brand of fiberglass asphalt roofing shingles—are
“plagued by design flaws that result in cracking, curling and
degranulation” and “will eventually fail, causing property
damage, and costing consumers substantial removal and
replacement costs.” 

The District Court entered
summary judgment in favor of Owens Corning, finding that
Wright and West’s claims had been discharged in bankruptcy
by the 2006 confirmation of Owens Corning’s reorganization
plan. We partially reversed, concluding that the claims were
not discharged. See Wright v. Owens Corning, 

,
108–09 (3d Cir. 2012). After the case was remanded to the



                              4
District Court, Appellants Jaime Gonzalez, Gerald Boehm,
and Edward and Diane Maag (together with Wright and West,
Plaintiffs) filed three similar suits in district courts in other
states, which were then transferred to the Western District of
Pennsylvania and consolidated with Wright and West’s case.

        Plaintiffs are homeowners from Pennsylvania, Illinois,
Texas, and California, on whose roofs Oakridge shingles
were installed prior to 2006. They allege that their shingles
have not performed as promised because they were
manufactured “in accordance with defective design
specifications.” 

. Of the named
plaintiffs, three reported property damage and two had their
roofs reshingled. The shingles were all subject to warranties
of 25 years or more, which Plaintiffs argue amounted to
affirmative representations about the shingles’ expected
useful life. Plaintiffs proposed two classes in the District
Court: (1) a class of property owners from their four home
states (the Four-State Class), asserting various combinations
of state-law causes of action against Owens Corning; and (2)
a nationwide class of property owners (the Nationwide Class)
seeking a ruling regarding the legal standard governing
whether Owens Corning can use a bankruptcy discharge
defense to shield itself from liability.

                               A

       Plaintiffs proffer the Four-State Class as either a
money damages class under Rule 23(b)(3) or an issue class
under Rule 23(c)(4). Plaintiffs define the class as follows: “all
individuals and entities that own a building or structure
physically located in the states of California, Illinois,
Pennsylvania, or Texas on which Owens Corning’s Oakridge-
brand shingles were installed from 1992 through 2012, and



                               5
where those shingles manifested any cracking, degranulation,
fragmentation, or deterioration during the warranty coverage

.

       During the proposed 20-year class period, Owens
Corning manufactured at least 23 kinds of Oakridge shingles
at 13 different plants around the country using more than 500
design specifications. Plaintiffs did not dispute that all of
these specifications met the applicable industry standard
(ASTM D3462), which prescribes minimum measurements
for newly manufactured shingles, such as “tear strength, net
mass, mat mass, asphalt mass, and mineral matter mass.” 

Plaintiffs contended that in lieu of industry standards,
defectiveness should be judged by the expected useful life of
the shingles as represented by the applicable warranty period.
According to Plaintiffs, compliance with ASTM D3462 did
not consistently yield shingles that would last until the
applicable warranty expired. In Plaintiffs’ view, Owens
Corning’s design specifications sometimes yielded shingles
vulnerable to “premature deterioration so that they . . . will
actually only last 15 to 20 years.” App. 247–48.

       Plaintiffs claimed that Oakridge shingles had a
propensity to fail before their warranties expired because of
one or more of the following design flaws: insufficient
asphalt quantity or quality, and insufficient mat mass and tear
strength. Based on testing he conducted on 298 shingles,
Plaintiffs’ expert Dean Rutila opined that “about half” of the
Oakridge shingles produced during the proposed 20-year
class period fell on the “low end” of Owens Corning’s
specifications, which meant they were manufactured at or
near the ASTM minimums. 

523, 528. Owens
Corning challenged the admissibility of Rutila’s opinions
under Daubert v. Merrell Dow Pharmaceuticals, Inc., 509



                               

(1993). The District Court ruled that all testimony
based on Rutila’s testing was inadmissible, but that he could
testify to the general effect of various measurements on
shingle performance based on his review of Owens Corning’s
internal documents, including design specifications and
internal communications. For example, Rutila could testify
that Oakridge design specifications produced shingles with a
significant range of measurements, with only the “high-end”
shingles capable of lasting for more than 20 years. 

. Rutila acknowledged, however, that the
“appropriateness of design specification measurements can
only be judged in relationship to each other,” 

and
that “a shingle would have to be individually inspected to
determine whether it is a nondefective or defective shingle,”

As a result, Plaintiffs could not point to “any
particular measurement or set of measurements for the
individual elements of a shingle that constitute a design
defect.” 

propose that the Nationwide Class proceed
under Rule 23(b)(1)(B) or Rule 23(b)(2), and that it include
“all individuals and entities that own a building or structure
physically located in the United States on which Owens
Corning’s Oakridge-brand shingles are currently installed,
where those shingles were purchased on or before September
26, 2006.” 

Through the Nationwide Class,
Plaintiffs seek to mitigate what they perceive is a risk of
inconsistent judgments resulting from the procedural history
of this case prior to the consolidation of Wright and West’s
action with those brought by the other named plaintiffs.




                              7
       On September 26, 2006 (the cutoff date for the
proposed class), the United States Bankruptcy Court for the
District of Delaware confirmed a reorganization plan for
Owens Corning after nearly six years of bankruptcy
proceedings. Pursuant to the order confirming the plan and 11
U.S.C. § 1141, all claims that existed against Owens Corning
as of that date were discharged. After Wright and West
brought suit but before Plaintiffs’ actions were consolidated,
Owens Corning moved for summary judgment, arguing that
the claims belonging to Wright and West existed prior to
September 26, 2006, and were thus discharged when the
reorganization plan was confirmed. The District Court
granted summary judgment in favor of Owens Corning, citing
this Court’s decision in JELD-WEN, Inc. v. Van Brunt (In re
Grossman’s Inc.), 

(3d Cir. 2010) (en banc)
(“Grossman’s”). In that case, we held that for purposes of
determining whether a claim is dischargeable in bankruptcy,
the claim arises when the claimant is exposed to the debtor’s
product or conduct, regardless of when an injury is
discovered. 

. For Wright and
West, this meant that their claim arose when they purchased
their shingles. Since that was before Owens Corning’s
reorganization plan was confirmed, the District Court
concluded that Wright and West’s claims were discharged.

       On appeal to this Court, we declined to apply the
Grossman’s rule retroactively, citing due process concerns.
See 

–09 (3d Cir. 2012). We concluded
that bankruptcy cases in which reorganization plans were
confirmed prior to the Grossman’s decision are governed by
the dischargeability standard articulated in the case that
Grossman’s overruled, namely Avellino & Bienes v. M.
Frenville Co. (In re M. Frenville Co.), 

(3d Cir.




                              8
1984). 

. Since Owens Corning’s
reorganization plan was confirmed prior to our decision in
Grossman’s, we held that Frenville governed whether Wright
and West’s claims were discharged. 

Frenville
standard, a court must look to the underlying state limitations
law to determine when a claim arises. 

& n.5. Thus,
for example, a claim brought under the law of a state in which
the discovery rule applies arises when the claimant discovers
the injury. We affirmed in part and reversed in part the
District Court’s summary judgment, agreeing with the Court
that Wright and West held “claims” under the Bankruptcy
Code, 

but concluding that the Court erred in
holding that those claims had been discharged, 

        On remand to the District Court, this case was
consolidated with the cases brought in other districts by
Gonzalez, Boehm, and the Maags. Because no class had been
certified at the time of our decision in Wright, that decision
did not bind unnamed putative class members. Hence
Plaintiffs’ desire to certify the Nationwide Class, through
which they seek a judgment declaring that Frenville governs
the dischargeability of class members’ claims—effectively
giving our decision in Wright binding classwide effect.
Owens Corning concedes that, as we made clear in Wright,
Frenville (not Grossman’s) applies in determining whether
the claims against it have been discharged in bankruptcy, and
it has repeatedly—including during oral argument—
foresworn any intention of raising a discharge defense against
Plaintiffs or future claimants. Plaintiffs nevertheless argue
that classwide adjudication of the question we faced in Wright
is necessary “[t]o avoid inconsistency and needless litigation
for other consumers.” Plaintiffs Br. 22.




                              9
                              C

       Plaintiffs moved for certification of the Four-State and
Nationwide Classes. On March 31, 2016, the District Court
denied Plaintiffs’ motion. With respect to the Four-State
Class, the District Court concluded that Plaintiffs had not met
their burden under Rule 23(b)(3) to show that “questions of
law or fact common to class members predominate over any
questions affecting only individual members,” 

, nor was it persuaded that certifying a class
under Rule 23(c)(4) to decide issues of liability was
appropriate. 1 The District Court concluded that the
Nationwide Class could not satisfy the commonality
requirement of Rule 23(a) because the only common question
it presented was not justiciable. Plaintiffs filed a timely
petition for Rule 23(f) review, which we granted.

                              II

       The District Court had jurisdiction under 28 U.S.C.
§ 1332(d). We have appellate jurisdiction over this
interlocutory appeal under 28 U.S.C. § 1292(e) and Rule
23(f).

       “We review a class certification order for abuse of
discretion, which occurs if the district court’s decision rests
upon a clearly erroneous finding of fact, an errant conclusion
of law or an improper application of law to fact.” Hayes v.
Wal-Mart Stores, Inc., 

, 354 (3d Cir. 2013)

      1
         Plaintiffs also moved for certification of the Four-
State Class under Rule 23(b)(2), which the District Court
denied. Plaintiffs have not appealed that ruling.




                              10
(internal quotation marks omitted). We review de novo
whether an incorrect legal standard has been used. 

this appeal, Plaintiffs argue that the District Court
denied certification of the Nationwide Class based on an
erroneous understanding of the requirements for justiciability
under Article III of the Constitution, and that its denial of
certification of the Four-State Class was, among other errors,
improperly grounded in its assessment of the merits, as
forbidden by the Supreme Court in Amgen Inc. v. Connecticut
Retirement Plans and Trust Funds, 

, 466 (2013).

       Every putative class action must satisfy the four
requirements of Rule 23(a) of the Federal Rules of Civil
Procedure: numerosity, commonality, typicality, and
adequacy. See Amchem Prods., Inc. v. Windsor, 

,
613 (1997). In addition to the Rule 23(a) requirements, a class
action must satisfy Rule 23(b)(1), (2), or (3). The party
seeking certification bears the burden of establishing each
element of Rule 23 by a preponderance of the evidence. See
In re Hydrogen Peroxide Antitrust Litig., 

, 307
(3d Cir. 2009). Echoing the Supreme Court, we have
repeatedly “emphasize[d] that ‘[a]ctual, not presumed
conformance’ with Rule 23 requirements is essential.” 

(quoting Newton v. Merrill Lynch, Pierce, Fenner &
Smith, Inc., 

, 167 (3d Cir. 2001)). A class action
“may only be certified if the trial court is satisfied, after a
rigorous analysis, that the prerequisites of Rule 23(a) have
been satisfied.” Gen. Tel. Co. of the Sw. v. Falcon, 

, 161 (1982). To proceed under Rule 23(b)(3), Plaintiffs’
proposed Four-State Class must meet the additional
requirements that “[common] questions of law or fact . . .



                              11
predominate over any questions affecting only individual
members” and that “a class action is superior to other
available methods for fairly and efficiently adjudicating the
controversy.” Fed. R. Civ. P. 23(b)(3).

       We agree with the District Court that the Nationwide
Class cannot satisfy Rule 23(a)’s commonality requirement
because the only common question it poses can be answered
only by way of an advisory opinion, which is forbidden by
Article III. As we shall explain, because we agree with the
District Court that the Four-State Class cannot satisfy
Rule 23(b)(3)’s predominance requirement, we need not
discuss the other requirements of Rule 23 as they relate to the
Four-State Class. Finally, the District Court’s conclusion that
a Rule 23(c)(4) issue class is not an appropriate vehicle for
the Four-State Class was not an abuse of discretion.

                               A

        Plaintiffs first argue that the District Court erred when
it failed to certify the Nationwide Class under Rule 23. Rule
23(a) requires that the named plaintiffs share with the rest of
the putative class at least one “question[] of law or fact.” Fed.
R. Civ. P. 23(a)(2); see Wal-Mart Stores, Inc. v. Dukes, 

, 359 (2011) (“[F]or purposes of Rule 23(a)(2) [e]ven
a single [common] question will do.” (second and third
alterations in original) (internal quotation marks omitted)).
And that common question must be justiciable under Article
III of the Constitution. See Neale v. Volvo Cars of N. Am.,
LLC, 

, 366 (3d Cir. 2015). In this appeal,
Plaintiffs must establish the justiciability of the following
question: what legal standard governs the dischargeability of
claims against Owens Corning? Failing that, the Nationwide
Class cannot satisfy the commonality requirement and



                               12
certification is inappropriate under either Rule 23(b)(1)(B) or
(b)(2). 

. We therefore examine the
requirements of Article III to determine whether the District
Court abused its discretion.

        “Article III . . . restricts the power of federal courts to
‘Cases’ and ‘Controversies.’” Chafin v. Chafin, 

,
171 (2013). Accordingly, federal courts may not give
“opinion[s] advising what the law would be upon a
hypothetical state of facts.” Lewis v. Cont’l Bank Corp., 

, 477 (1990) (internal quotation marks and citation
omitted). In determining whether an actual controversy exists,
we have considered “the adversity of the interest of the
parties, the conclusiveness of the judicial judgment and the
practical help, or utility, of that judgment.” Step-Saver Data
Sys., Inc. v. Wyse Tech., 

, 647 (3d Cir. 1990).
Step-Saver was a suit for declaratory relief in which the
plaintiff corporation sought a declaration that its suppliers
were responsible for any liability it may have had to its
customers as a result of the customers’ pending suits against
it. 

While Plaintiffs do not invoke the Declaratory
Judgment Act, 28 U.S.C. §§ 2201–02, they acknowledge that
their position is analogous to that of one seeking a declaratory
judgment. The Step-Saver factors are therefore appropriate
guides to our analysis. See, e.g., Coffin v. Malvern Fed. Sav.
Bank, 

, 853 (3d Cir. 1996) (applying Step-Saver
factors to justiciability analysis where relief sought was
“somewhat analogous to . . . a declaration of rights”).

        Despite Owens Corning’s assurances that it will not
raise a discharge defense, Plaintiffs maintain that the interests
of the parties are sufficiently adverse to give rise to a
justiciable case or controversy. They argue, contrary to the
District Court’s analysis, that Owens Corning would not be



                                13
precluded from raising a discharge defense. Because of the
discretionary nature of collateral estoppel when asserted by a
plaintiff who was not party to the original judgment—what
we call “non-mutual offensive collateral estoppel”—a court
could theoretically decline to find that Wright precludes a
discharge defense under these circumstances. So long as such
a possibility exists, Plaintiffs argue, the parties’ dispute
remains a live one and their interests are adverse.

        Plaintiffs’ analysis does not change the fact that the
relief they seek would come from an advisory opinion. Put in
terms of the Step-Saver factors, the judgment Plaintiffs seek
on behalf of the Nationwide Class would lack both
conclusiveness and practical utility. See 

. As Plaintiffs acknowledge, because the Frenville test
requires a fact-intensive analysis of each claim under the
applicable state limitations law, a court’s declaration that
“Frenville applies” would not determine whether the claims
held by individual members of the Nationwide Class were
discharged. That would depend on the outcome of the
Frenville analysis as applied to each claim. But Plaintiffs do
not seek this relief. Instead, they ask that we now declare—as
we did in Wright—that Frenville applies, but this time in the
class context. Plaintiffs do so as a preemptive strike in the
event Owens Corning might raise a discharge defense in
future litigation.

       This is the opposite of a conclusive judgment. It is
more akin to the bankruptcy court order we reviewed in
Coffin, which stated that a debtor’s bankruptcy confirmation
had not discharged a bank’s mortgage 

–
53. We concluded that the order constituted an advisory
opinion, since it decided “no actual controversy between the
parties.” 

The determination of whether the bank’s



                             14
lien had been discharged was “not before the court for
adjudication,” because the debtor had not moved for an order
of lien avoidance, and the bank had not yet attempted to
enforce its lien. 

would have to be
“made by another court in foreclosure proceedings[,] and the
bankruptcy court’s advice will have no legal effect.” 

Here, if Owens Corning were to raise a discharge
defense against a future claimant, the court hearing that case
would be required to decide what law governs. That
contingency is not before us. 2

      Nor are we persuaded that the judgment Plaintiffs seek
on behalf of the Nationwide Class would be of any utility to
them beyond what we held in Wright. Contrary to Plaintiffs’

      2
          Plaintiffs’ and amici’s citation to In re Nassau
County Strip Search Cases, 

(2d Cir. 2006), is
inapposite. There, plaintiffs challenged a county jail’s strip
search policy under 42 U.S.C. § 1983. The defendants
conceded the policy’s unconstitutionality, which the district
court found “removed all common liability issues,” since the
only remaining issues concerned whether individual class
members’ rights were violated. In re Nassau 

. The Second Circuit reversed, holding that rather
than defeating predominance under Rule 23(b)(3), the
concession acted as a form of classwide proof. 

But justiciability under Article III, not predominance under
Rule 23(b)(3), is at issue here. Because the Nassau County
plaintiffs still had § 1983 claims pending before the district
court, an “actual controversy between the parties” remained.

. By contrast, Plaintiffs requested that
the District Court here preempt a potential defense in a future
case.




                              15
suggestion, such a judgment would not adjudicate the rights
of the parties, since it would not decide whether, under
Frenville, a particular class member’s claims had been
discharged. It would merely repeat, on a larger scale, what
this Court has already held. We understand Plaintiffs’ concern
that Wright’s preclusive effect might be left to the discretion
of a distant tribunal in a hypothetical future case, but
Plaintiffs do not cite, and we cannot find, any support for the
proposition that this fact alone suffices to create a justiciable
question. 3

       For these reasons, we conclude that the sole common
question the Nationwide Class asked the District Court to

       3
          Plaintiffs correctly note that, as we explained in
Gayle v. Warden Monmouth County Correctional Institution,
“courts should exercise great caution before denying class
certification” on the basis that it is unnecessary. 

,
310 (3d Cir. 2016). But contrary to their suggestion, our
conclusion that certification is inappropriate here is consistent
with Gayle. In that case, we reversed a denial of class
certification predicated on the conclusion that it would serve
“[n]o useful purpose” after the district court’s merits rulings.

We held that “necessity is not a freestanding
requirement justifying the denial of class certification,” and
that district courts can consider it only to the extent it bears
on the ability of a putative class to satisfy the requirements of
Rule 23. 

Here, the District Court did not treat the
usefulness of the relief sought by Plaintiffs as a freestanding
requirement, but rather as a component of its justiciability
analysis. As we have explained, that analysis was necessary
to determine if Plaintiffs could satisfy the commonality
requirement of Rule 23.




                               16
answer was not justiciable under Article III. The District
Court rightly concluded that Plaintiffs could not satisfy the
commonality requirement of Rule 23(a) and therefore did not
err when it denied Plaintiffs’ motion to certify the Nationwide
Class. 4

                               B

        We turn next to the District Court’s decision regarding
the Four-State Class. The predominance requirement of
Federal Rule of Civil Procedure 23(b)(3) “tests whether
proposed classes are sufficiently cohesive to warrant
adjudication by representation.” In re Hydrogen 

–11 (quoting 

).
Predominance requires “[i]ssues common to the class [to]
predominate over individual issues[,]” In re Prudential Ins.
Co. Am. Sales Practice Litig., 

, 313–14 (3d Cir.
1998), a “far more demanding” standard than the
commonality requirement of Rule 23(a), 

–24. To assess whether predominance is met at the class
certification stage, a district court must determine whether the
essential elements of the claims brought by a putative class
are “capable of proof at trial through evidence that is common

       4
          Plaintiffs complain that despite their request at oral
argument for certification under Rule 23(b)(2) as an
alternative to certification under Rule 23(b)(1)(B), the District
Court’s opinion does not address the appropriateness of Rule
23(b)(2) as a vehicle for the Nationwide Class. The need for
such an analysis, however, was obviated by the District
Court’s conclusion that the Nationwide Class failed to satisfy
the commonality requirement of Rule 23(a), which applies to
class actions under both Rule 23(b)(1)(B) and (b)(2).




                               17
to the class rather than individual to its members.” In re
Hydrogen 

–12. This determination
requires a “rigorous assessment of the available evidence and
the method or methods by which plaintiffs propose to use the
evidence to prove” these elements. 

“If proof of the
essential elements of the cause of action requires individual
treatment, then class certification is unsuitable.” 

. Accordingly, we examine the essential elements
of Plaintiffs’ claims on behalf of the Four-State Class, as well
as the evidence they propose to use, “through the prism” of
the predominance requirement to determine whether the
District Court properly denied class certification. Marcus v.
BMW of N. Am., LLC, 

, 600 (3d Cir. 2012)
(citation omitted).

       Plaintiffs proffer numerous claims, but each can be
categorized as one of the following: breach of express
warranty, breach of implied warranty of merchantability,
violation of state consumer protection statutes, and unjust
enrichment. In the District Court, Plaintiffs pointed to two
questions central to all claims that, on their view,
predominated over individual issues: whether Oakridge
shingles suffer from a common defect, and if so, whether
Owens Corning misrepresented the shingles’ expected useful
life. As the District Court observed, the defect question is
primary, because success on each claim requires a finding
that Oakridge shingles are defectively designed. This is
because “[t]he only misrepresentations or omissions that
Owens Corning is accused of making are that Oakridge-brand
shingles will last for at least 25 years, or for the same number
of years as the limited shingle warranty.” 

. Owens Corning argued that “[i]f a class
member’s Oakridge shingle was not defective, no warranty




                              18
was breached. . . . Similarly, if [the] . . . shingles were not
defective, then Owens Corning’s conduct was not . . .
actionable under the state consumer fraud acts; nor has
Owens Corning unjustly retained any benefit.” App. 2824.
Defective design is therefore an essential element of
Plaintiffs’ misrepresentation-based claims, and whether it is
susceptible to classwide evidence is dispositive of whether
Plaintiffs can satisfy predominance.

        While the parties agreed that the defect question
underpins this case, they disagreed about the implications of
this fact for the predominance analysis. Plaintiffs argued that
the defect question was common and predominant for
purposes of Rule 23(b)(3). Owens Corning countered that
while the question may be essential to this litigation, it could
not be “proven using classwide evidence,” and thus neither
could Plaintiffs’ misrepresentation claims. App. 2824. During
the proposed class period, Owens Corning noted, 23 types of
Oakridge shingles were designed according to 500 different
specifications. As Plaintiffs acknowledged, the useful life of a
shingle could be estimated only upon inspection of that
shingle. Finally, Owens Corning argued that Plaintiffs’
admission that a significant proportion of Oakridge shingles
may in fact last the length of their warranties, i.e., lack any
defect, distinguishes this case from the defective-products
cases Plaintiffs cited in which courts certified a class after
finding a predominant common question.

       In response, Plaintiffs argued that their theory of
defective design did not require them to show that all
Oakridge shingles were prone to fail during their warranty
periods. According to Plaintiffs, all Oakridge shingles could
be considered defectively designed, regardless of their actual
measurements or performance, because Owens Corning’s



                              19
design specifications provided for a range of measurements
that resulted in some shingles having a higher-than-advertised
likelihood of failing before the warranties expired. It therefore
did not matter that Plaintiffs’ expert could not identify the
particular measurements that supposedly rendered the
Oakridge specifications defective. Regardless of the quality
of the shingles on their own roofs, all Oakridge customers had
unknowingly entered a “shingle lottery,” 

–79. Because this theory of defect applies to the
entire class, Plaintiffs argued, a predominant common
question existed and Rule 23(b)(3) was satisfied.

        The District Court rejected this argument and agreed
with Owens Corning that it would be “impossible for
plaintiffs to meet their burden to prove a design defect by
evidence common to the class.” 

Because the
presence of a design defect was an essential element of
Plaintiffs’ misrepresentation-based claims, the District Court
concluded that they could not be proved by common
evidence. Given that, as Plaintiffs acknowledged, the design
defect and misrepresentation issues form the core of their
theory of liability, the Court concluded that the inability to
prove these issues through classwide evidence was fatal to
predominance and thus to certification of the Four-State
Class.

        Plaintiffs’ theory of design defect—which absolves
them of the need to identify any particular problems with
their shingles—is the focus of Plaintiffs’ appeal. According to
Plaintiffs, the District Court’s rejection of this theory was an
abuse of discretion because the District Court: (1) misread
product-defect cases in this circuit and others; (2) improperly
excluded expert testimony; and (3) contravened Amgen by
assessing the merits of Plaintiffs’ claims at the class



                               20
certification stage. In the alternative, Plaintiffs argue that the
District Court should have certified a “liability-only class”
under Rule 23(c)(4) because that “resolution of the common
liability issues would materially advance the litigation.”
Plaintiffs Br. 21. We will address these arguments in turn.

                                1

        Plaintiffs argue that, because customers were “playing
roulette in assuming that Oakridge shingles will last for the
full warranty period,” Plaintiffs Br. 15, they “did not get the
benefit of the[ir] bargain,” regardless of their shingles’ actual
performance. Plaintiffs Br. 40. Consequently, they liken their
case to a number of product-defect cases in which courts have
certified a class despite an alleged defect not manifesting
itself in each product. For example, they cite Sixth and
Seventh Circuit cases involving Whirlpool washing machines
that tended to accumulate mold because of an alleged design
defect. In each case, class certification was upheld despite the
absence of mold accumulation in the majority of class
members’ washing machines. See In re Whirlpool Corp.
Front-Loading Washer Prods. Liab. Litig., 

(6th
Cir. 2013); Butler v. Sears, Roebuck & Co., 

(7th
Cir. 2012), cert. granted, judgment vacated, 

(2013), judgment reinstated on remand, 

(7th
Cir. 2013).

       The cases involving Whirlpool washing machines are
not on point because there is a critical distinction between a
latent defect and a non-existent one. The defect in the
washing machine cases was allegedly present in all washing
machines manufactured under a particular line, even if the
defect had not yet manifested itself. See In re Whirlpool
Corp. Front-Loading Washer Prods. Liab. 

847. Plaintiffs in those cases had adduced evidence at the
class certification stage indicating that the allegedly defective
washing machines had all been built according to nearly
identical designs. See 

the various types of
front-loading washing machines at issue shared “nearly
identical engineering”); see also 

(noting “all Kenmore-brand frontloading ‘high efficiency’
washing machines” were alleged to suffer from the same
defect). This meant that the lack of a mold problem in some
machines was the result of latency, not the absence of a
defect.

       By contrast, Plaintiffs here do not identify a particular
defect that can be attributed to all Oakridge shingles. They
instead admit that a great many Oakridge shingles will last
through the end of their warranty periods, and that a shingle-
by-shingle inspection is necessary to distinguish ones that are
likely to fail before the end of their warranty periods from
ones that are likely to perform as expected (i.e., that are not
defective). This case is thus unlike those in which the latency
of an alleged defect did not pose an obstacle to certification.

       For the same reasons, Plaintiffs’ citations to Rikos v.
Procter & Gamble Co., 

(6th Cir. 2015); Wolin
v. Jaguar Land Rover N. Am., LLC, 

(9th Cir.
2010); Pella Corp. v. Saltzman, 

(7th Cir. 2010)
(per curiam); and Daffin v. Ford Motor Co., 

(6th Cir. 2006) are not persuasive. In each case, the plaintiffs
were able to identify a particular defect that, whether it had
manifested itself, allegedly existed in each of the relevant
products. See 

(noting the allegation
that nutritional supplement is “snake oil” and yields no
benefit to any class members); 

(noting that a geometric defect in vehicle alignment was



                               22
alleged to be present in each class member’s car); 

(noting that all ProLine casement windows were
allegedly designed to allow water to seep behind aluminum
casing, accelerating wood rot); 

(observing that allegedly defectively designed throttle body in
named plaintiff’s vehicle was identical to that found in all
vehicles covered by the class definition). Not only do
Plaintiffs fail to identify an alleged defect common to all
Oakridge shingles, they fail to specify where within a range
of measurements a particular design “crosses the line from
producing nondefective products to producing defective
products or to quantify how often defective products, versus
nondefective products, were produced.” 

.

        As we noted, Plaintiffs attempt to circumvent the need
to identify a common defect by, in effect, redefining the
concept to include a subset of defective shingles.
Unsurprisingly, they cite no case sanctioning such a
remarkable proposition. Plaintiffs’ citation to In re IKO
Shingle Products Liability Litigation, 

(7th Cir.
2014), a roofing shingle case in which not all shingles were
manifestly defective, is unavailing. In that case, the Multi-
District Litigation Court’s order denying class certification
was vacated because it had incorrectly imposed a
commonality-of-damages requirement at the class
certification stage. 

. The plaintiffs in that
case—unlike Plaintiffs here—had no difficulty articulating a
defect common to all of their shingles, namely, their failure to
meet the manufacturer’s promise that they satisfied the
applicable ASTM standard. 

       Equally unavailing is Plaintiffs’ analogy to McManus
v. Fleetwood Enterprises, Inc., 

(5th Cir. 2003),



                               23
in which certification of a class was upheld on the ground that
class members “did not receive the benefit of their bargain.”

(internal quotation marks and citation omitted). The
plaintiffs in that case were nevertheless required to explain
why the motor homes they had purchased were defective, and
their explanation applied to all class members: none of the
motor homes could “safely tow a normal car without
supplemental breaks.” 

As in all of the other
cases Plaintiffs cite, members of the putative class in
McManus were required to articulate a defect allegedly
present in all relevant products. Regardless of the theory of
liability pursued, Rule 23 requires, if nothing else, that a
putative class must describe the product’s defect on a
classwide basis. If proponents of the class do not allege a
defect common to the class, the defectiveness of a given
product is, by necessity, not susceptible to proof by classwide
evidence.

        Finally, Plaintiffs’ reliance on Tyson Foods, Inc. v.
Bouaphakeo, 

(2016), is misplaced as well. In
that case, a class of employees claimed Tyson Foods violated
the Fair Labor Standards Act when it failed to compensate
them for time spent “donning and doffing” protective 

. Because Tyson had failed to keep records
for the relevant period, the class proposed to use a study
measuring uncompensated overtime using a representative
sample of employees. 

3. The Supreme Court held
that the class could extrapolate from the results of the study to
establish classwide liability, even though the parties agreed
that some class members were properly compensated,
because the study was “the only practicable means to collect
and present relevant data.” 

6. Tyson objected that
uninjured class members might share in or increase the




                               24
amount of the awarded damages, but the Court concluded that
this was an issue the parties could address upon
apportionment of the damages. 

9–50. Despite the
necessary inquiries into individual work times, the Court held
that the question of whether the time employees spent
“donning and doffing” the gear was compensable work under
the FLSA satisfied the predominance requirement. 

5–46. This appeal is unlike Tyson. The FLSA question in
Tyson was a common one because its resolution would
determine the scope of all class members’ rights and leave
only questions of individual damages. Here, by contrast,
resolving the defect issue can be done only by examining
each individual shingle or by accepting a speculative theory
of defect.

       In this case, instead of alleging a defect common to the
class that might be proved by classwide evidence, Plaintiffs
invite us to equate the existence of a defect with the mere
possibility that one might exist. We find no support in
Rule 23 or caselaw for class certification on such a
speculative basis. Far from abusing its discretion, the District
Court properly concluded that Plaintiffs’ novel reformulation
of the concept of a product defect could not be permitted to
work an end run around the requirements of Rule 23(b)(3).

                               2

        Plaintiffs argue that the District Court also abused its
discretion by ruling inadmissible the testimony proffered by
Rutila based on his testing of Oakridge shingles. The Court
ruled that Rutila could not testify because, among other
reasons, the results were tainted by selection bias and
statistically insignificant in light of the millions of Oakridge
shingles installed during the class period. As Rutila admitted,



                              25
the 298 shingles he tested had been returned in connection
with a warranty claim, so they were the antithesis of a random
sample of Oakridge shingles.

       Plaintiffs make several arguments as to why the
Court’s ruling on Rutila’s testimony was erroneous, but we
need not address them, because our Rule 23 analysis is
unaffected by the outcome. Even if all of the testimony
offered by Rutila were admissible, Plaintiffs would not have
been able to cure their inability to identify a meaningful
defect in Oakridge shingles susceptible to classwide evidence.
As the District Court noted, Rutila’s testing did not enable
him to identify when a shingle “crosses the line and becomes
defective.” 

. Nor would it have
negated Rutila’s own admissions that (1) one can identify a
shingle as defective, however that term is defined, only after
examining it individually, and (2) even shingles designed at
the low end of Owens Corning’s specifications will not
necessarily fail before the end of their warranty periods. The
fundamental problem thus remains: Plaintiffs cannot identify
what is defective about their shingles, let alone how the
unidentified defect is susceptible to proof common to all class
members. 5 It is therefore unnecessary for us to undertake a

      5
         Although the class definition’s limitation to shingles
that “manifested any cracking, degranulation, fragmentation,
or deterioration during the warranty coverage period,”

, might seem to solve this
problem, it is irrelevant to the predominance inquiry. The
definition limits who may be a member of the class, but it
does not purport to define the elements of the claims brought
on its behalf. Only the latter are relevant to the predominance
inquiry.




                              26
Daubert analysis to determine whether the District Court
abused its discretion in holding Rutila’s testimony
inadmissible. 6




       6
         After it excluded Rutila’s testing, the District Court
ruled that Rutila would be permitted to testify that mat mass
of approximately 1.5 pounds per 100 square feet would
ensure a non-defective shingle. The Court would not permit
him to testify, however, as to what measurements of asphalt
mass and net weight would ensure a non-defective shingle.
Even if all of these opinions were admissible, however, they
would not help Plaintiffs articulate a meaningful theory of
defect susceptible to classwide evidence, for two reasons.
First, an opinion that one measurement will guarantee
adequate performance does not mean that anything less is
defective. Second, as discussed at the Daubert hearing, “it is
the relationship between mat mass, asphalt mass and net
weight that must be coordinated in order to produce a shingle
that will be reliable.” App. 390. Thus, a shingle with an
inadequate mat mass might not fail before the end of the
applicable warranty period if its other measurements are
sufficient. The fact that this relationship can only be assessed
upon inspection of a particular shingle only compounds the
evidentiary problem for Plaintiffs. At any rate, Plaintiffs only
challenge the District Court’s exclusion of testimony based
on Rutila’s testing of returned shingles.




                              27
                                 3

      Plaintiffs argue that, to the extent the District Court
had reservations about their theory of defect, the Supreme
Court’s decision in Amgen required it to set them aside for
purposes of the class certification decision: “[t]he district
court may not have liked this theory, but class certification
was not the place to adjudicate it.” Plaintiffs Br. 41.

        In Amgen, the Supreme Court cautioned district courts
not to “put[] the cart before the horse” by allowing their
views of the merits to affect their analysis of the independent
question whether a putative class satisfies the requirements of
Rule 

. “[T]he office of a Rule 23(b)(3)
certification ruling is not to adjudicate the case; rather, it is to
select the method best suited to adjudication of the
controversy fairly and efficiently.” 

internal
quotation marks omitted). While in some cases it may be
inevitable that the class-certification analysis “entail some
overlap with the merits of the plaintiff’s underlying claim,”

(quoting 

), as a general
matter, “Rule 23 grants courts no license to engage in free-
ranging merits inquiries at the certification stage.” 

       Plaintiffs call the District Court’s opinion the
“antithesis of Amgen.” Plaintiffs Br. 36. According to
Plaintiffs, the merits of their theory of defect were irrelevant
to class certification, and thus off limits under Amgen. But the
District Court’s analysis, they correctly note, is laced with
merits determinations. For example, the Court characterized
Plaintiffs’ theory of defect as “seemingly novel and illogical,”

, and concluded that Plaintiffs
had “submit[ted] no legal authority to support their
proposition that a design specification that sets a range of



                                28
measurements, some of which will produce defective
products and some of which will not, can establish a design
defect claim,” 

Plaintiffs argue that these merits
determinations “infected the entire certification process,”
requiring a remand. Plaintiffs Br. 2.

        Because “all of the claims in the lawsuit will rise or
fall on the question of defect,” and at least some of the
evidence they propose to use—including portions of Rutila’s
testimony and internal Owens Corning communications—is
applicable to all class members, Plaintiffs claim they have
satisfied their burden under Amgen by identifying a “common
question[] that will yield [a] common answer[].” Reply
Br. 10. (emphasis omitted). To be sure, if defect is defined in
the manner Plaintiffs propose, it does not require a leap to
conclude that the alleged defect is susceptible to classwide
evidence, since class members would not be required to show
anything about their individual shingles. But Plaintiffs offer
no support for the idea that it is a violation of Amgen to
recognize at the certification stage that the “question of
defect” they propose is only superficially a “common
question,” just as any question becomes universal when it
includes the word “all.” Nothing about Amgen or Rule 23
limits district courts in this way. As the Amgen Court
recognized, merits determinations are permitted “to the
extent . . . relevant to determining whether the Rule 23
prerequisites for class certification are 

. As with the other requirements of Rule 23, the
predominance requirement of Rule 23(b)(3) is “not [a] mere
pleading rule[.]” In re Hydrogen 

.
To determine whether the requirement is met, a court may
“delve beyond the pleadings” and answer questions that are
often “enmeshed in the factual and legal issues comprising




                               29
the plaintiff’s cause of action.” 

(internal quotation marks and citations omitted). Courts “must
resolve all factual or legal disputes relevant to class
certification, even if they overlap with the merits—including
disputes touching on elements of the cause of action.” In re
Hydrogen 

.

        In this context, the District Court was not bound—by
Amgen or otherwise—to deem sufficient for certification the
“question of defect” as Plaintiffs presented it. It is
incongruous to argue simultaneously, as Plaintiffs do, that
their theory of defect is central to their claims and that its
coherence and legal foundation are irrelevant to the
certification decision. By rejecting Plaintiffs’ theory, the
District Court did not find as a matter of fact that Oakridge
shingles are not defective. It merely applied Rule 23’s
predominance requirement, under which Plaintiffs were
required at the certification stage to identify a defect that
affected all class members’ shingles, regardless of whether
the defect will ultimately be proved by a preponderance of the
evidence at the merits stage. It does not follow from Amgen
that a common question suffices for purposes of Rule 23 by
virtue of Plaintiffs’ ipse dixit. Instead, it remains the task of
district courts, through application of the rule’s requirements
to the facts and claims before it, to determine what constitutes
a “question[] of law or fact common to class members.” Fed.
R. Civ. P. 23(b)(3). The District Court’s determination in this
regard, while fatal to certification, was nonetheless squarely
within the discretion afforded it by Amgen and Rule 23. What
Plaintiffs attack as the District Court’s “obsession with the
merits,” Plaintiffs Br. 36, we conclude was instead nothing
more than the “rigorous analysis” required to determine




                               30
whether the proposed class actually met the requirements of
Rule 23. 

. 7

       Our conclusion that the District Court did not abuse its
discretion in rejecting Plaintiffs’ theory of defect for purposes
of Rule 23 ends our analysis of whether the Four-State Class



       7
          Plaintiffs quote several other passages in the District
Court’s opinion in which the Court discusses, or appears to
discuss, the merits of Plaintiffs’ defect claim, allegedly in
violation of Amgen. These passages do not show what
Plaintiffs suggest when read in context. For example,
Plaintiffs call “especially revealing” the District Court’s
observation that because it had ruled Rutila’s testing
inadmissible, “[P]laintiffs were tasked with proving their
design defect claim by way of [other] evidence,” 

. Plaintiffs Br. 35. The Court was not
suggesting that Plaintiffs were required to prove the merits of
their defect claim at the class certification stage, though that
might be inferred by Plaintiffs’ incomplete quotation, which
stops after the word “claim.” Rather, the Court was simply,
and correctly, noting that it was Plaintiffs’ burden under Rule
23 to show that the defect claim was susceptible to proof by
classwide evidence despite the inadmissibility of this portion
of Rutila’s testimony. The other passages Plaintiffs cite turn
out to be similarly innocuous. To the extent Plaintiffs argue
that the District Court’s Rule 23 analysis violated Amgen in
its treatment of their misrepresentation-based claims, we need
not address this issue, because, as explained, the inadequacy
of Plaintiffs’ theory of defect for purposes of Rule 23(b)(3) is
alone fatal to class certification.




                               31
should have been certified under Rule 23(b)(3). 8 As we have
explained, because Plaintiffs cannot show that liability for
defective design is susceptible to classwide evidence, they are
equally unable to show the same in support of their
misrepresentation claims. They are thus unable to show that
“[common] questions of law or fact . . . predominate over any
questions affecting only individual members,” as
Rule 23(b)(3) requires. Our conclusion that predominance is
lacking makes it unnecessary for us to discuss whether
Plaintiffs have satisfied the other requirements of Rule 23.
The District Court did not abuse its discretion in denying
certification of the Four-State Class under Rule 23(b)(3).




      8
         Despite its conclusion that Plaintiffs’ “inability to
prove the existence of a design defect by evidence that is
predominantly common to the class is fatal to [their]
misrepresentation-based legal claims,” the District Court
undertook an analysis of the latter “for the sake of
completeness.” 

. For reasons
already explained, we agree with this conclusion.




                              32
                                4

        Plaintiffs argue in the alternative that the District Court
abused its discretion in denying certification of the Four-State
Class as a liability-only issue class under Rule 23(c)(4),
which provides that “[w]hen appropriate, an action may be
brought or maintained as a class action with respect to
particular issues.” In Gates v. Rohm & Haas Co., 

(3d Cir. 2011), we enumerated several non-exhaustive
factors relevant to assessing whether certification of an issue
class under Rule 23(c)(4) is appropriate. Those factors
include “the efficiencies to be gained by granting partial
certification in light of realistic procedural alternatives” and
“the substantive law underlying the claim(s).” 

       Plaintiffs sought to pursue their claims as an issue
class under Rule 23(c)(4) because the issues involved rest on
a “common core of facts,” so efficiencies would be gained by
resolving them in a single proceeding. 

. The District Court disagreed, in part for the same
reasons it concluded that Plaintiffs could not satisfy the
predominance requirement of Rule 23(b)(3): because the
issues are not susceptible to “proof by common
evidence[,] . . . [n]o efficiencies are gained by litigating
[them] on a classwide basis.” 

decision to exercise its discretion under
Rule 23(c)(4), like any other certification determination under
Rule 23, must be supported by rigorous analysis.” Hohider v.
United Parcel Serv., Inc., 

, 200–01 (3d Cir.
2009). While Plaintiffs are correct to point out that the
appropriateness of certifying a Rule 23(c)(4) class is
analytically independent from the predominance inquiry
under Rule 23(b)(3), a case may present concerns relevant to



                                33
both. See 

–74. This is such a case.
Unlike a situation in which a Rule 23(c)(4) class might be
appropriate because liability is capable of classwide treatment
but damages are not, Plaintiffs offer no theories of liability for
which classwide treatment is apt. The District Court therefore
correctly determined that a Rule 23(c)(4) class would not
“materially advance resolution of the underlying claims,” 

any more than would a class under Rule 23(b)(3).
Accordingly, we cannot conclude that the District Court’s
denial of certification of the Four-State Class under
Rule 23(c)(4) was an abuse of discretion.

                               IV

        For the reasons stated, the District Court did not abuse
its discretion in denying Plaintiffs’ motion for class
certification. We will affirm.




                               34