FOR PUBLICATION

  UNITED STATES COURT OF APPEALS
       FOR THE NINTH CIRCUIT


SMITHKLINE BEECHAM                   No. 11-17357
CORPORATION, DBA
GlaxoSmithKline,                       D.C. No.
          Plaintiff-Appellee,     4:07-cv-05702-CW

             v.

ABBOTT LABORATORIES,
       Defendant-Appellant.



SMITHKLINE BEECHAM                   No. 11-17373
CORPORATION, DBA
GlaxoSmithKline,                       D.C. No.
         Plaintiff-Appellant,     4:07-cv-05702-CW

             v.
                                       OPINION
ABBOTT LABORATORIES,
        Defendant-Appellee.


     Appeal from the United States District Court
        for the Northern District of California
    Claudia Wilken, Chief District Judge, Presiding

               Argued and Submitted
    September 18, 2013—San Francisco, California
2    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

                      Filed January 21, 2014

       Before: Mary M. Schroeder, Stephen Reinhardt,
           and Marsha S. Berzon, Circuit Judges.

                   Opinion by Judge Reinhardt


                           SUMMARY*


                          Jury Selection

    Reversing the district court’s judgment in an antitrust case
concerning a licensing agreement and the pricing of HIV
medications, the panel held that classifications based on
sexual orientation are subject to heightened scrutiny, and in
jury selection, equal protection prohibits peremptory strikes
based on sexual orientation.

    The panel held that even though the Ninth Circuit had in
the past applied rational basis review, United States v.
Windsor, 

(2013) (holding Defense of
Marriage Act unconstitutional), required that heightened
scrutiny be applied to equal protection claims involving
sexual orientation. The panel held that in light of the history
of exclusion of gays and lesbians from democratic institutions
and the pervasiveness of stereotypes about the group, the
protection of Batson v. Kentucky, 

(1986), applies,
and equal protection forbids striking a juror on the basis of


  *
    This summary constitutes no part of the opinion of the court. It has
been prepared by court staff for the convenience of the reader.
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES               3

his sexual orientation. The panel remanded the case for a
new trial.


                        COUNSEL

Daniel B. Levin (argued), Jeffrey I. Weinberger, Stuart N.
Senator, Keith R.D. Hamilton, Kathryn A. Eidmann, Munger,
Tolles, & Olson LLP, Los Angeles, California; Krista Enns,
Winston & Strawn LLP, San Francisco, California; James F.
Hurst, Samuel S. Park, Winston & Strawn LLP, Chicago,
Illinois; Charles B. Klein, Steffen N. Johnson, Matthew A.
Campbell, Jacob R. Loshin, Winston & Strawn LLP,
Washington, D.C., for Defendant-Appellant/Cross-Appellee.

Lisa S. Blatt (argued), Sarah M. Harris, Arnold & Porter LLP,
Washington, D.C.; Brian J. Hennigan (argued), Alexander F.
Wiles, Carlos R. Moreno, Trevor V. Stockinger, Lillie A.
Werner, Christopher Beatty, Andrew Ow, Irell & Manella
LLP, Los Angeles, California; for Plaintiff-Appellee/Cross-
Appellant.

Shelbi D. Day, Tara L. Borelli, Jon W. Davidson, Lambda
Legal Defense and Education Fund, Inc., Los Angeles,
California, for Amicus Curiae.


                         OPINION

REINHARDT, Circuit Judge:

   The central question in this appeal arises out of a lawsuit
brought by SmithKline Beecham (GSK) against Abbott
Laboratories (Abbott) that contains antitrust, contract, and
4   SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

unfair trade practice (UTPA) claims. The dispute relates to
a licensing agreement and the pricing of HIV medications, the
latter being a subject of considerable controversy in the gay
community. GSK’s claims center on the contention that
Abbott violated the implied covenant of good faith and fair
dealing, the antitrust laws, and North Carolina’s Unfair Trade
Practices Act by first licensing to GSK the authority to
market an Abbott HIV drug in conjunction with one of its
own and then increasing the price of the Abbott drug
fourfold, so as to drive business to Abbott’s own,
combination drug.

    During jury selection, Abbott used its first peremptory
strike against the only self-identified gay member of the
venire. GSK challenged the strike under Batson v. Kentucky,

(1986), arguing that it was impermissibly made
on the basis of sexual orientation. The district judge denied
the challenge.

    This appeal’s central question is whether equal protection
prohibits discrimination based on sexual orientation in jury
selection. We must first decide whether classifications based
on sexual orientation are subject to a standard higher than
rational basis review. We hold that such classifications are
subject to heightened scrutiny. We also hold that equal
protection prohibits peremptory strikes based on sexual
orientation and remand for a new trial.

                               I.

   During jury selection, the district judge began by asking
questions of the potential jurors based on their questionnaires,
and then each party’s counsel had an opportunity to ask
additional questions. When the judge turned her attention to
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES                 5

Juror B, a male, she inquired first about his employment, as
she had done with each of the previous members of the
venire. Juror B stated that he worked as a computer
technician for the Ninth Circuit Court of Appeals in San
Francisco. During the course of the judge’s colloquy with
Juror B, the juror revealed that his “partner” studied
economics and investments. When the district judge
followed up with additional questions, the prospective juror
referred to his partner three times by using the masculine
pronoun, “he,” and the judge subsequently referred to Juror
B’s partner as “he” in a follow-up question regarding his
employment status. Responding to additional questions from
the judge, Juror B stated that he took an Abbott or a GSK
medication and that he had friends with HIV. When the time
arrived for Abbott’s counsel, Weinberger, to question Juror
B, the questioning was brief and limited. Counsel’s first
question concerned Juror B’s knowledge of the medications
that were the focal point of the litigation: “You indicated that
you know some people who have been diagnosed with
HIV. . . . Do you know anything about the medications that
any of them are on?” Juror B responded, “Not really.”
Abbott’s counsel then continued: “Do you know whether any
of them are taking any of the medications that we are going
to be talking about here[,] . . . Norvir or Kaletra or Lexiva,
any of those?” Juror B responded that he did not know
whether his friends took those medications, but that he had
heard of Kaletra. He added that he didn’t know much about
the drug and that he had no personal experiences with it. In
sum, Abbott’s counsel asked Juror B five questions, all
regarding his knowledge of the drugs at issue in the litigation.
Abbott’s counsel did not ask Juror B when he had taken
either an Abbott or GSK medication, how long ago, which
medication it was, or the purpose of the medication. He also
6   SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

failed to ask any questions as to whether Juror B could decide
the case fairly and impartially.

    When the time came for peremptory challenges, Abbott
exercised its first strike against Juror B. GSK’s counsel,
Saveri, immediately raised a Batson challenge, and the
following discussion ensued:

       Mr. Saveri: Okay. So, you know, the first
       challenge, your honor, is a peremptory
       challenge of someone who is — who I think is
       or appears to be, could be homosexual. That’s
       use of the peremptory challenge in a
       discriminatory way.

       The problem here, of course, your honor, is
       the litigation involves AIDS medication. The
       incidents [sic] of AIDS in the homosexual
       community is well-known, particularly gay
       men.

       So with that challenge, Abbott wants to
       exclude from — it looks like Abbott wants to
       exclude from the pool anybody who is gay.
       So I am concerned about that. I wanted to
       raise it.

       The Court: Well, I don’t know that, number
       one, whether Batson applies in civil, and
       number two, whether Batson ever applies to
       sexual orientation. Number three, how we
       would know — I mean, the evil of Batson is
       not that one person of a given group is
       excluded, but that everyone is. And there is
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES              7

       no way for us to know who is gay and who
       isn’t here, unless somebody happens to say
       something.

       There would be no real way to analyze it.
       And number four, one turns to the other side
       and asks for the basis for their challenge other
       than the category that they are in, and if you
       have one, it might be the better part of valor to
       tell us what it is.

       Mr. Weinberger: Well, he —

       The Court: Or if you don’t want to, you can
       stand on my first three reasons.

       Mr. Weinberger: I will stand on the first
       three, at this point, your honor. I don’t think
       any of the challenge applies. I have no idea
       whether he is gay or not.

       Mr. Saveri: Your honor, in fact, he said on
       voir dire that he had a male partner. So —

       Mr. Weinberger: This is my first challenge.
       It’s not like we are sitting here after three
       challenges and you can make a case that we
       are excluding anybody.

The district judge then stated that she would allow Abbott’s
strike and would reconsider her ruling if Abbott struck other
gay men.
8   SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

    At the conclusion of the four-week trial, the jury returned
with a mixed verdict. It held for Abbott on the antitrust and
UTPA claims, and for GSK on the contract claim. It awarded
$3,486,240 in damages to GSK.

    Abbott appealed the jury verdict on the contract claim,
and GSK cross-appealed. On cross-appeal, GSK contends
that a new trial is warranted on all counts, including the
contract claim, because Abbott unconstitutionally used a
peremptory strike to exclude a juror on the basis of his sexual
orientation. We hold that the exclusion of the juror because
of his sexual orientation violated Batson and we remand for
a new trial.

                              II.

    The Batson analysis involves a three-part inquiry. First,
the party challenging the peremptory strike must establish a
prima facie case of intentional discrimination. Kesser v.
Cambra, 

, 359 (9th Cir. 2006). Second, the
striking party must give a nondiscriminatory reason for the
strike. See 

court determines, on the basis of
the record, whether the party raising the challenge has shown
purposeful discrimination. 

district judge
applied the wrong legal standard in evaluating the Batson
claim, we review the Batson challenge de novo. United
States v. Collins, 

, 919 (9th Cir. 2009).

    To establish a prima facie case under Batson, GSK must
produce evidence that 1) the prospective juror is a member of
a cognizable group; 2) counsel used a peremptory strike
against the individual; and 3) “the totality of the
circumstances raises an inference that the strike was
motivated” by the characteristic in question. Collins,
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES                  

. “[A] defendant satisfies the requirements of
Batson’s first step by producing evidence sufficient to permit
the trial judge to draw an inference that discrimination has
occurred.” Johnson v. California, 

, 170 (2005).
The burden on the challenging party at the prima facie stage
is “not an onerous one.” Boyd v. Newland, 

,
1151 (9th Cir. 2004). It is a burden of production, not a
burden of persuasion. Crittenden v. Ayers, 

, 954
(9th Cir. 2010).

     GSK has established a prima facie case of intentional
discrimination. Juror B was the only juror to have identified
himself as gay on the record, and the subject matter of the
litigation presented an issue of consequence to the gay
community. When jury pools contain little racial or ethnic
diversity, we have held that a strike of the lone member of the
minority group is a “relevant consideration” in determining
whether a prima facie case has been established. 

We have further cautioned against failing to “look closely” at
instances in which the sole minority is struck from the venire;
this is because failure to do so would innoculate peremptory
strikes against Batson challenges in jury pools with scant
diversity. 

; see also United States v.
Chinchilla, 

, 698 n.5 (9th Cir. 1989)
(“[A]lthough the striking of one or two members of the same
racial group may not always constitute a prima facie case, it
is preferable for the court to err on the side of the defendant’s
rights to a fair and impartial jury.”).

    There is also reason to infer that Abbott struck Juror B on
the basis of his sexual orientation because of its fear that he
would be influenced by concern in the gay community over
Abbott’s decision to increase the price of its HIV drug. When
we analyzed whether the appellant had made out a prima
10 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

facie case in Johnson v. Campbell, 

(9th Cir.
1996), for instance, we found it significant that the struck
juror’s sexual orientation had no relevance to the subject
matter of the litigation. 

& n.1. The converse is true
as well. In J.E.B. v. Alabama, 

(1994), the
Supreme Court stated that when the gender of the juror
coincided with the subject matter of the case, the potential for
an impermissible strike based on sex increases substantially.

Here, the increase in the price of the HIV drug had
led to considerable discussion in the gay community. Upon
raising the Batson challenge, GSK’s counsel argued that the
subject matter of the litigation raised suspicions regarding the
purpose of the strike: “The problem here . . . is the litigation
involves AIDS medications. The incidents [sic] of AIDS in
the homosexual community is well-known, particularly gay
men.” The potential for relying on impermissible stereotypes
in the process of selecting jurors was “particularly acute” in
this case. Id.; see also Powers v. Ohio, 

, 416
(1991).1 Viewing the totality of the circumstances, we have
no difficulty in concluding that GSK has raised an inference
of discrimination and established a prima facie case.

  1
    In evaluating an ineffective assistance of counsel claim for failure to
raise a Wheeler claim, the California analog of a Batson claim, we stated
that asking Hispanic-surnamed venire members whether they would be
biased in evaluating a case involving a Hispanic defendant did not pose
any constitutional problem because “asking questions about potential bias
is the purpose of voir dire.” Carrera v. Ayers, 

, 1111 (9th
Cir. 2012) (en banc). Carrera suggests that if Abbott’s counsel was
concerned that gay members of the jury pool might be biased because the
price increase had gained some notoriety in the gay community, he could
have questioned Juror B about this potential bias. Instead of pursuing this
line of questioning about Juror B’s ability to assess the case fairly,
Abbott’s counsel struck him without any indication that he was biased,
thereby raising the inference that he had relied on an impermissible
assumption about Juror B’s ability to be impartial.
      SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 11

    Also, Abbott declined to provide any justification for its
strike when offered the opportunity to do so by the district
court. After the judge stated that she might reject the Batson
challenge on legal grounds that were in fact erroneous,2 she
told Abbott’s counsel that he could adopt those grounds,
although she advised him that “it might be the better part of
valor” to reveal the basis for his strike. Abbott’s counsel
replied that he would rely on the grounds given by the judge
and further explained, “I don’t think any of the challenge
applies. I have no idea whether he is gay or not.” He later
added that he could not have engaged in intentional
discrimination because this was only his first strike.

    Counsel’s statement that he did not know that Juror B was
gay is neither consistent with the record nor an explanation
for his strike. First, Juror B and the judge referred to Juror
B’s male partner several times during the course of voir dire
and repeatedly used masculine pronouns when referring to
him. Given the information regarding Juror B’s sexual
orientation that was adduced during the course of voir dire,
counsel’s statement was far from credible. See 

–83 (comparing counsel’s proffered reasons

  2
    The district judge offered her view that Batson did not apply in civil
cases or when only a single member of a protected group is struck. The
first statement — that Batson does not apply to civil cases — is clearly
incorrect. The Supreme Court held over twenty years ago that Batson
applies in the civil context. See Edmonson v. Leesville Concrete Co.,

, 631 (1991). Her statement that Batson does not apply when
only a single member of the given group is excluded is also a legal error
because “[t]he [C]onstitution forbids striking even a single prospective
juror for a discriminatory purpose.” United States v. Vasquez-Lopez,

, 902 (9th Cir. 1994); see also Snyder v. Louisiana, 

, 474 (2008) (citing and quoting Vasquez-Lopez). Her final statement
expressing uncertainty about whether Batson applies to sexual orientation
is the subject of this appeal.
12 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

with the plausible facts on the record). Second, the false
statement was non-responsive; it was simply a denial of a
discriminatory intent and it in no way provided a reason,
colorable or otherwise, for striking Juror B. Counsel’s denial
of a discriminatory motive had the opposite effect of that
intended. Because the denial was demonstrably untrue, it
undermines counsel’s argument that his challenge was not
based on intentional discrimination. Taking all these factors
together, including the absence of any proffered reason for
the challenge, a strong inference arises that counsel engaged
in intentional discrimination when he exercised the strike.3
Paulino v. Harrison (Paulino II), 

, 702–03 (9th
Cir. 2008); see also 

n.6 (“In the
unlikely hypothetical in which [counsel] declines to respond
to a trial judge’s inquiry regarding his justification for making
a strike, the evidence before the judge would consist not only
of the original facts from which the prima facie case was
established, but also [counsel’s] refusal to justify his strike in
light of the court’s request.”).

    Abbott’s counsel asked Juror B only five questions and
failed to question him meaningfully about his impartiality or
potential biases. See 

. Combined
with Abbott’s counsel’s statement, in the face of clear
evidence in the record to the contrary, that he did not know
that Juror B was gay, the voir dire reveals that Abbott’s strike
was based not on a concern for Juror B’s actual bias, but on
a discriminatory assumption that Juror B could not
impartially evaluate the case because of his sexual
orientation. See 

–62.


 3
   Abbott’s adoption of the court’s erroneous legal reasons why Batson
might be inapplicable to the type of trial before her does not, of course,
provide or even suggest any explanation as to why counsel struck Juror B.
     SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 13

     Finally, Abbott attempts to offer several neutral reasons
for the strike in its brief on appeal to our Court, but these
reasons are also belied by the record. See 

(“[I]f a
review of the record undermines . . . many of the proffered
reasons, the reasons may be deemed a pretext for racial
discrimination.”). Ordinarily, it does not matter what reasons
the striking party might have offered because “[w]hat matters
is the real reason [the juror was] stricken,” Paulino v. Castro
(Paulino I), 

, 1090 (9th Cir. 2004) (emphasis
in original): that is, the reason offered at the time of the
strike, if true. Here, Abbott offered no reasons for the strike
at the voir dire, but we know from the reasons offered on
appeal after full deliberation by highly respected and able
counsel that even the best explanations that counsel could
have offered are pretextual.4 See 

.



 4
    One reason advanced by Abbott on appeal is that Juror B was the only
juror who had lost friends to AIDS. We reject this reason because it is not
supported by the record. Nowhere does the record show that Juror B had
friends who died of complications due to HIV or AIDS.

    A second reason advanced by Abbott on appeal is that Juror B was
acquainted with many people in the legal field. Other jurors, however,
who were lawyers, and other jurors with close relatives who were lawyers
were not stricken but served on the jury.

    Third, Abbott speculates on appeal that because Juror B was a
computer technician at the Court, other jurors “might have given extra
weight” to his opinions. We have more respect for jurors than to credit the
idea that Juror B would have more influence on his fellow jurors than
would the other jurors, including the two lawyers who remained on the
panel. This is the kind of “highly speculative” rationale that the Supreme
Court rejected in 

.

   Finally, Abbott points out that Juror B was the only potential juror
who testified that he had heard of any of the three drugs at issue. When
14 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

    The record reflects that had the district judge applied the
law correctly, she would necessarily have concluded that
Abbott’s strike of Juror B was impermissibly made on the
basis of his sexual orientation. See United States v. Alanis,

, 969 (9th Cir. 2003). Because GSK has
established a prima facie case, Abbott offered no
nondiscriminatory reason for its strike of Juror B at trial, and
Abbott does not now offer in its brief on appeal any colorable
neutral explanation for the strike, only one result is possible
here. The prima facie evidence that the strike was based on
a discriminatory motive is unrefuted, and on appeal it is clear
that Abbott has no further credible reasons to advance nor
evidence to offer. Accordingly, we need not remand the
question whether a Batson violation occurred. See 

The record persuasively demonstrates that Juror B
was struck because of his sexual orientation. This Court may
therefore perform the third step of the Batson analysis and
conclude “even based on a ‘cold record,’ that [Abbott’s]
stated reasons for striking [Juror B] was a pretext for
purposeful discrimination.” 

n.5.



asked what he knew about the drug, however, Juror B replied, “not much,”
and stated that he had no personal experience with it.

     Here, three of the four reasons offered by Abbott are pretextual and
the record casts strong doubt on the fourth. In such a circumstance, we
follow the rule of our en banc decision in Kesser, and conclude that none
of those reasons can withstand judicial scrutiny. See 

(“A court
need not find all nonracial reasons pretextual in order to find racial
discrimination.”); see also 

court is left with only two
acceptable bases for the challenges. . . . Although these criteria would
normally be adequately ‘neutral’ explanations taken at face value, the fact
that two of the four proffered reasons do not hold up under judicial
scrutiny militates against their sufficiency.’” (quoting 

)).
     SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 15

                                   III.

    We must now decide the fundamental legal question
before us: whether Batson prohibits strikes based on sexual
orientation.5 In Batson, the Supreme Court held that the
privilege of peremptory strikes in selecting a jury is subject
to the guarantees of the Equal Protection 

. Batson, of course, considered peremptory strikes based
on race. At stake, the Court explained, were not only the
rights of the criminal defendant, but also of the individual
who is excluded from participating in jury service on the
basis of his race. 

Allowing peremptory strikes
based on race would “touch the entire community” because
it would “undermine public confidence in the fairness of our
system of justice.” 

Court held, the exclusion of
prospective jurors because of their race would require
reversal upon a finding of intentional discrimination. 

Eight years later, in J.E.B., the Court extended Batson
to peremptory strikes made on the basis of gender. While
expanding Batson’s ambit, J.E.B. explained the scope of its


 5
   Citing Johnson v. Campbell, Abbott urges us to avoid deciding whether
Batson applies to sexual orientation by holding that a prima facie showing
cannot be demonstrated because “‘an obvious neutral reason for the
challenge’ appears in the record.” As we have explained, there are no
“obvious neutral” reasons for Abbott’s strike in the record or even in
Abbott’s brief on appeal. In Campbell, we rejected a Batson challenge
based on sexual-orientation where (1) counsel “made no attempt to show
discriminatory motivation on the part of the opposing attorney,” (2) there
was no showing that opposing counsel was aware of the juror’s sexual
orientation, (3) there was an obvious neutral reason for the strike, and (4)
the juror’s sexual orientation had no bearing on the subject matter of the
case. 

. All of the factors that were absent in
Campbell are present here. Because the record shows that there was
purposeful discrimination here, the path we took in Campbell is not
available to us.
16 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

expansion. The Court stated that “[p]arties may . . . exercise
their peremptory challenges to remove from the venire any
group or class of individuals normally subject to ‘rational
basis’ 

; accord United States v.
Santiago-Martinez, 

, 423 (9th Cir. 1995). Thus,
if sexual orientation is subject to rational basis review,
Abbott’s strike does not require reversal.

    We have in the past applied rational basis review to
classifications based on sexual orientation. In High Tech
Gays v. Defense Industrial Security Clearance Office,

, 574 (9th Cir. 1990), and Philips v. Perry,

, 1425 (9th Cir. 1997), we applied rational
basis review when upholding Department of Defense and
military policies that classified individuals on the basis of
sexual orientation. More recently, in Witt v. Department of
the Air Force, 

(9th Cir. 2008), an Air Force
reservist brought due process and equal protection challenges
to her suspension from duty on account of her sexual
relationship with a woman. 

We considered the
meaning of the Supreme Court’s decision in Lawrence v.
Texas, 

(2003), and concluded that because
Lawrence relied only on substantive due process and not on
equal protection, it affected our prior substantive due process
cases, but not our equal protection rules. 

. As a result, although we applied heightened scrutiny to
the substantive due process challenge in Witt, we did not
change our level of scrutiny for the equal protection
challenge. 

that Lawrence “declined to address
equal protection,” and relying on Philips, our pre-Lawrence
decision, we continued to apply rational basis review to equal
protection challenges. 

Thus, we are bound here to
apply rational basis review to the equal protection claim in
the absence of a post-Witt change in the law by the Supreme
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 17

Court or an en banc court. See Miller v. Gammie, 

, 892–93 (9th Cir. 2003) (en banc). Here, we turn to the
Supreme Court’s most recent case on the relationship
between equal protection and classifications based on sexual
orientation: United States v. Windsor, 

(2013).
That landmark case was decided just last term and is
dispositive of the question of the appropriate level of scrutiny
in this case.

    Windsor, of course, did not expressly announce the level
of scrutiny it applied to the equal protection claim at issue in
that case, but an express declaration is not necessary.
Lawrence presented us with a nearly identical quandary when
we confronted the due process claim in Witt. Just as
Lawrence omitted any explicit declaration of its level of
scrutiny with respect to due process claims regarding sexual
orientation, so does Windsor fail to declare what level of
scrutiny it applies with respect to such equal protection
claims. Nevertheless, we have been told how to resolve the
question. 

. When the Supreme Court
has refrained from identifying its method of analysis, we have
analyzed the Supreme Court precedent “by considering what
the Court actually did, rather than by dissecting isolated
pieces of text.” 

we looked to three factors in determining that
Lawrence applied a heightened level of scrutiny rather than
a rational basis analysis. We stated that Lawrence did not
consider the possible post-hoc rationalizations for the law,
required under rational basis review. 

.
We further explained that Lawrence required a “legitimate
state interest” to “justify” the harm that the Texas law
inflicted as is traditionally the case in heightened scrutiny.

(quoting 

)
18 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

(internal quotation marks omitted). Finally, we looked to the
cases on which Lawrence relied and found that those cases
applied heightened scrutiny. 

. Applying
the Witt test here, we conclude that Windsor compels the
same result with respect to equal protection that Lawrence
compelled with respect to substantive due process: Windsor
review is not rational basis review. In its words and its deed,
Windsor established a level of scrutiny for classifications
based on sexual orientation that is unquestionably higher than
rational basis review. In other words, Windsor requires that
heightened scrutiny be applied to equal protection claims
involving sexual orientation.

     Examining Witt’s first factor, Windsor, like Lawrence, did
not consider the possible rational bases for the law in question
as is required for rational basis review. The Supreme Court
has long held that a law must be upheld under rational basis
review “if any state of facts reasonably may be conceived to
justify” the classifications imposed by the law. McGowan v.
Maryland, 

, 426 (1961). This lowest level of
review does not look to the actual purposes of the law.
Instead, it considers whether there is some conceivable
rational purpose that Congress could have had in mind when
it enacted the law.

    This rule has been repeated throughout the history of
modern constitutional law. In Williamson v. Lee Optical,

(1955), the Court repeatedly looked to what the
legislature “might have concluded” in enacting the law in
question and evaluated these hypothetical reasons. 

In United States Railroad Retirement Board v. Fritz, 

(1980), the Court emphasized that deference to post-hoc
explanations was central to rational basis review:
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 19

       Where, as here, there are plausible reasons for
       Congress’ action, our inquiry is at an end. It
       is, of course, “constitutionally irrelevant
       whether this reasoning in fact underlay the
       legislative decision,”. . . because this Court
       has never insisted that a legislative body
       articulate its reasons for enacting a statute.
       This is particularly true where the legislature
       must necessarily engage in a process of
       line-drawing.       The “task of classifying
       persons for . . . benefits . . . inevitably requires
       that some persons who have an almost equally
       strong claim to favored treatment be placed on
       different sides of the line,” . . . and the fact the
       line might have been drawn differently at
       some points is a matter for legislative, rather
       than judicial, consideration.

(internal citations omitted). More recently, the
Supreme Court has again stated that under rational basis
review, “it is entirely irrelevant for constitutional purposes
whether the conceived reason for the challenged distinction
actually motivated the legislature.” Fed. Commc’n Comm’n
v. Beach Commc’n, Inc., 

, 315 (1993).

    In Windsor, instead of conceiving of hypothetical
justifications for the law, the Court evaluated the “essence”
of the law. 

. Windsor looked to
DOMA’s “design, purpose, and effect.” 

This
inquiry included a review of the legislative history of DOMA.
Windsor quoted extensively from the House Report and
restated the House’s conclusion that marriage should be
protected from the immorality of homosexuality. 

Unlike in rational basis review, hypothetical reasons for
20 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

DOMA’s enactment were not a basis of the Court’s inquiry.
In its brief to the Supreme Court, the Bipartisan Legal
Advisory Group offered five distinct rational bases for the
law. See Brief on the Merits for Respondent the Bipartisan
Legal Advisory Group of the U.S. House of Representatives
at 28–48, Windsor, 

(2013) (No. 12-307),

. Windsor, however, looked behind these
justifications to consider Congress’s “avowed purpose:” “The
principal purpose,” it declared, “is to impose inequality, not
for other reasons like governmental efficiency.” 

, 2694. The result of this more fundamental
inquiry was the Supreme Court’s conclusion that DOMA’s
“demonstrated purpose” “raise[d] a most serious question
under the Constitution’s Fifth Amendment.” 

(emphasis added). Windsor thus requires not that we
conceive of hypothetical purposes, but that we scrutinize
Congress’s actual purposes.              Windsor’s “careful
consideration” of DOMA’s actual purpose and its failure to
consider other unsupported bases is antithetical to the very
concept of rational basis review. 

    Witt’s next factor also requires that we conclude that
Windsor applied heightened scrutiny. Just as Lawrence
required that a legitimate state interest justify the harm
imposed by the Texas law, the critical part of Windsor begins
by demanding that Congress’s purpose “justify disparate
treatment of the group.” 

(emphasis added). Windsor requires a “legitimate purpose”
to “overcome[]” the “disability” on a “class” of individuals.

As we explained in Witt, “[w]ere the Court
applying rational basis review, it would not identify a
legitimate state interest to ‘justify’ . . . .” the disparate
treatment of the group. 

.
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 21

    Rational basis is ordinarily unconcerned with the
inequality that results from the challenged state action. See

–26 (applying the presumption
that state legislatures “have acted within their constitutional
power despite the fact that, in practice, their laws result in
some inequality”). Due to this distinctive feature of rational
basis review, words like harm or injury rarely appear in the
Court’s decisions applying rational basis review. Windsor,
however, uses these words repeatedly. The majority opinion
considers DOMA’s “effect” on eight separate occasions.
Windsor concerns the “resulting injury and indignity” and the
“disadvantage” inflicted on gays and 

, 2693.

    Moreover, Windsor refuses to tolerate the imposition of
a second-class status on gays and lesbians. Section 3 of
DOMA violates the equal protection component of the due
process clause because “it tells those couples, and all the
world, that their otherwise valid marriages are unworthy of
federal recognition.” 

Windsor was thus
concerned with the public message sent by DOMA about the
status occupied by gays and lesbians in our society. This
government-sponsored message was in itself a harm of great
constitutional significance: “Under DOMA, same-sex married
couples have their lives burdened, by reason of government
decree, in visible and public ways.” 

with DOMA’s message follows our constitutional tradition in
forbidding state action from “denoting the inferiority” of a
class of people. Brown v. Bd. of Educ., 

, 494
(1954) (internal quotations omitted) (citation omitted). It is
the identification of such a class by the law for a separate and
lesser public status that “make[s] them unequal.” 

. DOMA was “practically a brand upon
them, affixed by the law, an assertion of their inferiority.”
22 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

Strauder v. West Virginia, 

, 308 (1879).
Windsor requires that classifications based on sexual
orientation that impose inequality on gays and lesbians and
send a message of second-class status be justified by some
legitimate purpose.

     Notably absent from Windsor’s review of DOMA are the
“strong presumption” in favor of the constitutionality of laws
and the “extremely deferential” posture toward government
action that are the marks of rational basis review. Erwin
Chemerinsky, Constitutional Law 695 (4th ed. 2013). After
all, under rational basis review, “it is for the legislature, not
the courts, to balance the advantages and disadvantages of the
new requirement.” Lee 

. Windsor’s
failure to afford this presumption of validity, however, is
unmistakable. In its parting sentences, Windsor explicitly
announces its balancing of the government’s interest against
the harm or injury to gays and lesbians: “The federal statute
is invalid, for no legitimate purpose overcomes the purpose
and effect to disparage and injure those whom the State, by its
marriage laws, sought to protect in personhood and 

(emphasis added). Windsor’s balancing is
not the work of rational basis review.

    In analyzing its final and least important factor, Witt
stated that Lawrence must have applied heightened scrutiny
because it cited and relied on heightened scrutiny cases. 

. Part IV, the central portion of Windsor’s
reasoning, cites few cases, instead scrutinizing Congress’s
actual purposes and examining in detail the inequality
imposed by the law. Among the cases that the Court cites are
Romer v. Evans, 

(1996), Department of
Agriculture v. Moreno, 

(1973), and Lawrence.
In Witt, we thought it noteworthy that Lawrence did not cite
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 23

Romer, a rational basis case. 

. The
citation to Moreno, however, is significant because the Court
recognized in Lawrence that Moreno applied “a more
searching form of rational basis review,” despite purporting
to apply simple rational basis review. 

. Our Court has similarly acknowledged that Moreno
applied “‘heightened’ scrutiny.” See Mountain Water Co. v.
Montana Dep’t of Pub. Serv. Regulation, 

, 599
(9th Cir. 1990). Further, the Court cited Lawrence, which we
have since held applied heightened scrutiny. 

. As we stated in Witt, Lawrence did not resolve
whether to apply heightened scrutiny in equal protection
cases, but, nevertheless, Lawrence is a heightened scrutiny
case. Because Windsor relies on one case applying rational
basis and two cases applying heightened scrutiny, Witt’s final
factor does not decisively support one side or the other but
leans in favor of applying heightened scrutiny.

    At a minimum, applying the Witt factors, Windsor
scrutiny “requires something more than traditional rational
basis review.” 

. Windsor requires that
when state action discriminates on the basis of sexual
orientation, we must examine its actual purposes and
carefully consider the resulting inequality to ensure that our
most fundamental institutions neither send nor reinforce
messages of stigma or second-class status. In short, Windsor
requires heightened scrutiny. Our earlier cases applying
rational basis review to classifications based on sexual
orientation cannot be reconciled with Windsor. See 

–93. Because we are bound by controlling,
higher authority, we now hold that Windsor’s heightened
scrutiny applies to classifications based on sexual orientation.
See 

–93; see also 

–17, 821.
24 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

    In sum, Windsor requires that we reexamine our prior
precedents, and Witt tells us how to interpret Windsor. Under
that analysis, we are required by Windsor to apply heightened
scrutiny to classifications based on sexual orientation for
purposes of equal protection. Lawrence previously reached
that same conclusion for purposes of due process. 

, 821. Thus, there can no longer be any
question that gays and lesbians are no longer a “group or
class of individuals normally subject to ‘rational basis’
review.” 

.

                              IV.

                               A.

     Having established that heightened scrutiny applies to
classifications based on sexual orientation, we must now
determine whether Batson is applicable to that classification
or group of individuals. In J.E.B., the Court did not state
definitively whether heightened scrutiny is sufficient to
warrant Batson’s protection or merely necessary. See 

& n.6, 143. The Court explained that striking
potential jurors on the basis of their gender harms “the
litigants, the community, and the individual jurors” because
it reinforces stereotypes and creates an appearance that the
judicial system condones the exclusion of an entire class of
individuals. 

It added that, when viewed against
the long history of women’s exclusion from jury service,
gender-based strikes send a message “that certain individuals
. . . are presumed unqualified by state actors to decide
important questions upon which reasonable persons could
disagree.” 

With J.E.B.’s concerns in mind and
given that classifications on the basis of sexual orientation are
subject to heightened scrutiny, we must answer whether equal
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 25

protection forbids striking a juror on the basis of his sexual
orientation. We conclude that it does.

    J.E.B. took Batson, a case about the use of race in jury
selection, and applied its principles to discrimination against
women. As the Supreme Court acknowledged, women’s
experiences differed significantly from the experiences of
African Americans. 

–36. The Court
did not require that, to warrant the protections of Batson,
women’s experiences had to be identical to those of African
Americans. 

remained constant in the
Court’s analysis was its willingness to reason from the actual
experiences of the group. For women, a history of exclusion
from jury service and the prevalence of “invidious group
stereotypes” led the Court to conclude that Batson should
extend to strikes on the basis of gender. 

140.
Here also we must reason from the unique circumstances of
gays and lesbians in our society.

    Gays and lesbians have been systematically excluded
from the most important institutions of self-governance.
Even our prior cases that rejected applying heightened
scrutiny to classifications on the basis of sexual orientation
have acknowledged that gay and lesbian individuals have
experienced significant discrimination. See High Tech 

; 

–25 (Canby, J.,
dissenting in part). In the first half of the twentieth century,
public attention was preoccupied with homosexual
“infiltration” of the federal government. Gays and lesbians
were dismissed from civilian employment in the federal
government at a rate of sixty per month. Michael J. Klarman,
From the Closet to the Altar 5 (2013). Discrimination in
employment was not limited to the federal government; local
and state governments also excluded homosexuals, and
26 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

professional licensing boards often revoked licenses on
account of homosexuality. 

the Supreme Court
denied certiorari in a case in which a woman had been fired
from her job as a guidance counselor in a public school
because of her sexuality. Rowland v. Mad River Local Sch.
Dist., 

(1985) (Brennan, J., dissenting from
denial of certiorari). Indeed, gays and lesbians were thought
to be so contrary to our conception of citizenship that they
were made inadmissible under a provision of our immigration
laws that required the Immigration and Naturalization Service
(INS) to exclude individuals “afflicted with psychopathic
personality.” See Boutilier v. INS, 

, 120 (1967).
It was not until 1990 that the INS ceased to interpret that
category as including gays and lesbians. William N.
Eskridge, Gaylaw: Challenging the Apartheid of the Closet
133–34 (1999). It is only recently that gay men and women
gained the right to be open about their sexuality in the course
of their military service. As one scholar put it, throughout the
twentieth century, gays and lesbians were the “anticitizen.”
Margot Canaday, The Straight State 9 (2009).

    Strikes exercised on the basis of sexual orientation
continue this deplorable tradition of treating gays and
lesbians as undeserving of participation in our nation’s most
cherished rites and rituals. They tell the individual who has
been struck, the litigants, other members of the venire, and
the public that our judicial system treats gays and lesbians
differently. They deprive individuals of the opportunity to
participate in perfecting democracy and guarding our ideals
of justice on account of a characteristic that has nothing to do
with their fitness to serve.

    Windsor’s reasoning reinforces the constitutional urgency
of ensuring that individuals are not excluded from our most
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 27

fundamental institutions because of their sexual orientation.
“Responsibilities, as well as rights, enhance the dignity and
integrity of the person.” 

. Jury
service is one of the most important responsibilities of an
American citizen. “[F]or most citizens the honor and
privilege of jury duty is their most significant opportunity to
participate in the democratic process.” 

. It gives gay and lesbian individuals a means of
articulating their values and a voice in resolving controversies
that affect their lives as well as the lives of all others. To
allow peremptory strikes because of assumptions based on
sexual orientation is to revoke this civic responsibility,
demeaning the dignity of the individual and threatening the
impartiality of the judicial system.

    Gays and lesbians may not have been excluded from
juries in the same open manner as women and African
Americans, but our translation of the principles that lie
behind Batson and J.E.B. requires that we apply the same
principles to the unique experiences of gays and lesbians.
Gays and lesbians did not identify themselves as such
because, for most of the history of this country, being openly
gay resulted in significant discrimination. See Kenji Yoshino,
Covering, 111 Yale L.J. 769, 814–36 (2002). The
machineries of discrimination against gay individuals were
such that explicit exclusion of gay individuals was
unnecessary — homosexuality was “unspeakable.” 

In J.E.B., the Court noted that strikes based on gender
were a recent phenomenon because women’s participation on
juries was relatively recent. 

. Being
“out” about one’s sexuality is also a relatively recent
phenomenon. To illustrate how recently the change occurred,
in 1985, only one quarter of Americans reported knowing
someone who was gay. By 2000, this number increased to 75
28 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

percent of Americans. Klarman, From the Closet, at 197. As
we have indicated, gays and lesbians who were “out” were
punished for their openness, sometimes through
imprisonment or exclusion from civil society.

    Batson must also protect potential jurors, litigants, and the
community from the serious dignitary harm of strikes based
on sexual orientation because, as in the case of gender, to
allow such strikes risks perpetuating the very stereotypes that
the law forbids. “It is well known that prejudices often exist
against particular classes in the community, which sway the
judgment of jurors, and which, therefore, operate in some
cases to deny to persons of those classes the full enjoyment
of that protection which others enjoy.” Miller-El v. Dretke
(Miller-El II), 

, 237 (2005) (quoting 

(internal quotation marks omitted)). These
stereotypes and their pernicious effects are not always known
to us. “Prejudice . . . rises not from malice or hostile animus
alone. It may result as well from insensitivity caused by
simple want of careful, rational reflection or from some
instinctive mechanism to guard against people who appear to
be different in some respects from ourselves.” Bd. of
Trustees of Univ. of Alabama v. Garrett, 

, 374
(2001) (Kennedy, J., concurring). Stereotypes of gays and
lesbians depict them as wealthy and promiscuous, and as
“disease vectors” or child molesters.                 Perry v.
Schwarzenegger, 

, 982–83 (N.D. Cal.
2010). Empirical research has begun to show that
discriminatory attitudes toward gays and lesbians persist and
play a significant role in courtroom dynamics. See Jennifer
M. Hill, The Effects of Sexual Orientation in the Courtroom:
A Double Standard, 39:2 J. of Homosexuality 93 (2000).
      SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 29

    As illustrated by this case, permitting a strike based on
sexual orientation would send the false message that gays and
lesbians could not be trusted to reason fairly on issues of
great import to the community or the nation. Strikes based on
preconceived notions of the identities, preferences, and biases
of gays and lesbians reinforce and perpetuate these
stereotypes.6 The Constitution cannot countenance “state-
sponsored group stereotypes rooted in, and reflective of,
historical prejudice.” 

.

    The history of exclusion of gays and lesbians from
democratic institutions and the pervasiveness of stereotypes
about the group leads us to conclude that Batson applies to
peremptory strikes based on sexual orientation.

                                    B.

    Abbott urges us to proceed with caution in light of the
significant sensitivities and privacy interests at stake in
applying Batson to strikes based on sexual orientation. We
agree that, as the California Court of Appeal put it when it
extended Wheeler protection, the state equivalent of Batson,
to gays and lesbians, “No one should be ‘outed’ in order to
take part in the civic enterprise which is jury duty.” People
v. Garcia, 

, 347 (Cal. Ct. App. 2000).
For gays and lesbians, keeping one’s sexual orientation
private has long been a strategy for avoiding the ramifications


  6
    True, attitudes toward gays and lesbians are rapidly changing, just as
attitudes toward women’s role in civic life had changed by the time the
Supreme Court decided J.E.B. in 1994. The central premise of J.E.B.,
however, was that the courtroom should not be a site for “ratify[ing] and
reinforc[ing] prejudicial views,” even if such prejudicial views are on the
decline. 

.
30 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

— job loss, being disowned by friends and family, or even
potential physical danger — that accompanied open
acknowledgment of one’s sexual orientation for most of the
twentieth century and sometimes even today. For some
individuals, being forced to announce their sexuality risks
intruding into the intimate process of self-discovery that is
“coming out,” a process that can be at once affirming and
emotionally fraught. Equally important, coming out for many
gays and lesbians is a life-defining moment of celebrating
one’s dignity and identity. Deciding when, and how, and to
whom one comes out is a vital part of this process, and it
should not be co-opted in the name of affording a group that
has long been discriminated against the constitutional rights
to which it is entitled.

    These concerns merit careful consideration, but they do
not warrant the conclusion that the Constitution necessitates
permitting peremptory strikes based on sexual orientation.
Concerns that applying Batson to sexual orientation will
jeopardize the privacy of gay and lesbian prospective jurors
can be allayed by prudent courtroom procedure. Courts can
and already do employ procedures to protect the privacy of
prospective jurors when they are asked sensitive questions on
any number of topics. Further, applying Batson to strikes
based on sexual orientation creates no requirement that
prospective jurors reveal their sexual orientation. A Batson
challenge would be cognizable only once a prospective
juror’s sexual orientation was established, voluntarily and on
the record. California’s successful application of Wheeler
protections to sexual orientation for the past thirteen years
illustrates that problems with administration can be
overcome, even in a large judicial system that comes in
contact with a diverse population of court users. See 

.
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 31

                              V.

    Abbott contends that any exclusion of a juror in violation
of Batson would have been harmless because none of GSK’s
claims should have been submitted to the jury. It asserts that
there was not sufficient evidence to support any of those
claims.

    We have held that “[t]here is no harmless error analysis
with respect to Batson claims,” Turner v. Marshall, 

, 1254 n.3 (9th Cir. 1997); see also Gray v. Mississippi,

, 668 (1987) (holding that the “right to an
impartial adjudicator, be it judge or jury” is among those
constitutional rights so basic “that their infraction can never
be treated as harmless error”). There are two reasons for this.

    First, it is impossible to determine whether a jury verdict
would have been different had the jury been constitutionally
selected. See Vasquez v. Hillery, 

, 263 (1986)
(“[W]hen a petit jury has been selected upon improper criteria
or has been exposed to prejudicial publicity, we have required
reversal of the conviction because the effect of the violation
cannot be ascertained.”). Second, even if it were possible to
find that a jury verdict had been unaffected by the error, this
would not render the error harmless, as the harm from
excluding a juror in violation of Batson is far greater than
simply the effect upon the verdict.

    In Powers v. Ohio, 

(1991), the Supreme
Court held that a defendant may object to the race-based
exclusion of jurors even if the defendant and the excluded
jurors are not of the same race. 

In so holding, the
Court explained that a Batson violation injures the
unconstitutionally stricken juror as well as the parties: “[a]
32 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

venireperson excluded from jury service because of race
suffers a profound personal humiliation heightened by its
public character.” 

–14. Moreover,
a Batson violation undermines the integrity of the entire trial:

        [The] wrongful exclusion of a juror by a race-
        based peremptory challenge is a constitutional
        violation committed in open court at the
        outset of the proceedings. The overt wrong,
        often apparent to the entire jury panel, casts
        doubt over the obligation of the parties, the
        jury, and indeed the court to adhere to the law
        throughout the trial of the cause. The voir dire
        phase of the trial represents the jurors’ first
        introduction to the substantive factual and
        legal issues in a case. The influence of the
        voir dire process may persist through the
        whole course of the trial proceedings.

(internal quotation omitted). In
Powers, the Court further stated that “discrimination in the
selection of jurors casts doubt on the integrity of the judicial
process” and “may pervade all the proceedings that follow.”

413; see also 

(“Discrimination in jury selection . . . causes harm to the
litigants, the community, and the individual jurors who are
wrongfully excluded from participation in the judicial
process. . . . The community is harmed by the State’s
participation in the perpetuation of invidious group
stereotypes and the inevitable loss of confidence in our
judicial system that state-sanctioned discrimination in the
courtroom engenders.”). Because the effect of excluding a
juror in violation of Batson is so pervasive, it cannot be
    SMITHKLINE BEECHAM V. ABBOTT LABORATORIES 33

deemed harmless, and therefore we do not subject such
violations to harmless error review.

    Abbott urges an exception to this rule, citing an
unpublished disposition, United States v. Gonzalez-Largo,
436 F. App’x 819, 821 (9th Cir. 2011), that relies on Nevius
v. Sumner, 

, 468 (9th Cir. 1988). In Nevius,
which was decided before Powers and J.E.B., we stated that
a Batson violation is harmless where the challenged juror
would have been an alternate who would not have been called
to serve as a juror in any event. 

.
Here, Abbott argues that the Batson error is harmless because
none of the claims should have been allowed to go to the jury
for various reasons, including insufficiency of evidence.
Even were we to accept Abbott’s harmlessness exception, it
would not apply here.

    As agreed by the parties, the contract claim is governed
by New York law. Abbott argues, first, that its conduct did
not violate any implied covenant in its contract with GSK
because that contract contained no agreement as to price.
There was evidence, however, from which a jury could find
that Abbott’s conduct had “injur[ed]” GSK’s right to “receive
the fruits of the contract,” and was meant to have that impact.
Such proof is sufficient under New York law to find a breach
of an implied covenant. See 511 W. 232nd Owners Corp. v
Jennifer Realty Co., 

, 500 (N.Y. 2002).
Abbott’s second argument, that the contract’s limitation-of-
liability clause bars any damages award, is premised on the
“jury[’s] reject[ion of] GSK’s theories involving tortious
gross negligence and intent to harm . . . .” As the jury
findings were tainted by the Batson violation, we cannot rely
34 SMITHKLINE BEECHAM V. ABBOTT LABORATORIES

on them to support enforcement of the limitation-of-liability
clause.7

    In conclusion, the district court properly found that
GSK’s contract claim does not fail as a matter of law.8 Thus,
even if Batson violations were subject to harmless error
analysis where the losing party should have prevailed as a
matter of law and no jury verdict should have been rendered,
the exclusion of a juror in violation of Batson was not
harmless here, as a jury was necessary to resolve the case.
Therefore, we remand for a new trial.9

                                    VI.

    We hold that heightened scrutiny applies to classifications
based on sexual orientation and that Batson applies to strikes
on that basis. Because a Batson violation occurred here, this
case must be remanded for a new trial.

         REVERSED AND REMANDED.


 7
   We have considered and rejected Abbott’s other arguments with regard
to the contract claim.
 8
   Abbott has argued only that structural error does not apply because no
claim should have gone to the jury. As we hold to the contrary with
regard to the implied covenant claim, we need not consider whether the
district court erred in submitting the UTPA and antitrust claims to the jury.
     9
     Our holding that the contract claim does not fail as a matter of law
resolves Abbott’s sole contention on direct appeal, that the district court
should have granted its 50(b) motion for judgment as a matter of law on
this claim. We need not address GSK’s remaining claim on cross-appeal
— that the UTPA verdict was inconsistent with the jury’s findings — as
we remand for a new trial and new findings.