UNITED STATES COURT OF APPEALS

FOR THE NINTH CIRCUIT

PERRY E. COLEMAN; BARBARA
J. COLEMAN, husband and wife,
Plaintiffs-Appellants,
                                                     No. 99-15885

v.
                                                     D.C. No.
                                                     CV-96-0828-PHX-RCB
THE QUAKER OATS COMPANY, a
New Jersey corporation,
Defendant-Appellee.

JERRY JENEY; PEGGY JENEY,
husband and wife,
Plaintiffs-Appellants,
                                                     No. 99-15886

v.
                                                     D.C. No.
                                                     CV-96-0822-PHX-RCB
THE QUAKER OATS COMPANY, a
New Jersey corporation,
Defendant-Appellee.

                               14913


PERRY E. COLEMAN; BARBARA
J. COLEMAN; JOHN RUSSELL;
JERRY JENEY; DENIS
SCHWEITZER, a divorced man;
JOHN TALLARITI; KENNETH
NERO; GARY PEEPLES; J.
THOMAS CHRISTENSON; SHARON
RUSSELL; PEGGY JENEY;
TERRINE TALLARITI; KATHLEEN
FLAMM NERO; JAN MARIE
                                                     No. 99-15887
IMMKER PEEPLES; KAYE
                                                     D.C. No.
CHRISTENSON; CAROL COLLINS,
                                                     CV-95-1516-PHX-RCB
a single person,
Plaintiffs,
                                                     OPINION

and

JOSEPH GENTILE; LORRAINE
GENTILE, husband and wife,
Plaintiffs-Appellants,

v.

THE QUAKER OATS COMPANY, a
New Jersey corporation,
Defendant-Appellee.

Appeals from the United States District Court
for the District of Arizona
Robert C. Broomfield, District Judge, Presiding

Argued and Submitted
February 17, 2000--San Francisco, California

Filed November 20, 2000

Before: Betty B. Fletcher, William C. Canby, Jr., and
Diarmuid F. O'Scannlain, Circuit Judges.

                               14914


Opinion by Judge O'Scannlain;
Dissent by Judge B. Fletcher

_________________________________________________________________



COUNSEL

Arthur Pederson (argued) and Karen Karr, Mohr, Hackett,
Pederson, Blakley & Randolph, P.C., Phoenix, Arizona, and
James Hill, Phoenix, Arizona, for the plaintiffs-appellants.

                               14920


Martin Harris (argued), Connelly Sheehan Moran, Chicago,
Illinois, and Antonio Dominguez, Dominguez & Associates,
P.C., Phoenix, Arizona, for the defendant-appellee.

_________________________________________________________________

OPINION

O'SCANNLAIN, Circuit Judge:

We must decide whether former employees have raised a
genuine issue of material fact as to whether they were ille-
gally fired because of age.

I

A

Jerry Jeney ("Jeney"), Joseph Gentile ("Gentile") and Perry
Coleman ("Coleman"), along with hundreds of other employ-
ees nationwide, were laid off by the Quaker Oats Company
("Quaker") in Arizona during a series of reductions-in-force
("RIFs") from 1994 to 1995.

1

Jeney, a 55-year-old white male,1 began working for
Quaker in 1987 after it had purchased the company at which
Jeney was employed. At Quaker, Jeney worked for six years
as a sales representative, a position equivalent to that which
he had held at other companies since 1969. Jeney's responsi-
bilities involved visiting supermarkets to rotate products, to
arrange displays, to ensure adequate stocks, and to check if
the supermarkets displayed all the company's products they
were supposed to carry.
_________________________________________________________________
1 All ages are as of the year in which the employees were terminated.

                               14921


In 1993, a RIF threatened Jeney's job. Impressed with
Jeney's skills, Ken Willis, a Quaker supervisor, created a new
position for him--Area Retail Manager ("ARM"). Jeney was
the only ARM in the Quaker organization nationwide and in
this position assisted individuals with various accounts.
Jeney's previous responsibilities as sales representative had
been outsourced to independent food brokers. His new posi-
tion required him to inspect and to grade the performance of
the outside brokers and to prepare diagrams showing how to
arrange products on the shelves. Jeney held this job until the
1994 RIF.

In 1994, Quaker implemented a nationwide business reor-
ganization which eliminated almost all lower level sales posi-
tions. Quaker laid off 300 employees, including Jeney. The
work of these employees was farmed out to outside brokers.
This 50 percent reduction in workforce added high-paying
jobs as it cut most low-ranking jobs. In order to select the
employees who would remain, Quaker sent a human
resources team to rate employees in several categories. Most
of these categories emphasized skills developed through head-
quarters (management) work.

The ranking evaluated employees in six categories: strate-
gic business planning, advanced fact-based selling, category
expertise, customer operations understanding, leadership abil-
ity, and communications skills. All employees were also eval-
uated on their learning agility and versatility. Jeney received
poor scores in those categories representing headquarters
skills with which Jeney had no experience: strategic business
planning, advance fact-based selling, and category expertise.
He received average scores in the remaining categories. Jeney
also received an average score in "Learning agility," defined
as one's ability to learn from past mistakes. In "Versatility,"
Jeney received a rating of well-placed, meaning that he was
good at his current job, but not likely to advance. Overall he
received a grade of "B." During this process, Quaker moni-
tored the RIF to ensure that it did not have an adverse effect

                               14922


on race or gender diversity in the company. No action was
ever taken because no such effect was identified.

Positions at Quaker are ranked by "hay points. " A one-level
promotion entails an increase of 100 hay points. Jeney's posi-
tion as an ARM was 366 hay points. His previous position as
a sales representative represented 301 hay points. Following
the evaluation process, Quaker had to fill six positions in the
Phoenix, Arizona area where Jeney worked. Ken Willis, the
same employee who had created the ARM position specially
for Jeney the previous year, did not select him to fill any of
these job openings.

The first four job openings were for account executives,
positions ranked at 636 or 732 hay points. The fifth position
was a broker field manager at 732 hay points. The final open-
ing was for a customer manager, a position with 438 hay
points, much closer to Jeney's then current level of 366. All
candidates selected for the account executive positions had
previously held comparable positions with 732 hay points.
Three had received an A rating and the fourth a B+ rating.
The fourth individual also had years of headquarters experi-
ence. The candidate chosen to fill the high-level broker field
manager position (732 hay points) had received a B+ rating
and had managed brokers previously. Finally, Willis chose
Bob Dumais (a forty-year-old white male) for the Customer
Manager position. Dumais had received a B+ rating and had
headquarters experience. All six individuals chosen for these
positions were white, five were men, and three were forty or
older.

Several of these employees, however, chose to leave the
company. One of the account executives resigned. He was
replaced by a candidate who had received an A rating, but
who also resigned within a month. His replacement was
Bryan Hardman, a customer manager from Los Angeles who
had received an A rating. In November 1994, when another

                               14923


account executive resigned, he was replaced by Bob Dumais,
the previous customer manager, leaving that position open.

Jeney was considered for the vacant customer manager
position after his former manager recommended him for it.
Quaker decided against Jeney, however, at least in part
because it did not consider him to be "promotable." Instead,
Rucker chose Ken Oliver (a 34-year-old white male) with a
degree in marketing who was pursuing an M.B.A. in the eve-
nings. Jeney was laid off. By the time the RIF was completed,
Quaker had laid off two-thirds of its employees aged 40 and
older, and one-third of those employees under age 40.

2

Gentile, a 53-year-old white male, began working for
Quaker in 1978 as a sales representative. He was later pro-

moted to customer manager, the position he had when he was
discharged in 1994. Gentile's responsibilities involved han-
dling large supermarket accounts for both breakfast and pet
foods. Although Gentile's performance had earned him vari-
ous awards and positive performance reviews, particularly
with regard to his relationships with the buyers, Gentile's
reviews had recently worsened as his position required greater
analytical responsibilities. Quaker also received a complaint
about him from an independent broker.

During the 1994 RIF, Quaker evaluated Gentile using the
same system employed with Jeney. Gentile received a B rat-
ing overall, garnering praise for his customer relationships,
but criticism for his analytical skills. In Arizona, where Gen-
tile worked, Quaker filled six sales positions. The first four
job openings were for account executives, the fifth position
was a broker field manager, and the final opening was for a
customer manager. All of these jobs went to employees who
had received higher ratings. All six individuals chosen for
these positions were white, five were men, and three were
forty or older.

                               14924


3

Coleman, a 49-year-old male, began working for Quaker in
1972. At Quaker, Coleman rose from account representative
to account executive, the position he held until he was dis-
charged. Coleman's responsibilities included preparing busi-
ness planning initiatives and managing relations with assigned
customers. Coleman survived the 1994 RIF that ended the
careers of Jeney and Gentile. In fact, at that time Quaker pro-
moted Coleman from customer manager to account executive.

Then, in 1995, Quaker sold its pet foods business and
undertook another, smaller RIF, eliminating two sales posi-
tions in Phoenix. The four account executives and one cus-
tomer manager were evaluated by their supervisors, Julie
Forbes and Bob Blair. Coleman ranked third with 13 points,
placing him in the middle of the group. Lisa Rodriguez (16
points) and Bryan Hardman (14 points) ranked above him;
Bob Dumais (12 points) and Ken Oliver (11 points) ranked
below him. On the basis of these rankings, Quaker decided to
terminate Dumais and Oliver. At this point, however, Forbes,
Dumais' supervisor, intervened. She argued that the differ-
ence between Coleman and Dumais was only 1 point and
resulted from the fact that she was a tougher evaluator than
Coleman's supervisor, Blair. Blair and Forbes sparred over
whose employee would be kept on, and Blair finally relented.
Thus Quaker kept Dumais and discharged Coleman.

Finally, Quaker's national headquarters agreed to allow one
more sales position in Phoenix, but only on condition that it
rank below account executive in order to keep down costs.
Oliver, who had scored lower than Coleman, was kept.
Quaker states that it did so because the position was two
levels below Coleman's level and would have meant a large
pay cut for him. Fearing the loss in morale and knowing that
Oliver was already working on the account assigned to that
position, Quaker selected Oliver and laid off Coleman.

                               14925


B

Believing that they had been terminated because of their
age, Jeney, Gentile, and Coleman filed complaints with the
Equal Employment Opportunity Commission ("EEOC"). On
May 27, 1998, the EEOC issued its determination that it
found reasonable cause to believe Quaker had discriminated
against older employees as a class and had violated the law
when it terminated Jeney, Gentile, and Coleman.

In the meantime, in October 1995, Jeney, Gentile, and
Coleman, together with seven other plaintiffs, filed this suit in
the United States District Court for the District of Arizona.
All ten plaintiffs had been fired in the RIFs and all alleged
that Quaker fired them because of their age in violation of the
Age Discrimination in Employment Act ("ADEA"), 29
U.S.C. S 621 et seq. Their complaints also included state law
claims for breach of contract and discharge in violation of
public policy. Quaker moved to sever the cases. The district
court granted Quaker's motion on April 8, 1996, ruling that
trying the cases together could cause jury confusion and prej-
udice Quaker. The court ordered the cases sent to the districts
in which the plaintiffs had worked for Quaker. Because Jeney,
Gentile, and Coleman had worked in Arizona, these cases
remained there. The three cases proceeded separately but in
close coordination with one another.

On November 11, 1996, Jeney and Gentile, but not Cole-
man, filed a second complaint with the EEOC charging dis-
crimination on the basis of sex and race. On March 14, 1997,
Jeney and Gentile filed a second amended complaint in dis-
trict court alleging that they had been fired because of their
race and sex in violation of Title VII of the Civil Rights Act,
42 U.S.C. S 2000e.

After a contentious discovery period, both sides moved for
summary judgment. On March 10, 1999, the district court
granted summary judgment in favor of Quaker on all the

                               14926


claims. The judge assumed Jeney, Gentile, and Coleman had
made out a prima facie case of intentional age discrimination,
but concluded that they had failed to raise a genuine issue of
material fact as to whether Quaker's stated reasons for firing
them were pretextual. In addition, the district court found that
they had not timely pled age discrimination under the dispa-
rate impact theory of liability and rejected their motion to
amend their complaint to add this theory.

With respect to Jeney and Gentile, the district court found
that their claims of race and sex discrimination were barred by
the statute of limitations and that in any case they had not
raised a general issue of material fact that they had been dis-
criminated against on these grounds. With respect to Cole-
man, the judge granted Quaker's motion to strike his
statistical evidence because most of the analysis concerned
the 1994 RIF, not the 1995 RIF in which Coleman lost his
job. The judge also excluded, under Federal Rule of Evidence
408, Quaker's settlement offer made after Coleman had filed
his EEOC charge.

As the state law claims rested on the same factual basis as
the federal claims, the district court granted summary judg-
ment for Quaker on Jeney's, Gentile's and Coleman's claims
that they had been discharged in violation of Arizona public
policy and in breach of their employment contract. In addi-
tion, the district court concluded that Quaker's Code of Ethics
did not create contractual obligations under Arizona law.2

Jeney, Gentile and Coleman timely appealed.
_________________________________________________________________
2 The district court also dismissed Jeney's, Gentile's and Coleman's
claims under ERISA, a decision they do not appeal.

                               14927


II

A

The Age Discrimination in Employment Act ("ADEA"), 29
U.S.C. SS 621 et seq., makes it "unlawful for an employer . . .
to fail or refuse to hire or to discharge any individual [who is
at least 40 years of age] . . . because of such individual's age."
Id. at SS 623(a), 631(a). To establish a violation of ADEA
under the disparate treatment theory of liability, Jeney, Gen-
tile, and Coleman "must first establish a prima facie case of
discrimination. If [they do so], the burden then shifts to
[Quaker] to articulate a legitimate nondiscriminatory reason
for its employment decision. Then, in order to prevail, [they]
must demonstrate that [Quaker's] alleged reason for the
adverse employment decision is a pretext for another motive
which is discriminatory." Wallis v. J.R. Simplot Co., 26 F.3d
885, 889 (9th Cir. 1994). Despite the burden shifting, the ulti-
mate burden of proof remains always on the former employ-
ees to show that Quaker intentionally discriminated because
of their age. See Rose v. Wells Fargo & Co., 902 F.2d 1417,
1420-21 (9th Cir. 1990).

To establish a prima facie case using circumstantial evi-
dence, the employees must demonstrate that they were (1)
members of the protected class (at least age 40); (2) perform-
ing their jobs satisfactorily; (3) discharged; and (4) replaced
by substantially younger employees with equal or inferior
qualifications. Nidds v. Schindler Elevator Co. , 113 F.3d 912,
917 (9th Cir. 1997). In Coleman's case, Quaker appears to
concede that he made a sufficient prima facie showing. In
Jeney's and Gentile's cases, Quaker does not dispute that they
were members of the protected class or that they were dis-
charged. In Jeney's case, with respect to the job performance
element, Quaker contends that he had performed satisfactorily
in his sales positions, but was not as qualified as other
employees for the new, higher-level positions that remained
after the RIF. In Gentile's case, with respect to the job perfor-

                               14928


mance element, Quaker contends that he had performed satis-
factorily in his previous position, but that his performance
slipped as this position required greater analytical skill.
Quaker states, however, that it did not discharge him for poor
performance directly; rather it laid him off as part of the wide-
spread RIF. For both Jeney and Gentile then, the second ele-
ment of the prima facie case ties into the fourth element, i.e.,
whether their qualifications were equal to or better than those
of the retained employees.

Where, as here, the discharge in question results from a
general reduction in workforce, Jeney and Gentile need not
show that they were replaced; rather they need show "through
circumstantial, statistical, or direct evidence that the discharge
occurred under circumstances giving rise to an inference of
age discrimination." Rose, 902 F.2d at 1421. This inference
"can be established by showing the employer had a continu-
ing need for [their] skills and services in that [their] various
duties were still being performed," Wallis , 26 F.3d at 891, or
by showing "that others not in [their] protected class were
treated more favorably." Washington v. Garrett, 10 F.3d
1421, 1434 (9th Cir. 1994).

To establish their prima facie case, Jeney and Gentile offer
the following proof: statistical evidence showing that older
workers were terminated at twice the rate of younger workers;
the EEOC reasonable cause determinations in their cases; the
alleged statement by Quaker officials that Jeney was not
"young and promotable;" and the fact that younger, less expe-
rienced employees than they were kept. "The requisite degree
of proof necessary to establish a prima facie case for Title VII
and ADEA claims on summary judgment is minimal . . .."
Wallis, 26 F.3d at 889.

[1] The statistical evidence is problematic. On its face, the
data shows that compared to employees under age 40, nearly
twice the percentage of employees over age 40 (the protected
class) were laid off. This observation, however, does not take

                               14929


into account any variables other than age. When the statistical
expert considered other variables, such as education, the per-
centage of older employees in eliminated low-level jobs, and
the job market, the results were far less dramatic. As the dis-
trict court found, "[W]hen broken down for extensive peer
analysis, his results were not even significant. " Moreover, the
EEOC probable cause determinations in these cases carry lit-
tle weight since they are conclusory and completely devoid of
analysis. Finally, Jeney's and Gentile's subjective evaluations
of their relative qualifications and Quaker's alleged use of the
phrase "young and promotable" do not take them far. Never-
theless, despite the weaknesses in the evidence offered by
Jeney and Gentile to establish their prima facie cases, given
the low threshold required, we assume, without deciding, that
all three--Jeney, Gentile and Coleman--have established
such a case. A prima facie case raises a rebuttable presump-
tion that Quaker has violated ADEA.

[2] The burden now shifts to Quaker to articulate legiti-
mate, non-discriminatory reasons for the adverse employment
action. See Wallis, 26 F.3d at 889. Quaker states that it termi-
nated Jeney and Gentile pursuant to a general reduction in its
workforce and that it did not select them for any of the
remaining positions because they were not as qualified as
those employees chosen. It supports its argument that Jeney
and Gentile were less qualified than the selected candidates
by pointing to its rating system and their work experience.
With respect to Coleman, Quaker states that he ranked lower
than all employees retained but one in the 1995 ratings system
and that the job of that employee was saved after his supervi-
sor fought harder than Coleman's to keep him. A RIF is a
legitimate nondiscriminatory reason for laying off an
employee. See Nidds, 113 F.3d at 918. Furthermore, lack of
proper qualifications is such a reason. See id.  Thus, Quaker
has met its burden of production.

[3] At this point in the analysis, the "presumption of unlaw-
ful discrimination `simply drops out of the picture.' " Wallis,

                               14930


26 F.3d at 889 (quoting St. Mary's Honor Ctr. v. Hicks, 509
U.S. 502, 511 (1993)). To survive summary judgment, the
employees must now introduce evidence sufficient to raise a
genuine issue of material fact as to whether the reasons
Quaker articulated are pretexts for age discrimination. They
may rely on the same evidence they used to establish a prima
facie case or put forth additional evidence. See Wallis, 26
F.3d at 892.

[4] While summary judgment should be used prudently in
ADEA cases involving motivation and intent, see Rose, 902
F.2d at 1420, the establishment of a prima facie case based on
the minimum evidence necessary to raise a presumption of
discrimination does not preclude summary judgment. See
Wallis, 26 F.3d at 890-91; Lindahl v. Air France, 930 F.2d
1434, 1437 (9th Cir. 1991) ("[A] plaintiff cannot defeat sum-
mary judgment simply by making out a prima facie case.").
Jeney, Gentile, and Coleman "must do more than establish a
prima facie case and deny the credibility of [Quaker's] wit-
nesses." Schuler v. Chronicle Broadcasting Co., Inc., 793
F.2d 1010, 1011 (9th Cir. 1986). "In response to[Quaker's]
offer of nondiscriminatory reasons, [they] must produce `spe-
cific, substantial evidence of pretext.' " Wallis, 26 F.3d at 890
(quoting Steckl v. Motorola, Inc., 703 F.2d 392, 393 (9th Cir.
1983)).

1

In response to the reasons articulated by Quaker for laying
him off, Jeney puts forth the evidence used to establish his
prima facie case and asserts that the system utilized by the
company to rate employees in order to decide whom to retain
was purposively made subjective so as to provide a
"smokescreen" for discrimination on the part of the raters.
Thus in determining whether Jeney produced enough specific
substantial evidence to raise a genuine question as to whether
Quaker's reasons were pretextual, we consider: the statistical
evidence; the EEOC determination; Quaker's alleged state-

                               14931


ment that they wanted someone "young and promotable;" his
qualifications relative to those of the retained employees; and
evidence regarding use of the rating system as a cover for dis-
crimination.

[5] As noted above, at first blush, the statistical disparity
between the effect of the RIF on older and younger workers
appears significant. Jeney produced evidence showing that the
likelihood that this disparity would occur by chance was 3 in
100 billion. The problem, however, is that Quaker never con-
tends that the disparity occurred by chance; just that it did not
occur for discriminatory reasons. When other pertinent vari-
ables were factored in, the statistical disparity diminished and
finally disappeared. To establish a prima facie case based
solely on statistics, let alone raise a triable issue of fact
regarding pretext, the statistics "must show a stark pattern of
discrimination unexplainable on grounds other than age."
Rose, 902 F.2d at 1423 (internal quotation marks omitted).

[6] We have previously held that statistics remarkably simi-
lar to these failed to raise a triable issue of fact of intent to
discriminate. In Rose, a vice president whose job had been
eliminated when Wells Fargo & Company purchased Crocker
National Bank brought suit against Wells Fargo alleging age
discrimination. The statistical evidence he proffered to sur-
vive the bank's motion for summary judgment revealed that
of individuals age 50 or over in the employee's division, 73.5
percent were terminated. In contrast, 34.1 percent of individu-
als in their 40s were terminated, as were only 28.2 percent of
those under age 40. Likewise, Jeney's figures show that 64.4
percent of employees over age 40 were terminated in contrast
to 36.8 percent of those under that age. See Rose, 902 F.2d at
1423. In Rose, we rejected such statistics as insufficient to
raise a question of intentional discrimination. See id. We
pointed to the fact that older individuals tended to occupy the
level of positions eliminated during the reorganization. See id.
Likewise, older individuals tended to occupy the sales posi-
tions eliminated and outsourced by Quaker. Jeney's statistics

                               14932


failed to account for obvious variables--including education,
previous position at the company, and distribution of age
groups by position--that would have affected the results of
the analysis. See Rea v. Martin Marietta Corp. , 29 F.3d 1450,
1456 (10th Cir. 1994) (holding that to raise an inference of
discrimination, statistical evidence should account for possi-
ble nondiscriminatory variables); Sheehan v. Daily Racing
Form, Inc., 104 F.3d 940, 942 (7th Cir. 1997) (refusing to
give weight to statistical evidence that does not account for
relevant variables). Cf. Wards Cove Packing Co. v. Atonio,
490 U.S. 642, 657 (1989) (Without a showing of causation
between the challenged action and the alleged disparities,
"employers [would be] potentially liable for the myriad of
innocent causes that may lead to statistical imbalances in the
composition of their work forces."). At this point of the dispa-
rate treatment analysis, moreover, Jeney's proof must raise
questions as to the veracity of the reasons Quaker gave for not
retaining Jeney specifically. "[W]hile statistics are often pro-
bative in determining the existence of a pattern or practice of
disparate treatment, their usefulness in proving pretext
depend[s] primarily upon their relevance to the specific deci-
sion affecting the individual plaintiff, [and not ] those that
affect [Jeney's] protected class in general. " Penk v. Oregon
State Board of Higher Education, 816 F.2d 458, 462 (9th Cir.
1987). Because they fail to account for many factors pertinent
to Jeney, we conclude that the statistics are not enough to take
this case to trial.

[7] Nor does the EEOC reasonable cause determination
create a genuine issue of material fact. An EEOC letter is a
"highly probative evaluation of an individual's discrimination
complaint." Plummer v. Western Int'l Hotels Co., Inc., 656
F.2d 502, 505 (9th Cir. 1981). As required by our case law,
the district court properly admitted the letter into evidence.
Such letters, however, "are not homogeneous products; they
vary greatly in quality and factual detail." Johnson v. Yellow
Freight System, Inc., 734 F.2d 1304, 1309 (8th Cir. 1984).
The EEOC letter in this case stated simply that "The Commis-

                               14933


sion's investigation finds reasonable cause to believe that
[Quaker] has violated the [ADEA]. The Commission further
finds that employees over the age of 40, as a class, were laid
of in violation of [ADEA]." It is impossible from this letter
to know what facts the EEOC considered and how it analyzed
them. Examining similarly conclusory EEOC letters, other
circuits have concluded that when the letters only report "bare
conclusions," they have little probative value. See Cortes v.
Maxus Exploration Co., 977 F.2d 195, 201-02 (5th Cir. 1992);
see also Goldberg v. B. Green & Co., Inc., 836 F.2d 845, 848
(4th Cir. 1988). In previous cases, we have upheld summary
judgment in cases in which the EEOC itself has brought suit.
See, e.g., EEOC v. Insurance Co. of N. Am., 49 F.3d 1418 (9th
Cir. 1995). If the EEOC's suing is insufficient to create a gen-
uine issue of material fact, then, a priori, a conclusory EEOC
reasonable cause letter, at least by itself, does not create an
issue of material fact.

Next, Jeney asserts that Quaker refused to hire him because
it was looking for someone "young and promotable, " or at
least that Quaker's use of the word "promotable " was an ille-
gal synonym for "young." Quaker's employees testified to
using the word "promotable," but stated that they never used
the word "young." Jeney's own testimony is mixed on this
issue. Although in his affidavit Jeney stated that he had heard
that they had said "young or promotable," he cast doubt on
this declaration when during his deposition he stated that
"Boyd said that Rucker told him it had to go to somebody
young and mobile or promotable, or whatever term you want
to use."3
_________________________________________________________________
3 Curiously, Jeney's brief does not cite to (nor do his excerpts of record
include) the deposition of Steve Shields, a Quaker account executive, who
also reported that Quaker's management would not retain Jeney because
they wanted someone "younger." In contrast, Gentile does cite to the
Shields deposition. The Shields deposition is in the Jeney record, however,
and we have considered it.

                               14934


[8] In the context of this case, the word "promotable" by
itself does not give rise to an inference of discriminatory
motive. As Quaker explained, and Jeney does not contradict,
the reorganization of the company eliminated hundreds of
low-level positions, the type of jobs that individuals like
Jeney once kept for years without any expectation that they
would need to advance in the company. Once these positions
were farmed out to independent brokers, the remaining posi-
tions were higher level: a few customer management positions
and many account executive positions. Hence, Quaker
expected its customer managers to be promotable to account
executive positions within two years. Promotability became
an increasingly important consideration in hiring customer
managers. Cf. Furr v. Seagate Technology, Inc. , 82 F.3d 980,
987 (10th Cir. 1996) (stating that promotability is a legitimate
employer consideration). Given this background, none of
which is contradicted by Jeney, the use of the word promot-
able does not give rise to suspicions of age discrimination.
Moreover, we have held that the use of far more suggestive
words on the part of management do not create a genuine
issue of material fact as to age discrimination. See Nesbit v.
Pepsico, Inc., 994 F.2d 703, 705 (9th Cir. 1993) (holding that
employer's use of the phrase "[w]e don't necessarily like grey
hair" did not support an inference of discriminatory motive);
Nidds, 113 F.3d at 918-19 (holding that employer's use of the
phrase "old timers" did not support inference of discrimina-
tory motive); Rose, 902 F.2d at 1423 (holding that employer's
use of the phrase "old-boy network" did not support inference
of discriminatory motive); Insurance Co., 49 F.3d at 1421
(holding that rejecting employee because he was "overquali-
fied" did not support inference of discriminatory motive). Cf.
Hazen Paper Co. v. Biggins, 507 U.S. 604, 611 (1993) (hold-
ing that even adverse employment actions motivated by fac-
tors strongly correlated with age are not illegal under ADEA).
Although Jeney has proferred sufficient evidence to raise a
question of fact as to whether one member of Quaker's man-
agement used the words "young and promotable, " this is not
sufficient to raise a question of fact as to whether the reasons

                               14935


Quaker gave for laying Jeney off were pretexts for age dis-
crimination. See Merrick v. Farmers Ins. Group , 892 F.2d
1434, 1438-39 (9th Cir. 1990) (holding that the hiring super-
visor's statement that he hired another employee over the
plaintiff because he was "a bright, intelligent, knowledgeable
young man" was not enough to raise an inference of age dis-
crimination).

[9] Fourth, Jeney attacks the rating system used by Quaker
as subjective and thus a cover for age discrimination. While
a subjective evaluation system can be used as cover for illegal
discrimination, subjective evaluations are not unlawful per se
and "their relevance to proof of a discriminatory intent is
weak." Sengupta v. Morrison-Knudsen Co., Inc. , 804 F.2d
1072, 1075 (9th Cir. 1986). Most of Jeney's criticism of the
ratings system centers on his contention that it did not do a
good job of evaluating the employees and other methods, such
as standardized testing, would have done better. Jeney's
expert witnesses concluded as much. But this allegation does
little to help Jeney establish that Quaker used a subjective sys-
tem in order to discriminate against older employees. That
Quaker made unwise business judgments or that it used a
faulty evaluation system does not support the inference that
Quaker discriminated on the basis of age. See Sharpe v. AT&T

Co., 66 F.3d 1045, 1050 (9th Cir. 1995) (applying state law);
Cotton v. City of Alameda, 812 F.2d 1245, 1249 (9th Cir.
1987) ("The ADEA does not make it unlawful for an
employer to do a poor job of selecting employees. It merely
makes it unlawful to discriminate on the basis of age."). Jeney
offers one item of proof that could create such an inference:
employees over age 40 received significantly lower rankings
on average than employees under age 40. Ultimately though,
these statistics suffer from the same infirmity as the statistics
regarding the layoffs--they do not account for any relevant
factors that might explain the variation.

[10] Fifth, Jeney asserts that younger, less qualified indi-
viduals were hired for the jobs for which he was qualified. As

                               14936


evidence of his qualifications, Jeney relies primarily on his
years of experience and his strong performance reviews. But
"[t]he question is not whether [Jeney] in the abstract had bet-
ter qualifications than the other candidates. The question is
whether the other candidates are more qualified with respect
to the criteria that [Quaker] actually employs." Cotton, 812
F.2d at 1249 (rejecting the ADEA claim of an employee who
argued he had more experience than those hired). Jeney's
"subjective personal judgments of [his] competence alone do
not raise a genuine issue of material fact." Bradley v. Har-
court, Brace and Co., 104 F.3d 267, 270 (9th Cir. 1996).

Of the six jobs filled in the Phoenix area following the rat-
ings, four were account executive positions. These positions
would have required a promotion of four levels for Jeney. As
Jeney himself attests, such a promotion is unheard of in the
company. The fifth position--broker field manager--would
have also represented a comparable promotion. This left the
opening for a customer manager, a job requiring a one-level
promotion and a position for which Jeney was considered.
Thus, Jeney must raise a genuine issue that Quaker's reasons
for hiring others instead of him for the customer manager
position are pretexts. Jeney argues that he was more qualified
than Lisa Rodriguez, Bryan Hardman, Glen Synoground,
Robert Dumais, Ken Oliver, and Stacey Ring. First, Rodri-
guez, Hardman, and Synoground were hired as account exec-
utives, the position four levels above Jeney's. Rodriguez had
already held a comparable position to the senior account exec-
utive job she was given. Robert Dumais, a 40-year-old white
male who had already held an account executive position in
San Antonio, was selected as a customer manager, but within
months was promoted to account executive when another
employee retired.

Ken Oliver, a 34-year-old white male with a degree in mar-
keting and who was studying for his M.B.A. during the eve-
nings, fit Quaker's emphasis on analytical skills. Quaker
stated that it also hired Oliver on the basis of a recommenda-

                               14937


tion from an outside broker who knew both Oliver and Jeney.
Cf. Merrick, 892 F.2d at 1437 (stating that the hiring supervi-
sor may properly rely on the recommendation of others in
making his decision). Quaker believed he was more likely
than Jeney to be promoted to account executive, an important
consideration in the new company. Stacey Ring, a 26-year-old
white female, had a degree in quantitative economics and
Quaker believed she was more qualified that Jeney to partici-
pate in analytical projects at headquarters.

Jeney's subjective evaluation of his qualifications cannot
raise an issue of material fact, and the one non-subjective
qualification he points to is his number of years of experience.
All but one of these years of experience was as a sales repre-
sentative. The little headquarters experience Jeney did have
came as an assistant for a year to two supervisors and
involved making schematics for shelf placement. See Merrick,
892 F.2d at 1438 (holding that a plaintiff has not raised an
inference of discrimination where he introduces evidence that
he may have been more qualified in some respects than the
chosen employee, but introduces no evidence that the compa-
ny's stated reasons for not hiring him were pretextual); Cot-
ton, 812 F.2d at 1249 (holding that the number of years of
experience is not relevant where this is not one of the qualifi-
cations sought by the employer). Furthermore, his case is
weakened by the fact that Ken Willis, who created the special
ARM position for Jeney in 1993, was the same individual
who made the decision not to rehire him initially."[W]here
the same actor is responsible for both the hiring and the firing
of a discrimination plaintiff, and both actions occur within a
short period of time, a strong inference arises that there was
no discriminatory motive." Bradley, 104 F.3d at 270-71. In
both Bradley and in Jeney's case, the decisions to hire and
then to terminate were about a year apart. See id. at 269.

Finally, Jeney attacks many of the grounds Quaker gives
for firing him because he was not told of these reasons when
he was let go. Jeney is correct that doubt is cast on an employ-

                               14938


er's proffered reasons for why an employee was laid off
where a straightforward answer was not given when he or she
was terminated, but later is provided during litigation. See
Lindhal, 930 F.2d at 1438 ("[T]he computer explanation
would have been such a straightforward answer to[the
employee's] inquiries that one might expect that[her supervi-
sors] would have mentioned it if it really were the explana-
tion."). In Jeney's case, however, Quaker informed all the
employees that they were being terminated pursuant to the
RIF in which low-level sales positions were eliminated and
the employees retained would be in higher level supervisory
or management positions. The ratings systems Quaker used to
help determine who remained included many of the specific
reasons Quaker gave during the litigation to explain why
Jeney was not as qualified as other employees, including
Jeney's lack of headquarters experience and skills and the fact
they considered him unlikely to be promoted. As the later rea-
sons mainly detail the earlier one, they are "not properly
described as `shifting reasons.' " Nidds, 113 F.3d at 918
(holding that employer's first explanation that "lack of work"
was the reason for the layoffs and later statement that "lack
of seniority and poorer performance relative to other mechan-
ics" was the reason for the particular employee's termination
did not create a genuine issue of material fact as to whether
employer's reasons were pretextual). Quaker has consistently
explained that the reorganized company required employees
with increased analytical abilities. Jeney has not cast doubt on
the sincerity of this explanation or on Quaker's conclusion
that his skills were not as strong in this area as those of the
individuals retained.

Jeney had over twenty years of experience as a sales repre-
sentative. But Quaker was no longer employing sales repre-
sentatives. Their responsibilities were filled by outside
brokers. All of the positions at Quaker now required the
employees to have analytical and management skills. In the
RIF context, Jeney must show that the skills he had continued
to be needed by the company. "Without the demonstration of

                               14939


a need for the same services and skills that [Jeney] pos-
sessed," Jeney's case falters. Sengupta, 804 F.2d at 1075.
Jeney has not made such a demonstration. Simply put, Quaker
no longer employed anyone to do the job Jeney had done for
over 20 years. " `In a reduction-in-force case, there is no
adverse inference to be drawn from an employee's discharge
if his position and duties are completely eliminated. . . . If
[Jeney] cannot show that [Quaker] had some continuing need
for his skills and services in that his various duties were still
being performed, then the basis of his claim collapses.' "
Rose, 902 F.2d at 1421 (quoting Leichihman v. Pickwick Int'l,
814 F.2d 1263, 1270 (8th Cir. 1987)).

[11] To survive summary judgment, the burden is on Jeney
to "produce enough evidence to allow a reasonable factfinder
to conclude either: (a) that the alleged reason for[his] dis-
charge was false, or (b) that the true reason for his discharge
was a discriminatory one." Nidds, 113 F.3d at 918. Looking
at the evidence in the light most favorable to Jeney, no rea-
sonable juror could conclude--on the bases of the statistics
that did not account for obvious factors, the evidence regard-
ing Quaker's use of "promotable" or "younger," Jeney's own
evaluations of his relative qualifications, and his assertion that
the ratings were unduly subjective--that Quaker intentionally
discriminated against him because of his age. Thus, we affirm
the district court's grant of summary judgment on Jeney's dis-
parate treatment claim under ADEA.

2

[12] Gentile makes essentially the same case as Jeney does
in opposing summary judgment on whether Quaker's reasons
for firing him are pretexts for age discrimination. To the
extent that Gentile's case rests on the statistical evidence, the
EEOC determination, Quaker's alleged statements regarding
Jeney's youth and promotability, and the rating system, and
his contention that Quaker's stated reasons for terminating
him have shifted, we reject it for the reasons explained with

                               14940


respect to Jeney. Gentile's assertion that Quaker hired youn-
ger, less qualified individuals for the jobs for which he was
qualified, however, requires separate analysis.

As evidence of his qualifications, Gentile relies primarily
on his years of experience as a customer manager and his
strong performance reviews. But as with Jeney "[t]he question
is not whether [Gentile] in the abstract had better qualifica-
tions than the other candidates. The question is whether the
other candidates are more qualified with respect to the criteria
that [Quaker] actually employs." Cotton , 812 F.2d at 1249. In
addition, his more recent performance reviews reveal that
Gentile failed to conduct analyses and improperly focused on
sales volume rather than profitability.

Of the six jobs filled in the Phoenix area following the rat-
ings, four were account executive positions, one was for a
broker field manager, and one for a customer manager. Gen-
tile asserts that he is more qualified than three individuals
chosen for the Phoenix office: Lisa Rodriguez, Bob Dumais
and Glen Synoground. He also states that he is more qualified
than two other employees retained in the Los Angeles office
as customer managers: Bryan Hardman and Stacey Ring.

Lisa Rodriguez and Glen Synoground were retained as
account executives, positions which would have required a
promotion for Gentile. While Gentile was a customer man-
ager, Rodriguez was already a senior account executive and
Synoground an account executive before the 1994 RIF. Quak-
er's failure to fire two higher ranking employees in order to
promote Gentile in the midst of a large restructuring involving
hundreds of layoffs does not raise an inference of discrimina-
tion. All three of the remaining employees hired as customer
managers rated higher than Gentile; there is ample evidence
of Gentile's weak analytical abilities, a crucial consideration
at a time when the nature of the job was changing. Two of the
individuals retained as customer managers, Hardman and
Ring, worked in the Los Angeles office. We have previously

                               14941


noted that a company need not transfer an employee to
another office during a RIF in order to retain him. See Rose,
902 F.2d at 1422. During the 1994 nationwide RIF, no
employee who received Gentile's low rating was moved to
another office. Bob Dumais, retained as a customer manager,
had already held an account executive position in the San
Antonio office. Gentile proffers no evidence that his analyti-
cal abilities are superior to those of the employees Quaker
retained, the very skills Quaker has given as the rationale for
its employment decisions. Gentile's "subjective personal
judgments of [his] competence alone do not raise a genuine
issue of material fact." Bradley, 104 F.3d at 270. Furthermore,
Ken Willis, who in 1993 retained both Gentile and Coleman
over a third employee when one of three positions was elimi-
nated, was the supervisor responsible for filling the positions
for which Gentile was not selected in 1994. As with Jeney,
this fact gives rise to an inference that no discrimination took
place. See id. at 270-71.

Like Jeney, Gentile has failed to "produce enough evidence
to allow a reasonable factfinder to conclude either: (a) that the
alleged reason for [his] discharge was false, or (b) that the
true reason for his discharge was a discriminatory one."
Nidds, 113 F.3d at 918. Accordingly, the district court did not
err when it granted Quaker's motion for summary judgment
with respect to Gentile's ADEA claim of disparate treatment.

3

[13] Although Coleman was terminated a year later during
the 1995 RIF, his case remains analytically similar to those of
Jeney and Gentile. Quaker assumes, as will we, that Coleman
has established a prima facie case of disparate treatment on
the basis of age. Quaker states that it had initially decided to
retain Coleman over another employee, Bob Dumais, in 1995,
but terminated him after Dumais' supervisor fought harder
than Coleman's to save his job. In response to the explanation
articulated by Quaker for laying him off in 1995, Coleman

                               14942


puts forth the following evidence: statistics regarding the
1994 RIF; the EEOC finding of reasonable cause; Quaker's
retention of younger, less experienced employees; and Quak-
er's settlement offer.

We have already addressed the weaknesses of the statistical
evidence in the context of Jeney's case. In Coleman's case,
the district court struck this evidence from the record. The
court noted that almost all the analysis was based on the 1994
data, that the limited analysis of the 1995 data considered no
relevant variables, and that the 1995 data analysis included all
post-1994 terminations, not just the 1995 RIF discharges.
Although Coleman uses the evidence in his brief to our court,
he never challenges the district court's order striking the evi-
dence. Accordingly, he has waived this issue, and the statisti-
cal evidence cannot be considered part of the record. See
Alaska Ctr. for Env't v. U.S. Forest Serv., 189 F.3d 851, 858
n.4 (9th Cir. 1999).4 Therefore, Coleman's attack on the sub-
jective nature of the rating system is besides the point. More-
over, Coleman's 1994 promotion indicates that he benefitted
from this system.

The EEOC reasonable cause determination is as conclusory
in Coleman's case as it is in Jeney's and Gentile's. In fact it
appears to have been a form letter. Indicative of its conclusory
nature and lack of probative value, the EEOC letter in Cole-
man's case includes the recitation that Quaker discriminated
against older employees as a class despite the fact that there
was no class in 1995 when Coleman was terminated. Coleman
was the only employee laid off in that RIF who filed an
EEOC charge, and the EEOC had no information regarding
other employees terminated that year.
_________________________________________________________________
4 Even in his reply brief, Coleman does not directly challenge the district
court's order striking the evidence, although he does cite to cases support-
ing the proposition that evidence of past discrimination can be relevant to
a later case. In any case, issues cannot be raised for the first time in a reply
brief. See Alaska Ctr., 189 F.3d at 858 n.4.

                               14943


Like Jeney and Gentile, Coleman asserts that younger, less
qualified individuals were retained in jobs for which he was
qualified. Specifically, Coleman points to Lisa Rodriguez,
Bart Wassom, Quentin Burrell, Bryan Hardman, Robert
Dumais, and Ken Oliver. As evidence of his qualifications,
Coleman relies primarily on his years of experience and his
strong performance reviews. In order to survive summary
judgment, Coleman must raise a genuine issue of material fact
as to whether the reasons Quaker has given for retaining these
six individuals instead of him are pretexts. First, Coleman
concedes that he should not have been retained over Rodri-
guez. Second, Bart Wassom and Quentin Burrell were not
hired in Phoenix, they were hired in Los Angeles. Quaker had
no duty to transfer Coleman to another part of the country.
See Rose, 902 F.2d at 1422 ("When an employer reduces its
work force for economic reasons, it incurs no duty to transfer
the employee to another position within the company."). In
the 1995 RIF, the decisions on which employees to retain
were purely local. Quaker did not transfer employees to other
regions when it discharged Coleman. Coleman points to no
evidence to the contrary.

Despite Coleman's assertion that he was more qualified
than Hardman, Hardman scored higher in the 1995 ratings.
Unlike Jeney and Gentile with respect to the 1994 evalua-
tions, Coleman does not attack the ratings method used in the
1995 RIF as a cover for age discrimination. We have held that
where the plaintiff does not even attack the selection criteria
as having a disparate impact on the protected class,"[w]e will
not second guess the selection criteria used by an employer."
Cotton, 812 F.2d at 1249. Quaker has said that it retained
Hardman because he was more qualified. Hardman scored
better than Coleman, and Coleman does not challenge the rat-
ing as biased. Coleman's "subjective personal judgments of
[his] competence alone do not raise a genuine issue of mate-
rial fact." Bradley, 104 F.3d at 270.

Coleman and Dumais competed for the final position.
Quaker initially selected Coleman. After a protracted struggle

                               14944


between their respective managers, Forbes, Dumais's man-
ager, won out. Dumais was retained and Coleman discharged.
Coleman points to no evidence that this account is not credi-
ble. He states only that Forbes's statements may be believed
or disbelieved by the jury. This is not enough to withstand
summary judgment. See Wallis, 26 F.3d at 890 ("[The plain-
tiff] must do more than establish a prima facie case and deny
the credibility of the [defendant's] witnesses."); Lindahl, 930
F.2d at 1437-38 ("The plaintiff cannot carry this burden sim-
ply by restating the prima facie case and expressing an intent
to challenge the credibility of the employer's witnesses on
cross-examination."). Nor, in the context of this decision, are
his assertions that he is more qualified than Dumais to the
point. In this case, Quaker admits that Coleman scored one
point above Dumais. In fact because of this, it had initially
decided to retain Coleman. Quaker's legitimate, non-
discriminatory reason is that Forbes convinced Coleman's
manager that Dumais should be kept over Coleman, and Cole-
man has introduced no evidence to cast doubt on this version
of the hiring decision. The "favoritism" of managers is "not
age discrimination." Shutt v. Sandoz Crop Protection Corp.,
944 F.2d 1431, 1433 (9th Cir. 1991).

The last employee to whom Coleman points as less quali-
fied than he is Oliver. Quaker admits that Oliver, a customer
manager, is less qualified in the abstract than Coleman, an
account executive, but states that because of this Oliver was
a more appropriate fit for the customer manager position. For
Coleman to take the position would have meant a demotion
and a sharp cut in salary. While we have recognized that
"overqualification" as a reason for firing an employee may be
a mask for age discrimination, an employer may choose not
to hire an employee because he is overqualified for a position
without violating ADEA. See Insurance Co., 49 F.3d at 1420-
21. This is particularly true where the criteria that make the
employee overqualified and the reasons for the employer's
concern are defined. See id. at 1421. Here, Coleman does not
contradict Quaker's argument that placing him in this position

                               14945


would have meant a large pay cut and a two-step demotion.
Coleman's previous position as an account executive is an
objective criterion and Quaker defined its concern that a
demotion would have meant a loss of morale. Thus explained,
Quaker's rejection of Coleman as overqualified is a legitimate
nondiscriminatory reason and Coleman has introduced no evi-
dence controverting Quaker's position on this issue which
would allow him to survive summary judgment. See id. In
sum, Coleman has not raised a genuine issue of fact as to
whether the reasons Quaker gave for not retaining him instead
of these six employees were pretextual.

Finally, Coleman argues that the district court should have
admitted into evidence Quaker's offer of additional medical
benefits in exchange for Coleman's release of claims against
the company. The district court excluded this evidence under
Federal Rule of Evidence 408.5 In Cassino v. Reichhold
Chemicals, Inc., 817 F.2d 1338 (9th Cir. 1987), we created an
exception to the general rule against admitting offers to settle
a case. In Cassino, we held that it is not an abuse of discretion
for the trial judge to admit evidence of an employer's offer of
a "severance pay package in exchange for a release of all
potential claims" when the offer is "contemporaneous" with
the termination of the employment relationship. Id. at 1342.
The court made clear, however, that offers made "in connec-
tion with the negotiation and settlement of disputes arising
after the termination" remain inadmissable. Id. Thus, where
an employer offered outplacement services, in addition to its
ordinary severance package, three weeks after the employee's
discharge in exchange for a release, the district court properly
excluded evidence of the offer even where the employee had
yet to file any claims. See Mundy v. Household Finance
Corp., 885 F.2d 542, 546-47 (9th Cir. 1989).
_________________________________________________________________
5 "Evidence of (1) furnishing or offering or promising to furnish . . . a
valuable consideration in compromising or attempting to compromise a
claim which was disputed as to either validity or amount, is not admissible
to prove liability for or invalidity of the claim or its amount." Fed. R.
Evid. 408.

                               14946


In Coleman's case, he had been informed in March 1995 of
his discharge and he received his severance package without
condition. On May 8, he filed his age discrimination charge
with the EEOC. On May 12, Quaker offered him additional
medical benefits in exchange for a release of claims. As the
settlement offer was not contemporaneous with termination
and the consideration was additional to the standard severance
benefits, the district court did not abuse its discretion in
excluding the offer. See Mundy, 885 F.2d at 547. Coleman's
contention that the district court should not have excluded the
evidence because Quaker may not have known of the EEOC
charge when it made him the offer is foreclosed by Mundy. In
Mundy, the plaintiff had only hired an attorney; he had not
brought a charge, much less made his charge known to the
company. See id. Still, we held that the trial judge did not
abuse his discretion in excluding the evidence. See id. Here,
Coleman had already filed his charge with the EEOC.
Accordingly, this evidence cannot be considered to determine
whether Coleman has raised a genuine issue of material fact.

To survive summary judgment, the burden is on Coleman
to "produce enough evidence to allow a reasonable factfinder
to conclude either: (a) that the alleged reason for[his] dis-
charge was false, or (b) that the true reason for his discharge
was a discriminatory one." Nidds, 113 F.3d at 918. Looking
at the evidence in the light most favorable to Coleman, no rea-
sonable juror could conclude--on the basis of the EEOC
determination and the evidence regarding the reasons for
Quaker's hiring decisions--that Quaker intentionally discrim-
inated against him because of his age. Thus, we affirm the
district court's grant of summary judgment on Coleman's
ADEA disparate treatment claim.

III

[14] When challenging an adverse employment action
under the ADEA, an employee may proceed under two theo-
ries of liability: disparate treatment or disparate impact. Under

                               14947


the disparate impact theory, employees must prove that a
facially neutral employment practice had a discriminatory
impact on older workers. See Rose, 902 F.2d at 1421. They
need not prove motive or intent, but the company can defend
by establishing that the challenged practice was based on
legitimate business reasons, such as job-relatedness or busi-
ness necessity. See id. at 1424. At that point, the employee
would have to show that other selection practices could serve
the same business interest identified by the company without
having a discriminatory effect. See id.

The district court dismissed Jeney's, Gentile's, and Cole-
man's disparate impact claims on the grounds that they had
not pled this theory of liability in their complaint. Jeney, Gen-
tile and Coleman do not dispute that neither their first nor sec-
ond amended complaints included this theory of liability.
They admit that the first time they raised this claim was in
their motions for summary judgment. Rather, the employees
argue that either they need not specifically allege the disparate
impact theory in the complaint or the district court should
have allowed them to amend their complaints to add this the-
ory at the summary judgment stage.

We have yet to decide whether both theories of liability--
disparate impact and disparate treatment--must be alleged in
the complaint in order to provide the basis for later trial or
summary judgment.6  But cf. Klein v. Boeing Co., 847 F.
Supp. 838, 844 (W.D. Wash. 1994) ("Klein never asserted
this claim prior to his opposition to Boeing's motion for sum-
mary judgment, nor has he moved to amend his complaint to
_________________________________________________________________
6 The lack of case law on this issue in the courts of appeal stems in large
part from the fact that several of our sister circuits do not recognize a dis-
parate impact cause of action under ADEA after Hazen Paper Co. v. Big-
gins, 507 U.S. 604 (1993). See, e.g., Ellis v. United Airlines, Inc., 73 F.3d
999 (10th Cir. 1996); Gehring v. Case Corp., 43 F.3d 340 (7th Cir. 1994).
We have continued to assume that this theory is still viable after Hazen.
See Mangold v. California Pub. Util. Comm'n, 67 F.3d 1470, 1474 (9th
Cir. 1995).

                               14948


add such a claim. Thus, this claim is not properly before the
court."). Jeney, Gentile, and Coleman rely on Gomes v. AVCO
Corp., 964 F.2d 1330 (2nd Cir. 1992), for the proposition that
disparate impact need not be separately alleged. In Gomes, the
district court had dismissed the discriminatory impact theory
of liability because it found that the plaintiff had not raised it
in his EEOC compliant, which must be filed before a case is
brought. See id. at 1334. Reversing, the Second Circuit held
that Gomes' EEOC complaint would have caused the defen-
dant reasonably to expect that the discriminatory impact of its
facially neutral rule would also be investigated by the EEOC.
See id. The employees argue that, similarly, the allegation of
discriminatory treatment in their district court complaint gave
Quaker notice that both theories of liability could be pursued.

Thus the issue is whether Jeney, Gentile and Coleman, hav-
ing failed to allege the discriminatory impact theory of liabil-
ity, may nevertheless proceed under that theory after the close
of discovery in their motion for summary judgment. Their
case is more akin to Josey v. John R. Hollingsworth Corp.,
996 F.2d 632 (3d Cir. 1993). In Josey, the Third Circuit held
that where a plaintiff sets forth only the disparate treatment
theory in his pleadings and does not move to amend his com-
plaint until summary judgment following the close of discov-
ery, the plaintiff is barred from proceeding on the disparate
impact theory. See id. at 641-42. The court reasoned that
adding a new theory of liability at the summary judgment
stage would prejudice the defendant who faces different bur-
dens and defenses under this second theory of liability. See id.
at 642. The court held that the plaintiff should have moved to
amend his pleadings during discovery. See id. ; see also Kalsi
v. New York City Transit Auth., 62 F.Supp.2d 745, 760-61
(E.D.N.Y. 1998), aff'd, 189 F.2d 462 (2nd Cir. 1999).

[15] We agree. Allowing Jeney, Gentile and Coleman to
proceed with their disparate impact theory after the close of
discovery would prejudice Quaker. A complaint guides the
parties' discovery, putting the defendant on notice of the evi-

                               14949


dence it needs to adduce in order to defend against the plain-
tiff's allegations. A disparate impact theory, lacking the
requirement that the plaintiff prove intent and focusing on sta-
tistical analyses, requires that the defendant develop entirely
different defenses, including the job relatedness of the chal-
lenged business practice or its business necessity. Neither of
these are necessary to defend against a disparate treatment
theory. This case illustrates the problem. At no time prior to
summary judgment, did Jeney, Gentile, or Coleman identify
which facially neutral Quaker employment practice they chal-
lenged as having a discriminatory impact. Cf. Josey, 996 F.2d
at 642. The district court judge's opinion indicates that after
more than two years of discovery, he had no idea until the
employees' motions for summary judgment were filed that
they intended to pursue this legal theory. The lack of notice
on this issue central to the cause of action makes it difficult,
if not impossible, for Quaker to know how to defend itself.
After having focused on intentional discrimination in their
complaint and during discovery, the employees cannot turn
around and surprise the company at the summary judgment
stage on the theory that an allegation of disparate treatment in
the complaint is sufficient to encompass a disparate impact
theory of liability.

The Second Circuit's decision in Gomes does not teach oth-
erwise. See Kalsi, 62 F.Supp.2d at 760-61 (following Josey in
the Eastern District of New York). Cf. AFSCME v. County of
Nassau, 96 F.3d 644, 647 (2d Cir. 1996) (stating in its recita-
tion of the facts that "[b]ecause AFSCME never amended its
complaint to adequately plead a disparate impact claim,
AFSCME was required to introduce evidence from which it
could be inferred that the County intentionally discriminated
in setting the pay scale in order to prevail under a disparate
treatment theory."). In Gomes, the district court had dismissed
the complaint containing both theories of ADEA liability
because it was broader than the plaintiff's EEOC charge.
Gomes did not confront the very different situation we face
here in which the plaintiffs proceeded at summary judgment

                               14950


on a new theory. An EEOC charge leads only to an investiga-
tion, and once a complaint is filed in court, the defendant has
ample opportunity to prepare its defenses during discovery.
Thus the concerns that animated the Third Circuit in Josey
were not present in Gomes. In Gomes, there had been no dis-
covery, and the parties had not spent years developing the evi-
dence to prove their cases.

The dissent cites two Ninth Circuit cases, one which is not
relevant and another which supports our holding. First, the
dissent urges that the language of a footnote in EEOC v. Local
350, Plumbers and Pipefitters, 998 F.2d 641(9th Cir. 1993),
be applied outside of the very limited and special context of
that case. See post at 14960-61. Local 350 reversed the dis-
trict court's grant of summary judgment in favor of the union
on the EEOC's action under ADEA. See id. at 643. While
Local 350 was pending, Hazen Paper Co. v. Biggins, 507 U.S.
604 (1993), was decided. Hazen, as noted in Local 350, set
forth "the parameters of disparate treatment in the age dis-
crimination context." Id. at 648 n.2. The court also noted
Hazen's articulation of the disparate impact claim.7  See id. In
this same footnote, the panel acknowledges that "both theo-
ries were potentially available to the EEOC." Id. The court
then remanded to the district court with the exceptional order
to "direct the EEOC to articulate its theory or theories of dis-
crimination in accordance with the principles set forth in
Hazen." Id.

The principles set forth in Hazen, as discussed in Local
350, concerned the requirements for disparate treatment and
disparate impact claims. See id. The court recognized that the
_________________________________________________________________
7 Hazen, while discussing the requirements of a disparate impact claim,
specifically left open the question of whether such a claim was cognizable
under the ADEA. See Hazen Paper Co. v. Biggins , 507 U.S. 604, 609-10.
We have continued to assume that this theory is still viable after Hazen.
See Mangold v. California Pub. Util. Comm'n, 67 F.3d 1474 (9th Cir.
1995).

                               14951


EEOC filed its complaint prior to the decision in Hazen and
amid confusion of what claims were cognizable under the
ADEA. The remand order gave the EEOC an opportunity to
articulate its claims in light of Hazen's explication of poten-
tial ADEA claims. Plaintiffs filing ADEA claims after Hazen
would presumably be aware of their potential Hazen-based
claims. Since Jeney's, Gentile's, and Coleman's complaints
were filed after Hazen was decided, the footnote in Local 350
cited by the dissent is simply not applicable to them.

The dissent also incorrectly relies on Arnett v. California
Public Employees Retirement System, 179 F.3d 690 (9th Cir.
1999), vacated and remanded on other grounds, _______ U.S.
_______, 120 S.Ct. 930 (2000), to support its proposition that
unspecific statements in the pleadings will suffice to raise a
claim of disparate impact. See post at 14961-62. Arnett, how-
ever, supports our analysis. There, the court reversed the dis-
trict court's dismissal of plaintiff's disparate impact and
disparate treatment claims. See id. at 692. The facts of Arnett
are vastly different from those here. First, in Arnett, the plain-
tiffs had pled a "disparate impact" claim. See id. at 693. Sec-
ond, the court determined that the pleading was sufficient
because the plaintiffs alleged "that the consequences of [the
disputed practice] f[e]ll more harshly on those employees
hired at age 40" and had documented the impact of the dis-
puted practice. Id. at 697. Here, the plaintiffs have not ful-
filled any of the conditions examined by the court: They have
not stated that they were alleging a "disparate impact" claim,
nor alleged that the consequences of a practice fell more
harshly on them, nor provided documentation of the disparity
with their complaint.

Thus, we hold that the plaintiffs, who clearly stated ADEA
claims of disparate treatment but sought also to pursue claims
of disparate impact, were required either (1) to plead the addi-
tional disparate impact theory in their complaints, or (2) to
make known during discovery their intention to pursue recov-
ery on the disparate impact theory omitted from their com-

                               14952


plaints. Only if the defendants have been put on notice may
the plaintiffs proceed on a disparate impact theory at the sum-
mary judgment stage. Because Jeney, Gentile and Coleman
raised the disparate impact theory of liability for the first time
at summary judgment, the district court did not err when it did
not allow them to proceed on it.8

IV

Jeney, Gentile, and Coleman argue that, even if they are
required to plead the disparate impact theory in their com-
plaints, the district court should have granted their motions to
amend their pleadings to include such a claim. The employees
moved to amend their complaints in their replies to Quaker's
response to their motion for summary judgment. The district
court denied their motions because they did not show good
cause for failing to amend their complaints earlier.

Generally, Federal Rule of Civil Procedure 15(a) liberally
allows for amendments to pleadings. In this case, however,
the district court correctly found that it should address the
issue under Federal Rule of Civil Procedure 16 because it had
filed a pretrial scheduling order that established a timetable
for amending the pleadings, and the deadline had expired
before Jeney, Gentile, and Coleman moved to amend. See
Johnson v. Mammoth Recreations, Inc., 975 F.2d 604, 607-09
(9th Cir. 1992). Under Rule 16(b), Jeney, Gentile, and Cole-
man must show good cause for not having amended their
_________________________________________________________________
8 Palmer v. United States, 794 F.2d 534 (9th Cir. 1986), raised in the
employees' reply briefs, is not to the contrary. In Palmer, the district court
considered the disparate impact claim despite the fact that it was not spe-
cifically alleged in the complaint because the attachments submitted with
the complaint raised the theory. See id. at 538. The defendants did not
challenge this action, and we only mentioned it as a recitation of the facts.
See id. In contrast, Jeney, Gentile, and Coleman gave Quaker no notice
either in the complaint, in documents submitted with the complaint, or in
any document prior to their motions for summary judgment that they
intended to argue this theory.

                               14953


complaints before the time specified in the scheduling order
expired. See id. at 608-09. This standard "primarily considers
the diligence of the party seeking the amendment. " Id. at 609.

[16] Jeney, Gentile, and Coleman did not move to amend
their complaints to add the disparate impact theory of liability
until their reply to Quaker's motion for summary judgment.
Despite having hired a statistical expert years before the sum-
mary judgment motions and having received the first statisti-
cal report noting the disparities between the retention of older
and younger employees over a year before filing for summary
judgment, the employees had never moved to amend their
complaints. In fact, Jeney and Gentile had the statistical report
before they amended their complaint to add the Title VII alle-
gations. Jeney, Gentile, and Coleman do not offer any expla-
nation for their failure to amend their complaints earlier. See
Acri v. International Ass'n of Machinists and Aerospace
Workers, 781 F.2d 1393, 1398 (9th Cir. 1986) (stating that
even under the liberal Rule 15 standard "late amendments to
assert new theories are not reviewed favorably when the facts
and the theory have been known to the party seeking amend-
ment since the inception of the cause of action."). Moreover,
granting the request to amend the complaint would likely
have required reopening discovery so that Quaker could
develop its evidence to prepare its defenses to this theory. "A
need to reopen discovery and therefore delay the proceedings
supports a district court's finding of prejudice from a delayed
motion to amend the complaint." Lockheed Martin Corp. v.
Network Solutions, Inc., 194 F.3d 980, 986 (9th Cir. 1999).
This prejudice to Quaker, although not required under Rule
16(b), supplies an additional reason for denying the motion.
See Johnson, 975 F.2d at 609. Because Jeney, Gentile, and
Coleman have failed to show diligence, "the inquiry should
end." Id. Thus, the district court did not abuse its discretion
when it denied their motions to amend their complaints.

                               14954


V

A

In Jeney's and Gentile's original discrimination claims with
the EEOC, they charged only that they had been discriminated
against because of their age. During discovery, they came
upon a document entitled "Workforce 2001" describing
Quaker's goals for a workforce diverse in terms of race and
sex. Jeney and Gentile then filed an additional complaint with
the EEOC charging that they had been discriminated against
because they were white males and filed a second amended
complaint alleging that Quaker had violated 29 U.S.C.
S 2000e ("Title VII").9 The district court allowed this amend-
ment, but then granted summary judgment to Quaker on this
issue on the grounds that their allegation was untimely. Alter-
natively, the judge ruled that Jeney and Gentile could not
make out a prima facie case of race or sex discrimination.

B

The analysis under Title VII is the same as that under
ADEA. See Wallis, 26 F.3d at 888. Thus, because Jeney and
Gentile were fired in the context of the RIF, they must pro-
duce statistical, direct or circumstantial evidence that their
"discharge occurred under circumstances giving rise to an
inference of" discrimination on account of race or sex. Id. at
891. Despite the fact that the standard is very low, they can
not meet it. In contrast to their age discrimination claims,
Jeney and Gentile have no statistics that demonstrate any sort
of disparate treatment in layoffs on the basis of race or sex.
Of the six positions filled in Phoenix for which Jeney might
have been considered, all six went to other white employees,
five of whom were men. Of the ten employees retained by
_________________________________________________________________
9 Coleman never filed this additional charge with the EEOC or moved
to amend his complaint to allege Title VII violations.

                               14955


Quaker and mentioned by Gentile as his competition, nine
were white and eight were men.

To support an inference of discrimination, Jeney and Gen-
tile point only to the existence of the Workforce 2001 pro-
gram, with its stated goal of increasing diversity in
management, and to the fact that the RIF was monitored to
determine whether it had any impact on women or minorities.
Quaker states that it never acted pursuant to the affirmative
action plan or the monitoring during the RIF. Jeney and Gen-
tile point to no evidence showing that they were affected by
the Workforce 2001 plan or the monitoring. Furthermore, they
do not point to any evidence showing that Quaker took any
action in response to its monitoring of the RIF."[T]he mere
fact of an affirmative action plan's existence is not relevant to
proving discrimination unless the employer acted pursuant to
the plan." Cerrato v. San Francisco Community College Dist.,
26 F.3d 968, 976 (9th Cir. 1994). Because Jeney and Gentile
have not raised a question of material fact as to whether they
were discriminated against on the basis of their age and sex,
the district court did not err when it granted summary judg-
ment in favor of Quaker on this claim.10

VI

Jeney, Gentile, and Coleman argue that the district court
improperly severed their original case, along with seven other
plaintiffs, against Quaker. Under Federal Rule of Civil Proce-
dure 20, joinder is proper if (1) the plaintiffs asserted a right
to relief arising out of the same transaction and occurrence
and (2) some question of law or fact common to all the plain-
tiffs will arise in the action. See Fed. R. Civ. P. 20(a); Desert
Empire Bank v. Insurance Co. of N. Am., 623 F.2d 1371, 1375
_________________________________________________________________
10 Because we conclude that Jeney and Gentile have not made out a
prima facie case of discrimination of the basis of their race or sex, we do
not reach the issue of whether they filed their EEOC charges in a timely
fashion.

                               14956


(9th Cir. 1980). Even once these requirements are met, a dis-
trict court must examine whether permissive joinder would
"comport with the principles of fundamental fairness" or
would result in prejudice to either side. Desert Bank, 623 F.2d
at 1375. Under Rule 20(b), the district court may sever the
trial in order to avoid prejudice. See Fed. R. Civ. P. 20(b).

Ten plaintiffs joined in the first complaint. All ten were
fired in the RIFs and all ten alleged age discrimination. Con-
sidering Quaker's motion for severance, the district court first
found that the plaintiffs' case met both prongs of Rule 20(a).
The court then examined the potential prejudice to Quaker
resulting from joinder and concluded that severance was
appropriate. The district court found that Quaker would be
prejudiced by having all ten plaintiffs testify in one trial. See
Moorhouse v. Boeing Co., 501 F. Supp. 390, 393 n.4 (E.D.
Pa.) (stating that "even the strongest jury instructions could
not have dulled the impact of a parade of witnesses, each
recounting his contention that defendant laid him off because
of his age"), aff'd, 639 F.2d 774 (3d Cir. 1980). In addition,
the district court emphasized the danger of jury confusion. For
each plaintiff, the jury would have had to examine individu-
ally his or her employment history as well as the explanations
given by Quaker for not retaining him or her, explanations
that would require the testimony of each employee's supervi-
sors and raters. Legal confusion was also likely because the
plaintiffs had worked for Quaker in six different states, and
the jury would have had to evaluate their state law claims in
light of the different laws of each state. The court found that
the likelihood of prejudice and confusion outweighed the
gains from judicial economy and any potential prejudice to
the plaintiffs who would try their cases in the states in which
they lived and worked.

Jeney, Gentile, and Coleman argue now that the district
court abused its discretion because severing the cases led to
duplicative discovery. As noted above, the court weighed the
interests of judicial efficiency and found them outweighed by

                               14957


the potential prejudice to Quaker. Discovery in this case was
not as duplicative as it might first appear. Jeney, Gentile, and
Coleman, whose cases remained before this district court in
Arizona, benefitted from common discovery and counsel,
completely coordinating their cases. The district court prop-
erly considered the potential prejudice to Quaker created by
the parade of terminated employees and the possibility of fac-
tual and legal confusion on the part of the jury. Given the
broad discretion with which the district court is vested to
make a decision granting severance and the fact that the dis-
trict court carefully weighed the arguments in favor of and
against joinder, we conclude that the district court did not
abuse its discretion when it granted Quaker's motion to sever
the cases.

VII

Jeney, Gentile, and Coleman contend that the district court
abused its discretion when it refused to sanction Quaker for
its delays in producing documents. We will not disturb the
district court's decisions with respect to sanctions "unless we
have a definite and firm conviction that the court committed
a clear error of judgment in the conclusion it reached upon a
weighing of the relevant factors." Payne v. Exxon Corp., 121
F.3d 503, 507 (9th Cir. 1997). We have reviewed the record
and fully considered the employees' argument in this regard,
as well as Judge Broomfield's conclusions. In addition to the
six specific rulings appealed by the employees, Judge Broom-
field considered many motions by the employees to compel
evidence and to sanction Quaker. His orders and the corre-
spondence between the attorneys reveal this to have been an
acrimonious discovery process. At times Judge Broomfield
granted the employees' requests for attorney's fees as sanc-
tions, more often he did not and ordered the parties to cooper-
ate with each other. Throughout these proceedings, he
properly considered the legal arguments made by each side
and the degree of cooperation they exhibited. We simply can-
not say that the court made "a clear error of judgment in . . .

                               14958


weighing . . . the relevant factors." Payne , 121 F.3d at 507.
The district court did not abuse its discretion when it denied
the employees' motions for sanctions.

VIII

Under ADEA, Title VII, and their state breach of contract
claim, Jeney, Gentile, and Coleman might be entitled to attor-
neys fees had they prevailed. See 29 U.S.C.SS 216(b), 626;
42 U.S.C. S 2000e-5(k); Ariz. Rev. Stat. S 12-341.01. As we
have affirmed the district court's grant of summary judgment,
the employees have not prevailed and thus are not entitled to
attorney's fees.11

IX

Because Jeney, Gentile and Coleman have failed to raise a
question of material fact as to whether Quaker discriminated
against them on the basis of age, race or sex, the district
court's grant of summary judgment in favor of Quaker is

AFFIRMED.12

_________________________________________________________________

B. FLETCHER, Circuit Judge, Dissenting:

I respectfully dissent. The Quaker Oats Company
("Quaker") embarked on a major restructuring of its work
force involving considerable downsizing. It sought consulta-
_________________________________________________________________
11 Thus, we do not reach the employees' argument that obtaining a rever-
sal of summary judgment, the only relief they sought in this appeal, would
have entitled them to such fees.
12 Jeney's, Gentile's, and Coleman's state law claims for breach of con-
tract and discharge in violation of public policy necessarily fail following
summary judgment on the federal claims. Thus we need not reach the
issue of whether Quaker's Code of Ethics modified the at will nature of
the plaintiffs' employment.

                               14959


tion to prevent gender and racial discrimination in the process,
but none to prevent age discrimination. Nobody denies that
the statistics in this case show that the reduction in force ter-
minations seriously impacted the older employees -- much
more so than the younger employees.1 So regardless of intent,
the real issue in this case is whether the process had a dispa-
rate impact on the older protected members of the work force.

The majority blinds itself to our own circuit precedent and
to what all the parties knew this case was about in refusing to
look at disparate impact. For this reason, I must dissent.

To reach its result -- finding disparate impact not suffi-
ciently or timely pled -- the panel adopts a rule that is incon-
sistent with our own prior precedent regarding the Age
Discrimination in Employment Act (ADEA). Yet, even under
the erroneous rule the panel adopts, the plaintiffs in this case
have stated a discrimination claim using the disparate impact
theory and the district judge should be required to analyze
whether the evidence is sufficient to go to trial on that theory.

In Equal Employment Opportunity Commission v. Local
350, Plumbers and Pipefitters, we held that plaintiffs may use
the disparate impact theory to prove discrimination under the
ADEA. See EEOC v. Local 350, Plumbers and Pipefitters,
998 F.2d 641, 648 n. 2 (9th Cir. 1993) (as amended on recon-
sideration). Hazen Paper Co. v. Biggins, decided while EEOC
v. Local 350 was pending, explicitly left that issue open. See
Hazen Paper Co. v. Biggins, 507 U.S. 604 (1993). In EEOC
v. Local 350, we further held that a plaintiff need not use the
words "disparate impact" in order to state a claim in the com-
plaint that is cognizable using such a theory.2  See EEOC v.
_________________________________________________________________
1 See majority opinion at 14932-33, providing examples of the findings
and characterizing the statistical disparities as "striking"; id. at 14936,
explaining that the plaintiffs' statistics demonstrated that employees over
40 received significantly lower rankings on average than those under 40.
2 This holding was foreshadowed by  Cotton v. City of Alameda, 812
F.2d 1245, 1247 (9th Cir. 1987), where we allowed a liberal construction

                               14960


Local 350, 998 F.2d at 648 n. 2. Rather, we held that a com-
plaint that identifies a company policy that falls more harshly
on one group than another is "cognizable as a disparate
impact challenge." Id. Examining the pleadings in that case,
we said "[t]he EEOC does not specify in its pleadings or
briefs on appeal whether it is proceeding under a disparate
treatment or disparate impact theory of discrimination.
Because in this circuit a plaintiff may challenge age discrimi-
nation under a disparate impact analysis, both theories were
potentially available to the EEOC." Id.

In Arnett v. California Public Employees Retirement Sys-
tem, 179 F.3d 690, 697 (9th Cir. 1999), vacated and
remanded on other grounds, _______ U.S. _______, 120 S.Ct. 930
(2000), we reaffirmed that plaintiffs may bring suit using a
"disparate impact" theory under the ADEA. 3 We were asked
_________________________________________________________________
of an ADEA claim as both a disparate treatment and a disparate impact
claim, and by Palmer v. United States, 794 F.2d 534, 538 (9th Cir. 1986),
where we allowed a claim that "did not specifically allege discrimination
under the disparate impact theory" to go forward based on the notice pro-
vided by attachments to the complaint. In Palmer , the district court had
determined that statistical studies attached to the complaint that strongly
suggested a discriminatory impact constituted notice of an impact claim.
Id.
Similarly, in the Title VII context, we have construed complaints that
simply challenged discriminatory policies and practices without specifying
the theory under which they were proceeding as disparate impact claims.
Golden v. Local 55 of the International Association of Firefighters, 633
F.2d 817, 821 (9th Cir. 1980).
3 We reaffirmed the Arnett  holding in a recent opinion. In Frank v.
United Airlines, Inc., 216 F.3d 845, 856 (9th Cir. 2000), we explained that
the Arnett opinion "squarely decided that a disparate impact claim is cog-
nizable in an ADEA case." We affirmed that this holding was not rejected
by the Supreme Court when it vacated Arnett in light of Kimel v. Florida
Bd. of Regents, _______ U.S. _______, 120 S.Ct. 631 (2000). The Kimel Court held
that ADEA does not abrogate the Eleventh Amendment. As we explained
in Frank, "The Eleventh Amendment issue is irrelevant to a case, such as
this one, in which a private rather than state entity is a defendant, and the
Court's vacation of our decision has no bearing on the correctness of our
conclusions that a disparate impact claim is cognizable under the ADEA.
We see no reason to depart from our conclusion in Arnett and we again
hold that a disparate impact claim is cognizable under the ADEA." Frank,
216 F.3d at 856.

                               14961


whether an individual who challenges an employment policy
whose consequences "fall more harshly" on those in the pro-
tected class states a claim of disparate impact under the
ADEA; we answered in the affirmative. Id. at 692. In Arnett,
we reinstated a disparate impact claim after the district court
granted a motion to dismiss the plaintiffs' complaint because
we found plaintiffs' allegations sufficient to survive dismissal
at the pleadings stage. Id. at 693. While plaintiffs had
amended their complaint to state the impact claim explicitly,
we focused on the adequacy of the pleading as a whole, not
on the use of the words "disparate impact." Id.

Rather than confronting and relying on our own precedent,
the majority turns to a third circuit case, Josey v. Hollings-
worth Corporation, 996 F.2d 632 (3d Cir. 1993), which has
facts very different from those in our case. In Josey, the Third
Circuit held that the plaintiffs had not provided notice of their
disparate impact claim when they raised it only after the close
of discovery. See id. at 641-2. However, in that case the origi-
nal complaint did not identify a specific employment practice
that had a disparate impact. See id. at 642. Here, the plaintiffs'
claims are grounded in a challenge to the termination policies
of Quaker in the context of its downsizing; the claims were
therefore always directed at the impact of a facially neutral
policy. Furthermore, in this case, unlike in Josey, the plain-
tiffs did give notice, both in the wording of the complaint and
by the discovery they conducted, including the resulting sta-
tistical evidence they put forward.

Even under the erroneous standard that the majority adopts,
the plaintiffs in this case sufficiently stated a claim of dispa-
rate impact in their complaints. The inartful language in the
complaints uses neither "disparate treatment" nor "disparate
impact" to describe the allegations against Quaker. Instead,
they use the phrase "discrimination under the ADEA" to
describe their claims. Since a plaintiff may prove discrimina-
tion using either theory, the complaint was sufficient to state
both a disparate treatment and a disparate impact claim. Fur-

                               14962


ther, the complaint language that described Quaker's "prac-
tices," "policies" and "procedures" as discriminatory applies
more aptly to impact than to treatment. No amendment of the
complaint was required to allow the plaintiffs to prove their
claim using a disparate impact approach. The district court
considered the disparate treatment claims based on this
generic language; it likewise should have considered the dis-
parate impact claims.

Furthermore, the majority's argument that Quaker did not
have notice of the disparate impact claims fails to recognize
that all of the evidence of impact was in the company's hands
and under its control. The discovery of the statistical evidence
that suggests disparate impact was extensive and gave suffi-
cient notice to Quaker of the disparate impact approach. This
is especially true because statistics, while admissible to prove
a disparate treatment claim, usually are unnecessary. Dispa-
rate impact claims, on the other hand, rely heavily and usually
necessarily on statistics. In short, the extensive statistical dis-
covery in this case surely placed Quaker on notice of
plaintiffs' impact theory. Any claim to the contrary is disin-
genuous.

Even if we were to take a very narrow view and conclude
that there was no notice to Quaker of the disparate impact
claim until the summer of 1998 when that claim was fully
briefed and plaintiffs moved for summary judgment on the
claim, that was time enough. All of the relevant data was in
Quaker's possession. Considerable time remained for Quaker
to prepare its defense. No trial date had yet been set. The
judge took nine months to determine the outcome of the sum-
mary judgment motions.

The majority errs in adopting a standard that conflicts with
our prior precedent. Even under the standard adopted by the
majority, we should remand each of these cases to the district
court to analyze whether the evidence is adequate to go to
trial on the impact theory.

                               14963