United States Court of Appeals
                             FOR THE EIGHTH CIRCUIT
                                ________________

                                   No. 04-3000
                                ________________

Carla Rodgers,                            *
                                          *
             Appellant,                   *
                                          *      Appeal from the United States
      v.                                  *      District Court for the
                                          *      Western District of Missouri.
U.S. Bank, N.A.,                          *
                                          *
             Appellee.                    *

                                ________________

                          Submitted: April 15, 2005
                              Filed: August 5, 2005 (corrected August 11, 2005)
                               ________________

Before MELLOY, COLLOTON and GRUENDER, Circuit Judges.
                       ________________

GRUENDER, Circuit Judge.

       Carla Rodgers (“Rodgers”) appeals the decision of the district court1 granting
summary judgment to U.S. Bank, N.A. (“U.S. Bank”), on her claims of employment
discrimination based on race under 42 U.S.C. § 1981, Title VII of the Civil Rights
Act of 1964, and the Missouri Human Rights Act. We agree with Rodgers that she
established a prima facie case of racial discrimination. However, because we


      1
       The Honorable Dean Whipple, United States District Judge for the Western
District of Missouri.
conclude that Rodgers failed to raise a triable question of material fact as to whether
U.S. Bank’s legitimate nondiscriminatory reason for her termination was a pretext for
discrimination, we affirm the district court.

I.    BACKGROUND

       Rodgers, an African-American woman, was employed at the Picture Hills
Branch of U.S. Bank as a drive-through teller for approximately eleven months in
2002. When Rodgers joined U.S. Bank, she received the employee handbook and
attended three days of teller training. The employee handbook contains a number of
policies, including U.S. Bank’s policy prohibiting employees from processing
transactions on their own accounts and its policy on progressive discipline.2 Rodgers
was trained on the proper way to do a “multi” transaction–a type of transaction in
which a teller processes a combination of transactions for a customer. She also was
trained on the bank’s policy against employees processing transactions on their own
accounts.

      In November 2002, a suspicious-activity report generated by U.S. Bank’s Loss
Prevention Group triggered an investigation into Rodgers’s account activity.
Suspicious-activity reports identify transactions where the last name of the customer
matches the last name of the employee performing the transaction. The reports are
then reviewed and analyzed by loss-prevention analysts in the Employee Fraud
Detection Group.

      Rodgers’s suspicious-activity report showed that she had processed five
transactions on one of her personal bank accounts involving credits of more than


      2
      The progressive-discipline policy provides for progressive discipline “when
appropriate.” It also provides that “[i]n some cases, U.S. Bancorp may determine
immediate termination is warranted.”
                                          -2-
$35,000. Many of the credits appeared to come from CD/IRA or Loan in Process
accounts. Rodgers did not have any of these accounts and, as a drive-through teller,
never processed transactions on these types of accounts. A subsequent investigation
revealed four additional suspicious transactions, including a credit to her account in
the amount of $620,003.14. Because there was no paper support for these
transactions, they washed out of the system at the end of each day, and Rodgers never
tried to remove money from her account while credits were posted there.

       A loss-prevention analyst contacted Beth Money, East Region District
Manager, and Jan Scaletta, Human Resources Manager, and asked that they follow
up on the report. Money and Scaletta passed along the information to Pathe Price,
corporate security investigator for U.S. Bank. Price, along with Vickie Gabbert,
District Operations Manager, conducted the investigation.

       Price and Gabbert interviewed Rodgers as part of their investigation. Rodgers
explained that she had used her account number to get out of a multi-transaction when
she entered the function accidentally and continued to enter data. Although there are
other ways to get out of the multi-transaction function, Rodgers used her account
number to “zero out” the transaction. Rodgers further explained that she did not
believe there was anything wrong with this procedure because she had been taught
the “trick” by a senior teller, Kathy Nichols. Although she offered to show them the
mistakes on the keyboard, neither investigator had her do so.

       Price and Gabbert did not find Rodgers’s explanation credible for several
reasons. First, the multi-transaction function could not be entered in the manner
Rodgers described. Rodgers said that she had entered the multi-transaction function
accidentally by hitting the enter key twice, but the function can only be entered by
hitting another particular key on the keyboard. The investigators also had difficulty
locating a corresponding transaction which Rodgers would have zeroed out with the
transactions she processed on her account. In addition, several of the transactions

                                         -3-
Rodgers processed, such as CDs and loans, were not normally handled by drive-
through tellers. There were other suspicious circumstances, including the presence
of an overdraft-fee waiver on Rodgers’s account, which was not permitted, and the
fact that Rodgers had repeatedly checked her account balance on one of the days on
which she processed some of the erroneous transactions.3 Price formed the opinion
that Rodgers was testing the bank’s computer system by putting money into her
account and seeing if anything would “stick.”

       Price and Gabbert reported their findings and opinions to Jennifer Crim, the
branch manager. Crim discussed the findings with her manager, Money. Based on
the information obtained during the investigation, Money recommended that Rodgers
be terminated. Crim agreed with Money’s recommendation because she, like Price
and Gabbert, did not find Rodgers’s explanation to be credible. As a result, Crim
terminated Rodgers’s employment because she believed the risk associated with the
suspicious transactions was too great to allow Rodgers to continue her employment.

       The investigators did not interview Nichols because she was on vacation during
U.S. Bank’s investigation of Rodgers. However, as a result of Rodgers’s explanation
that she had learned from Nichols that she could use her account number to “zero out”
a transaction, U.S. Bank conducted a follow-up investigation of Nichols. Nichols had
been a teller at U.S. Bank and its predecessors for more than seventeen years. Shortly
before the investigation of Rodgers, Nichols had been placed on written “final notice”
for violating a U.S. Bank dual-control security policy which prohibits one person




      3
        Rodgers asserts that the most likely reason for checking her balance the first
time that day was to see if her paycheck had been deposited, and the reason for the
other times she checked was probably to joke with the other tellers about her large
account balance because of the back-out transactions.
                                         -4-
from having both the key and the combination to a vault.4

       The Employee Fraud Detection Group investigated Nichols’s account activity
to determine whether she had used her own account number for any transactions.
Investigators found one such transaction. However, in contrast to Rodgers’s account
activity, they were able to determine how Nichols had gotten to the point where she
found it necessary to use her account number and the specific transaction she was
attempting to reverse. Additionally, Nichols immediately reversed the transaction on
her own account rather than allowing it to wash out at the end of the day. The branch
manager found Nichols’s explanation to be credible and did not discipline her.

       Rodgers filed a complaint against U.S. Bank alleging that her termination
constituted racial discrimination in violation of 42 U.S.C. § 1981, Title VII of the
Civil Rights Act of 1964, and the Missouri Human Rights Act. Specifically, she
alleged that she suffered disparate treatment based on race when she was terminated
for violating the same U.S. Bank policy as Nichols, a white employee who was not
disciplined for violating the policy. U.S. Bank moved for summary judgment. The
district court granted U.S. Bank’s motion, concluding that Rodgers did not establish
a prima facie case of racial discrimination. Rodgers now appeals.




      4
        Nichols was placed on final notice for a period of ninety days. The notice
indicated that during the 90-day period, Nichols was expected to meet the following
conditions: (1) no further violations of teller and security procedures; and (2) all teller
duties and other areas of performance must meet standards. The notice stated that
“[f]ailure to meet these expectations at anytime within the 90-day period may result
in further disciplinary action up to and including termination.”
                                            -5-
II.   DISCUSSION

      A.     Introduction

       We review the district court’s grant of summary judgment to U.S. Bank de
novo. Pope v. ESA Services, Inc., 

, 1006 (8th Cir. 2005). “Summary
judgment is appropriate only when there is no genuine issue of material fact and the
moving party is entitled to judgment as a matter of law.” Randolph v. Rodgers, 

, 856 (8th Cir. 1999). On summary judgment, “[t]he burden of
demonstrating that there are no genuine issues of material fact rests on the moving
party.” Winthrop Resources Corp. v. Eaton Hydraulics, Inc., 

, 468 (8th
Cir. 2004). The non-moving party, however, must still “present[] evidence
sufficiently supporting the disputed material facts that a reasonable jury could return
a verdict in [her] favor.” Gregory v. City of Rogers, Ark., 

, 1010 (8th
Cir. 1992) (en banc). “[W]e must view the evidence and the inferences that may be
reasonably drawn from the evidence in the light most favorable” to Rodgers, the non-
moving party. Winthrop 

. While summary judgment
“should seldom be utilized” in employment discrimination cases, Stidham v.
Minnesota Mining & Mfg., Inc., 

, 937 (8th Cir. 2005), “there is no
‘discrimination case exception’ to the application of Fed. R. Civ. P. 56, and it remains
a useful pretrial tool to determine whether or not any case, including one alleging
discrimination, merits a trial.” Berg v. Norand Corp., 

, 1144 (8th Cir.
1999).

       In this case, we employ the familiar McDonnell Douglas burden-shifting
framework. Turner v. Honeywell Fed. Mfg. & Techs., LLC, 

, 720 (8th
Cir. 2003) (citing McDonnell Douglas Corp. v. Green, 

, 801-04 (1973)).
“Under the McDonnell Douglas framework, a presumption of discrimination is
created when the plaintiff meets [her] burden of establishing a prima facie case of
employment discrimination. A minimal evidentiary showing will satisfy this burden

                                          -6-
of production.” 

(citations omitted). Once the plaintiff
establishes a prima facie case of discrimination, “[i]t then falls to the employer to
promulgate a non-discriminatory, legitimate justification for its conduct, which rebuts
the employee’s prima facie case.” Sprenger v. Fed. Home Loan Bank of Des Moines,

, 1111 (8th Cir. 2001) (citing St. Mary’s Honor Center v. Hicks, 

, 507 (1993)). If the employer meets its burden, “the presumption of
discrimination disappears, requiring the plaintiff to prove that the proffered
justification is merely a pretext for discrimination.” 

. The
plaintiff has the burden of persuasion at all times. 

Facie Case of Discrimination

       In order to establish a prima facie case of discrimination on the part of U.S.
Bank, Rodgers must show that: (1) she is a member of a protected group; (2) she was
qualified for her position; (3) she was discharged; and (4) the discharge occurred
under circumstances giving rise to an inference of discrimination. See Davenport v.
Riverview Gardens Sch. Dist., 

, 944-45 (8th Cir. 1994). The parties agree
that Rodgers satisfies the first three elements. Rodgers can prove the fourth element
by putting forth facts that similarly situated employees, who are not African-
American, were treated differently. Wheeler v. Aventis Pharms., 

, 857
(8th Cir. 2004).

       The district court held that Rodgers failed to prove the fourth element of her
prima facie case–that she and Nichols were similarly situated. The district court
applied a standard articulated in Gilmore v. AT&T, requiring that Rodgers and
Nichols be “similarly situated in all respects.” 

, 1046 (8th Cir. 2003)
(citing Clark v. Runyon, 

, 918 (8th Cir. 2000)). To satisfy this standard,
“[t]he individuals used as comparators ‘must have dealt with the same supervisor,
have been subject to the same standards, and engaged in the same conduct without
any mitigating or distinguishing circumstances.’” 

218 F.3d at
918).

                                          -7-
       Applying this standard, the district court held that while Rodgers and Nichols
violated the same bank policy, the degree and frequency of Rodgers’s violations
differed significantly from that of Nichols’s single violation. The district court
compared Rodgers’s nine transactions involving erroneous credits in excess of
$650,000 to Nichols’s one erroneous transaction involving $1,200. The district court
agreed with U.S. Bank that the circumstances surrounding Rodgers’s transactions
were suspicious. Accordingly, the district court held that “[i]t therefore cannot be
said that Rodgers and Nichols ‘engaged in the same conduct without any mitigating
or distinguishing circumstances.’” Rodgers v. U.S. Bank, N.A., No. 03-388-CV-W-
DW, slip op. at 6 (W.D. Mo. June 16, 2004) (quoting 

).
Consequently, the district court held that Rodgers failed to establish her prima facie
case because Rodgers and Nichols were not similarly situated. Rodgers argues that
the district court erred in concluding that she failed to establish a prima facie case of
racial discrimination.

       There appear to be conflicting lines of cases in our Circuit regarding the
standard for determining whether employees are similarly situated at the prima facie
stage of the McDonnell Douglas burden-shifting framework. One line sets a “low
threshold” for employees to be considered similarly situated, requiring only that the
employees “are involved in or accused of the same or similar conduct and are
disciplined in different ways.” See 

(citing Williams v. Ford
Motor Co., 

, 1308-09 (8th Cir. 1994), the Court reasoned that at the
prima facie stage the test for determining whether employees are similarly situated
“is ‘not onerous,’ however at the third stage (proving pretext) it is ‘rigorous’”); see
also 

n.3 (noting that the meaning of “similarly situated”
varies, depending on whether it is applied at the prima facie stage or pretext stage).
The other line, upon which the district court relied, is more rigorous at the prima facie
stage and requires that the employees must be “similarly situated in all respects.” See




                                           -8-


(citing 

).5

      5
        Gilmore relied on Clark v. Runyon. See 

(citing

). However, Clark’s use of a more rigorous standard at the
prima facie stage appears misplaced. For example, Clark relied on Harvey v.
Anheuser-Busch, Inc., 

(8th Cir. 1994), and Lynn v. Deaconess Medical
Center, 

(8th Cir. 1998). See 

. Harvey and Lynn,
however, used the more rigorous “similarly situated in all relevant respects” standard
at the pretext stage, not the prima facie stage. See 

; 

.

        We respectfully disagree with the concurring opinion’s position that the
definitions this Circuit has used for the term “similarly situated” at the prima facie
stage–“involved in or accused of the same or similar conduct” and “similarly situated
in all relevant respects”–are functionally equivalent.

       For example, the Court in Williams, using the “involved in or accused of the
same or similar conduct” definition at the prima facie stage, concluded that the
plaintiff, an African-American, established his prima facie case simply because he
provided evidence that he and nine white employees violated the employer’s “five
day letter” rule, yet he was treated differently because he was refused reinstatement.

. The employer’s “five day letter” rule provided that once
an employee’s medical leave expired, the employee received a letter directing him
either to report to work or to provide medical evidence of the need for an extension
within five days of receiving the letter. Failure to respond within the five-day period
resulted in immediate termination. However, the employee could apply for a
reinstatement waiver, which, if granted, would allow the employee to return to work
on probation. 

       In stark contrast, using the “similarly situated in all relevant respects”
definition at the pretext stage, the Court in Williams concluded that the plaintiff failed
to prove pretext because he:

      presented no evidence regarding the prior work histories of the reinstated
      Caucasian employees as compared to himself . . ., the respective excuses
      for violating the “five day letter” rule, the number of days they remained
      absent before returning to work, the nature or severity of their medical
                                           -9-
       When there are conflicting lines of cases in our Circuit, we can choose which
line to follow. See Kostelec v. State Farm Fire and Cas. Co., 

, 1228 n.8
(8th Cir. 1995) (noting that while we are not at liberty to overrule an opinion filed by
another panel, we are free to choose which line of cases to follow); United States v.
Anderson, 

, 570 n.1 (8th Cir. 2003).

       At the prima facie stage of the McDonnell Douglas burden-shifting framework,
we choose to follow the low-threshold standard for determining whether employees
are similarly situated. We believe this more accurately reflects Supreme Court
precedent. The Supreme Court has advised: “The burden of establishing a prima
facie case of disparate treatment is not onerous.” Texas Dep’t of Cmty. Affairs v.
Burdine, 

, 253 (1981). We also think that using the low-threshold
standard at the prima facie stage is more in line with our Court’s prior admonition that
“[t]he district court must not conflate the prima facie case with the ultimate issue of
discrimination.” 

. Using a more rigorous standard at the
prima facie stage would “conflate the prima facie case with the ultimate issue of
discrimination,” thereby effectively eliminating the burden-shifting framework the
Supreme Court has directed us to use.

      Under the low-threshold standard, Rodgers must show that she and Nichols
were “involved in or accused of the same or similar conduct and [were] disciplined


     conditions, or the extent to which their medical conditions were
     corroborated by independent, medical documentation. Moreover, the
     evidence in this case indicates that [the plaintiff] may have falsified his
     medical records, a circumstance not shown by [the plaintiff] to exist in
     other cases of reinstatement of Caucasian persons.

As exemplified in Williams, the different results obtained at the prima
facie stage and the pretext stage demonstrate that the “involved in or accused of the
same or similar conduct” definition and the “similarly situated in all relevant
respects” definition are not functionally equivalent.
                                          -10-
in different ways.” 

(quoting 

).
Rodgers satisfies this burden. Rodgers and Nichols are similarly situated because
they both violated the same bank policy by processing transactions on their own
accounts, and they were disciplined differently because Rodgers was terminated and
Nichols received no punishment. As the district court points out, there are differences
in the severity and frequency of their violations and the surrounding circumstances;
however, these differences are relevant to the issue of whether Rodgers and Nichols
are similarly situated for purposes of showing pretext. Accordingly, we hold that
Rodgers has established a prima facie case of disparate treatment.

       The burden of production now shifts to U.S. Bank to show that it had a
legitimate, nondiscriminatory reason for terminating Rodgers. U.S. Bank’s proffered
reason is that Rodgers violated the bank’s policy against employees processing
transactions on their own accounts. “The burden to articulate a nondiscriminatory
justification is not onerous, and the explanation need not be demonstrated by a
preponderance of the evidence.” Floyd v. State of Missouri Dept. of Soc. Servs., Div.
of Family Servs., 

, 936 (8th Cir. 1999). Accordingly, the ultimate
burden falls on Rodgers to produce evidence sufficient to create a genuine issue of
material fact regarding whether U.S. Bank’s proffered nondiscriminatory reason is
a pretext for discrimination.

      C.     Pretext

       Rodgers attempts to prove pretext with evidence of the following: (1) disparate
treatment of her and similarly situated white employees; (2) an inadequate
investigation into her account activity; and (3) a substantial change in U.S. Bank’s
legitimate non-discriminatory reason for her termination. We affirm the district
court’s grant of summary judgment to U.S. Bank because we conclude that Rodgers
failed to raise a triable question of material fact as to whether U.S. Bank’s legitimate



                                          -11-
nondiscriminatory reason for her termination was a pretext for discrimination.6

             1.     Disparate treatment

             Rodgers attempts to prove disparate treatment by demonstrating that she
was treated less favorably than similarly situated employees outside of her protected
group. See E.E.O.C. v. Kohler Co., 

, 776 (8th Cir. 2003). Specifically,
Rodgers argues that she was terminated for violating the same U.S. Bank policy as
Nichols, a white employee who was not disciplined for violating the policy. She also
argues that only white employees benefitted from U.S. Bank’s progressive discipline
policy.

               Rodgers has failed to show that she was similarly situated to any of these
white employees. At the pretext stage of the McDonnell Douglas burden-shifting
framework, the test for determining whether employees are similarly situated to a
plaintiff is a rigorous one. See 

; 

.
Rodgers must show that she and the employees outside of her protected group were
similarly situated in all relevant respects. 

. “To be probative
evidence of pretext, the misconduct of more leniently disciplined employees must be
of ‘comparable seriousness.’” 

(quoting Lanear v. Safeway
Grocery, 

, 301 (8th Cir. 1988)); see also 

(concluding that plaintiff and white employee were not similarly situated because
their actions “involved differing levels of misconduct”).

             Although Rodgers and Nichols violated the same bank policy, the

      6
        We note that we “may affirm a district court’s order, including an order
granting summary judgment, on any basis supported by the record, even if that
ground was not considered by the district court.” Saulsberry v. St. Mary’s Univ. of
Minn., 

, 866 (8th Cir. 2003) (quoting Viking Supply v. Nat’l Cart Co.,
Inc., 

, 1097 (8th Cir. 2002)) (internal quotation omitted).
                                          -12-
frequency and seriousness of Rodgers’s violations and the suspicious circumstances
surrounding her violations set her apart. Rodgers violated U.S. Bank’s policy against
processing transactions on her own account nine times in ten days, resulting in over
$650,000 being credited to her account. Some of these transactions were on accounts
not normally handled by drive-through tellers, such as CDs and loans. Although
Rodgers asserted that she was using her account to correct a mistake, the investigators
were unable to ascertain from the record of her transactions what alleged mistake she
was trying to correct. Rodgers also repeatedly checked her account balance on one
of the days on which some of the erroneous transactions occurred. In addition,
Rodgers had an impermissible overdraft-fee waiver code on her account. In contrast,
Nichols’s single violation involved $1,200. She reversed the transaction
immediately. And the investigators could clearly follow on the record what mistake
Nichols was attempting to correct. Nichols’s activity did not suggest that she was
testing the system to determine whether deposits that were not reversed would stay
in her account. Rodgers’s and Nichols’s violations were not of “comparable
seriousness.”

              When Nichols violated U.S. Bank’s policy against employees processing
transactions on their own accounts, she also was on final notice for violating the
bank’s dual-control vault security policy. Rodgers argues that this should be
considered in the similarly-situated analysis in the same way a plaintiff’s employment
history is considered. See, e.g., Forrest v. Kraft Foods, Inc., 

, 691-92
(8th Cir. 2002) (holding that the plaintiff was not similarly situated when the plaintiff
had a history of similar violations and this was the other employee’s first such
violation). It is important to note that Nichols and two other employees were written
up for violating the dual-control policy. Crim testified that these dual-control policy
violations resulted from poor management and that she was not concerned about any
suspicious activity on the part of the employees involved. Even if we were to accept
Rodgers’s argument and consider Nichols’s single violation of bank policy against
employees processing transactions on their own accounts in conjunction with her

                                          -13-
final notice status, the severity of Rodgers’s violations and the suspicious surrounding
circumstances cause us to conclude that Rodgers and Nichols are not similarly
situated in all relevant respects.

              Rodgers also argues that only white employees benefitted from U.S.
Bank’s progressive discipline policy. She points to the fact that rather than being
terminated, Nichols and two other white employees at her branch were put on final
notice for violating the bank’s dual-control vault security policy. She also notes that
Crim, her branch manager who is white, was demoted and given written warnings for
poor performance and behavior issues rather than being terminated. Rodgers has
failed to show how she is similarly situated to any of these white employees. See

(“It is not up to the employer to prove dissimilarity.”).
Moreover, U.S. Bank’s policy does not require progressive discipline in all cases, and
“the employment-discrimination laws have not vested in the federal courts the
authority to sit as super-personnel departments reviewing the wisdom or fairness of
the business judgments made by employers, except to the extent that those judgments
involve intentional discrimination.” Hutson v. McDonnell Douglas Corp., 

, 781 (8th Cir. 1995).

             2.     Inadequate Investigation

              Next, Rodgers attempts to prove that U.S. Bank’s proffered justification
for terminating her is a pretext for discrimination by arguing that U.S. Bank
conducted an inadequate investigation into her account activity. She argues that Price
and Gabbert did not speak with Nichols regarding her involvement in teaching her
how to use her account number to “zero out” a transaction. Rodgers also points to the
facts that Crim did not give her the opportunity to demonstrate the typing mistakes
she asserts she made and that Crim did not contact U.S. Bank’s Human Resources
Department before terminating her.



                                          -14-
             This essentially boils down to an argument that the decision to terminate
Rodgers was based on erroneous conclusions. The issue before the Court, however,
is not whether U.S. Bank’s suspicions were correct; instead, the issue is whether U.S.
Bank conducted a thorough investigation and whether it made credibility
determinations reasonably and in good faith. See 

(citing
Euerle-Wehle v. United Parcel Serv., 

, 900 (8th Cir. 1999)). We
conclude U.S. Bank did so. Price, Gabbert, Money and Crim considered Rodgers’s
explanation but found it incredible in light of the suspicious circumstances, and
Rodgers has produced no evidence showing that they did not believe she was guilty
of misconduct. See Griffith v. City of Des Moines, 

, 738 (8th Cir. 2004);
see also Waggoner v. City of Garland, Tex., 

, 1166 (5th Cir. 1993) (“To
the extent [plaintiff’s] summary judgment evidence relates to his innocence of the
sexual harassment charge, it is irrelevant. He must, instead, produce evidence
demonstrating that [his employer] did not in good faith believe the allegations, but
relied on them in a bad faith pretext to discriminate against him . . . .”).

             3.    Change in Proffered Reason

               Finally, Rodgers attempts to prove pretext by arguing that U.S. Bank
changed its reason for terminating her. She notes that in its MCHR/EEOC response,
U.S. Bank asserted that Rodgers was terminated for violating bank policy, but later,
in litigation, the bank brought up the frequency of her violations and the suspicious
circumstances surrounding her violations. While it is true that “[s]ubstantial changes
over time in the employer’s proffered reason for its employment decision support a
finding of pretext,” Kobrin v. Univ. of Minn., 

, 703 (1994), this does not
mean that an employer cannot elaborate on its proffered reason. See Smith v. Allen
Health Systems, Inc., 

, 835 (8th Cir. 2002). U.S. Bank has not backed
off from its original reason for terminating Rodgers; rather, U.S. Bank simply points
to “additional aspects” of Rodgers’s violation of bank policy. See 

aspects are relevant to U.S. Bank’s argument that Rodgers and Nichols are

                                         -15-
not similarly situated. This is not a case of U.S. Bank changing its story.

             Accordingly, Rodgers has failed to present any material evidence calling
into question U.S. Bank’s proffered legitimate, nondiscriminatory reason for
terminating her employment.

III.   CONCLUSION

      For the foregoing reasons, we conclude that Rodgers established a prima facie
case of racial discrimination. However, because we conclude that Rodgers failed to
raise a triable question of material fact as to whether U.S. Bank’s legitimate
nondiscriminatory reason for her termination was a pretext for discrimination, we
affirm the district court.

COLLOTON, Circuit Judge, concurring in the judgment.

       I concur in the judgment of the court. It seems unnecessary, in a case like this
one, to devote extensive analysis to whether the plaintiff established the elements of
a prima facie case of race discrimination. Given that U.S. Bank proffered a
legitimate, non-discriminatory reason for discharging Carla Rodgers, and Rodgers
presented whatever evidence she had to show that the proffered reason was a pretext
for discrimination, the summary judgment record before the district court includes
everything the parties would present even assuming Rodgers did make a prima facie
case. As the Supreme Court said in the context of a case that had been through trial,
“[t]he prima facie case method established in [McDonnell Douglas Corp. v. Green,

(1973),] was never intended to be rigid, mechanized, or ritualistic,” and
“[w]here the defendant has done everything that would be required of him if the
plaintiff had properly made out a prima facie case, whether the plaintiff really did so
is no longer relevant.” U.S. Postal Service Bd. of Governors v. Aikens, 

,
715 (1983) (internal quotation omitted). I agree with the court that Rodgers has not

                                         -16-
generated a genuine issue for trial on the “ultimate question of discrimination vel
non,” 

and that ought to be enough to resolve this appeal. See, e.g., George
v. Leavitt, 

, 411-412 (D.C. Cir. 2005) (applying Aikens in summary
judgment context); Fairchild v. Forma Scientific, Inc., 

, 572 (7th Cir.
1998) (same); Wixson v. Dowagiac Nursing Home, 

, 170 (6th Cir. 1996)
(same).

       The court, however, addresses what it describes an apparent conflict in our
cases concerning the standard for determining the existence of a prima facie case in
a case of alleged unlawful disparate treatment. I respectfully disagree with its
conclusion on this point. The court’s analysis concerns the definition of “similarly
situated” employees who are offered for comparison at the prima facie stage of the
McDonnell Douglas burden-shifting analysis. The majority labels one definition of
“similarly situated” a “low-threshold standard,” while the district court’s definition
is said to be “rigorous,” and therefore incorrect. Ante, at 8-9. In my judgment, the
divergence in definitions is more apparent than real, but to the extent there is a
material difference, the standard applied by the district court is more deeply rooted
in our precedents, more consistent with the law of other circuits, and more consonant
with the legal presumption that accompanies the establishment of a prima facie case.
Therefore, I would reaffirm our use of the district court’s standard – whether the
plaintiff is “similarly situated in all relevant respects” with the employees offered for
comparison – and uphold the district court’s conclusion that Rodgers did not provide
sufficient evidence to raise an inference that her termination was more likely than not
the result of discrimination based on race.

       First, a prima facie standard that inquires whether employees are “similarly
situated in all relevant respects” is more deeply rooted in our precedents. This
standard was employed as early as 1992 in Jones v. Frank, 

(8th Cir.
1992), where we explained that Jones’s claim of unlawful disparate treatment based
on her sex “must rest on proof that she and the three men are ‘similarly situated in all

                                          -17-
relevant respects,’” 

and then held that Jones had “failed to prove a prima
facie case of sex discrimination.” 

We have equated the phrase “similarly
situated in all relevant respects” with the statement that “[e]mployees are similarly
situated when they are involved in or accused of the same offense and are disciplined
in different ways.” Wheeler v. Aventis Pharm., 

, 858 (8th Cir. 2004)
(internal quotation and emphasis omitted). This latter formulation of “similarly
situated” dates to Boner v. Board of Commissioners, 

(8th Cir. 1982),
where we held that a plaintiff “did not establish a prima facie case of discrimination”
under this standard. 

697 (quoting Rohde v. K.O. Steel Castings, Inc., 

, 322 (5th Cir. 1981)). Several more recent cases have applied the standard
of “similarly situated in all relevant respects” or its equivalent in determining the
existence of a prima facie case. Tatum v. City of Berkeley, 

, 553 (8th
Cir. 2005) (rejecting plaintiffs’ prima facie case of disparate treatment where
comparator was “not similarly situated to them in one material respect”); Gilmore v.
AT&T, 

, 1046 (8th Cir. 2003); LaCroix v. Sears, Roebuck & Co., 

, 693-94 (8th Cir. 2001); Clark v. Runyon, 

, 918 (8th Cir. 2000);
Tolen v. Ashcroft, 

, 882-83 (8th Cir. 2004) (applying Gilmore and
Clark); Marquez v. Bridgestone/Firestone, Inc., 

, 1038 (8th Cir. 2004)
(per curiam) (applying Clark); see also Philip v. Ford Motor Co., No. 04-1735, 

, at *3 (8th Cir. July 7, 2005) (Heaney, J., dissenting) (“Our cases
require a plaintiff to show that he or she is similarly situated to employees offered for
comparison in all relevant respects to meet the prima facie burden.”) (emphasis in
original).

       The court seems to trace the view that we should employ a different, “low-
threshold” definition of “similarly situated” in assessing a plaintiff’s prima facie case
to the decision in Williams v. Ford Motor Co., 

(8th Cir. 1994). I
respectfully submit that a better reading of Williams indicates no such deviation from
our precedents in Boner and Jones. Williams explicitly relied on Boner for its
definition of “similarly situated,” stating that “for purposes of establishing a prima

                                          -18-
facie case we consider whether the employees are involved in or accused of the same
or similar conduct and are disciplined in different ways. See Boner v. Board of
Comm’rs, 

, 697 (8th Cir. 1982).” 

Williams also followed
the precedent of Jones, saying that “we analyze Williams’ prima facie case according
to the elements set forth in Jones,” 

and concluding that “[u]nder the Jones
model, this court holds that Williams established a prima facie case of race
discrimination.” 

be passing strange to interpret an opinion that explicitly
applied the “Jones model” and supported its definition of “similarly situated” with a
citation of Boner as rejecting the very definitions of “similarly situated” employed
in Jones and Boner at the prima facie stage as too “rigorous” for use in determining
the existence of a prima facie case.

       I therefore respectfully disagree with the reading of Williams advanced in our
recent opinion in Wheeler v. Aventis Pharmaceuticals. Wheeler interpreted the
Williams definition of similarly situated, which was supported by the citation to
Boner, and which merely paraphrased a passage from Boner by changing the words
“same offense” to “same or similar conduct,” as establishing a “low 

. Wheeler then introduced the notion that the definition of “similarly
situated” at the pretext stage of a discrimination case is relatively “rigorous,” id.,7 and
concluded that the very passage from Boner cited in Williams at the prima facie stage
constituted a “rigorous” definition of “similarly situated” to be used only at the
pretext stage. 

       This interpretation of the Williams analysis of a prima facie case would mean
that the court cited a so-called “rigorous” definition of “similarly situated” used at the
prima facie stage in Boner, but then paraphrased the definition in order to establish


      7
        Although Wheeler placed the term “rigorous” in quotation marks and followed
it with a citation to 

, the word “rigorous” does not appear
in Williams.
                                           -19-
a “low threshold” at the prima facie stage in Williams. We should try to read our
precedents as following, rather than circumventing, our earlier decisions, and the
better reading of Williams is that it merely applied the functional equivalent of the
Jones and Boner definitions of “similarly situated,” and concluded that the plaintiff
in Williams had established a prima facie case under those legal standards. Indeed,
that is how the Eleventh Circuit has interpreted Williams. Holifield v. Reno, 

, 1562 (11th Cir. 1997) (equating Williams statement of prima facie case
standard with “similarly situated in all relevant respects”).8 It turns out, therefore,
that the second “conflicting line” of cases described by the court can be limited to one
case decided only last year, and given the option to “choose which line to follow,”
ante at 10, we would perform a better service by returning the law of the circuit to its
moorings. Even assuming the majority’s view that Williams must be read as
conflicting with Jones and Boner, ante, at 8 n.5, the preferable course would be to
follow the earlier precedents and the majority of our cases. See Kostelec v. State
Farm Ins. Co., 

, 1228 n.8 (8th Cir. 1995) (explaining that panel may
recognize that earliest panel decisions “properly control, as they should have
controlled” in subsequent cases that created conflicting lines of precedent).



      8
        Williams noted that the phrase “similarly situated” appeared in both the
opinion’s discussion of a prima facie case and in a discussion whether the employer’s
explanation for adverse action was a pretext for discrimination, and then explained
that “the meaning of ‘similarly situated’ varies according to the context in which it
is 

n.3. Williams did not, however, say that the meaning of the
term is more or less “rigorous” at either stage of the McDonnell Douglas burden-
shifting analysis. Williams may be harmonized with Jones, Boner, and our similar
precedents by recognizing that the plaintiff’s argument concerning “similarly
situated” employees differed at the two stages of the litigation. At the prima facie
stage, the plaintiff in Williams compared himself directly with other employees of a
different race. 

& n.2. At the pretext stage, he advanced what the court
called a “statistical comparison,” apparently based on the relative percentages of
employees of each race who were subjected to the same adverse action as the
plaintiff. 

                                          -20-
        Second, a prima facie standard of “similarly situated in all relevant respects”
is used widely among our sister circuit courts of appeals. A majority of the regional
circuits has applied this familiar standard or its equivalent at the prima facie stage of
litigation under Title VII. See Cooper v. Southern Co., 

, 735 (11th Cir.
2004); Sartor v. Spherion Corp., 

, 279 (7th Cir. 2004); Barbour v.
Browner, 

, 1345-47 (D.C. Cir. 1999); 

(Tatel, J. dissenting);
Ercegovich v. Goodyear Tire & Rubber Co., 

, 352-53 (6th Cir. 1998);
Martinez v. Northern Rio Arriba Elec. Coop., 

, at *1 (10th Cir. 1998)
(unpublished opinion); Shumway v. United Parcel Serv., 

, 64 (2d Cir.
1997); see also Mayberry v. Vought Aircraft Co., 

, 1090 (5th Cir. 1995)
(“To establish a prima facie case in this manner, Mayberry must show that white
employees were treated differently under circumstances ‘nearly identical’ to his.”);
Patches v. City of Phoenix, 

, at *1 (9th Cir. 2003) (unpublished);
cf. Ortiz Garcia v. Toledo Fernandez, 

, 24 (1st Cir. 2005) (per curiam).
The court cites no other decision that has adopted different definitions of the term
“similarly situated” for the purpose of setting different thresholds at different stages
of a discrimination case, and I have not found any precedent (other than Wheeler) for
the approach endorsed by the court today.

        Third, the standard of “similarly situated in all relevant respects” also furthers
the purposes of the prima facie case in the burden-shifting model of McDonnell
Douglas. While the burden of establishing a prima facie case is not “onerous,” Texas
Dep’t of Cmty. Affairs v. Burdine, 

, 253 (1981), the proof of a prima
facie case has important legal consequences. It establishes a legal presumption that
the employer unlawfully discriminated against the employee, because the alleged acts,
“if otherwise unexplained, are more likely than not based on impermissible factors.”

(internal quotation omitted). If the employer remains silent in the face of
a prima facie case that is proved by a preponderance of the evidence, then the
employee is entitled to judgment as a matter of law. St. Mary’s Honor Center v.
Hicks, 

, 510 n.3 (1993).

                                           -21-
       Under that framework, it makes good sense to consider whether an employee
who seeks to make a prima facie case of disparate treatment by comparisons with
another employee of a different race has demonstrated that the comparators are
similarly situated in all relevant respects. For if they are not – that is, if the
employees offered for comparison are dissimilar in respects that are relevant to the
disputed employment action – then it is likely that the relevant dissimilarities explain
the disparate treatment. In that case, the evidence has not “eliminate[d] the most
common nondiscriminatory reasons” for the employer’s action, and it does not justify
the legal presumption that the employer’s acts “are more likely than not based on
impermissible factors.” 

(internal quotation omitted).

       This is not to say that employees offered for comparison must be similarly
situated in all respects. If the evidence supports an inference that a particular
dissimilarity is not relevant to the disputed act of the employer, then the presence of
that dissimilarity will not preclude the employee from establishing a prima facie case.
See 

. But it is difficult to see why there is anything
unduly rigorous about requiring a plaintiff to show that a jury could find that the
employees offered for comparison are similarly situated in those respects that are
relevant to the dispute, before there arises a legal presumption that the employer acted
based on an unlawful discriminatory motive.9



      9
        Of course, a discharged employee need not rely on comparisons with similarly
situated employees to prove unlawful discrimination. For example, to make a prima
facie case under the McDonnell Douglas framework, the employee may produce
evidence that her position remained open after the discharge and ultimately was filled
by a person of a different race. See 

. Or an employee could
attempt to prove race discrimination through direct evidence in the form of actions
or remarks by the employer that reflect discriminatory intent. See 

n.4. Rodgers, however, argues only that a comparison of her termination with
the employer’s treatment of another employee establishes a prima facie case and
intentional race discrimination.
                                          -22-
      For essentially the reasons discussed by the court in Part II.C.1 of its opinion,
I agree with the district court that Rodgers did not establish a prima facie case of
unlawful discrimination. Accepting for the sake of argument the court’s view that she
did make a prima facie case, I agree that Rodgers did not raise a genuine issue of
material fact as to whether U.S. Bank’s stated reasons for her termination were a
pretext for discrimination. Therefore, I concur in the judgment.
                        ______________________________




                                         -23-