Bryant v. Farmers Insurance Exchange

F I L E D United States Court of Appeals Tenth Circuit December 20, 2005 PUBLISH UNITED STATES COURT OF APPEALS Clerk of Court TENTH CIRCUIT JUDITH A. BRYANT, Plaintiff - Appellant, v. No. 03-3234 FARMERS INSURANCE EXCHANGE, Defendant - Appellee. Appeal from the United States District Court for the District of Kansas (D.C. No. 01-2390-CM) Michael J. Gallagher, Gallagher & Kaiser, LLP, Kansas City, Missouri, for Plaintiff-Appellant. James R. Holland II, (Shelley Schiebel Patterson with him on the briefs), Bioff Finucane Coffey Holland & Hosler, LLP, Kansas City, Missouri, for Defendant- Appellee. Before EBEL, BRISCOE and TYMKOVICH, Circuit Judges. EBEL, Circuit Judge. This case involves various age and gender discrimination claims brought by Plaintiff-Appellant Judith A. Bryant (“Bryant”) against her former employer, Defendant-Appellee Farmers Insurance Exchange (“Farmers”). The district court granted summary judgment in favor of Farmers. On appeal, Bryant raises two issues: (1) whether the district court abused its discretion in excluding portions of her summary judgment affidavit as inadmissable; and (2) whether there existed a genuine dispute of material fact regarding the pretextual nature of Farmers’ proffered nondiscriminatory reasons for terminating Bryant. As to Bryant’s affidavit, we hold, inter alia, that so long as the other prerequisites of Fed. R. Civ. P. 56(e) are met, it is permissible to submit simple statistical calculations such as averages without first being designated as an expert witness. Regarding pretext, we conclude that Bryant’s evidence calling into question the veracity of Farmers’ nondiscriminatory reasons for firing her are sufficient to establish pretext for summary judgment purposes. Thus, taking jurisdiction under 28 U.S.C. § 1291, we REVERSE. BACKGROUND I. The Parties Bryant is a sixty-one-year-old white female who worked for Farmers in various capacities from March 1992 until her termination in March 2000. 1 1 The record indicates that Bryant technically worked for a number of (continued...) 2 Farmers is a large, national insurance exchange based in California that sells a variety of policies, including coverage for recreational products, the division in which Bryant worked when she was terminated. II. Facts At the time of her termination, Bryant was a claims director within the Specialty Claims unit of the Western division of Farmers. Specifically, Specialty Claims handled claims from five types of insurance products: (1) Recreational Products Insurance (“RPI”); (2) Marine Insurance; (3) Loss Reporting; (4) Antique Auto; and (5) National Recovery. Each insurance product had its own individual director of claims. RPI dealt primarily with recreational vehicles and motorcycles. Within RPI, there existed two groups of claims adjusters: Liability and Adjusted Property of Physical Damage (“APD”). The liability section handled third-party claims where Farmers’ insureds were accused of causing injury to others. APD, on the other hand, handled first-party claims for property damage to vehicles covered by RPI policies. In March 1992, Bryant began working for Farmers as a Regional Marketing Manager in Denver, Colorado. Over the next few years, Bryant earned a number of promotions and moved throughout the company. By 1996, Bryant was director 1 (...continued) Farmers’ predecessor companies during the period from March 1992 to March 2000. For the sake of simplicity, we will refer to these predecessor companies (along with the defendant corporation) collectively as “Farmers.” 3 of claims for RPI. Along with the directors of claims for the other four insurance products in Specialty Claims, Bryant answered to Jack Honore, the division claims manager since 1998. One of Bryant’s subordinate managers was Steve Nagle, who headed the Liability section of RPI. Nagle, a 35-year-old male, eventually replaced Bryant after Honore terminated her in March 2000. A. Farmers’ internal auditing procedure On January 1, 1998, Farmers instituted a new quality control program that included the “Balanced Scorecard,” an attempt objectively to measure an office’s performance in four categories: (1) customer satisfaction; (2) financial perspective; (3) internal perspective; and (4) organizational learning/innovation. As part of the Balanced Scorecard evaluation, executives conducted a “Quality Assurance Review,” which was basically an audit of claims for each individual insurance product. Under the company procedure, auditors would review a statistical sample of claims files and measure two factors: “compliance” and “leakage.” The “compliance” score measured the thoroughness of the files reviewed. In other words, the auditors would make sure that each file had the necessary documentation, claimant contacts, and paperwork. After reviewing the files for completeness, the auditors assigned a percentage score based on how much of the 4 required information was missing or improperly recorded. The “leakage” score measured the percentage of overpayments. There were two components to this score: hard leakage and soft leakage. Hard leakage referred to objective overpayment. These were amounts that absolutely should not have been paid because, for example, the policy did not cover the injury or because the claim exceeded the policy limit. Soft leakage was a more subjective determination. It was the amount of overpayment on a claim based on the auditor’s evaluation of the claim’s worth. Thus, if the adjuster paid out $5,000 on a claim that the auditor felt was worth only $4,500, this would represent $500 in soft leakage. The hard and soft leakage amounts were totaled up and measured against the total payments of all files reviewed in order to determine the overall percentage of leakage for the reviewed files. B. Events leading up to Bryant’s termination From the time Farmers implemented the Quality Assurance Review in 1998 to the time Honore terminated Bryant in 2000, RPI was audited five times—three times by the home office and twice by Honore. 2 Three of these audits took place 2 The parties’ factual assertions about the results of these audits and the prescribed goals for RPI are often inconsistent with the record. What’s worse, the documents in the record are not exactly clear and often contradict each other. The main area of disagreement is what the “target” numbers for these audits were. Bryant claims that the standard for “passing” each audit was a compliance score of 90%, a 2% leakage for APD claims, and a 2.5% leakage for liability (continued...) 5 in the twelve months preceding Bryant’s termination. 3 1. January 1998 Audit RPI’s first audit was intended to serve as a baseline reading for the department’s compliance with new protocols and standards implemented by the company. RPI’s compliance score was 85.16%; it had no hard leakage, but soft leakage came in at 97.9%. For this audit, RPI’s goal was 91% compliance and 7.5% total leakage. 2. September 1998 Audit Farmers once again audited RPI after it had been given a few months to get oriented to the new quality control standards. RPI scored 84.88% compliance and 5.18% in overall leakage. As before, RPI’s goal was 91% compliance and 7.5% 2 (...continued) claims. Farmers, on the other hand, agrees that a 90% compliance score was needed but claims that leakage requirements were different. According to Farmers, to pass an audit RPI needed to obtain a hard leakage of 2% or below, and a soft leakage of 2% or below. Neither party is completely correct. Farmers’ position is based on vague admissions that Bryant made (in response to leading questions) during a particularly confusing part of her deposition. Looking more carefully at the original documents, it appears that Bryant’s position is closer to the truth, particularly for the audits Honore conducted. However, Bryant is incorrect in stating that the “passing” scores were the same for all five audits. The record indicates that the earlier audit goals were more lenient. 3 Bryant claims that RPI was audited four times over this period. This is because she counts a “mini-audit” that Honore conducted in August 1999 as a separate audit. In fact, Honore conducted two “mini-audits” in August and September 1999 and reported the combined results. Thus, we have treated the August and September “mini-audits” as one audit for the purposes of this opinion. 6 total leakage. One month after this Audit, Farmers hired Jack Honore to supervise Bryant. In her declaration before the district court, Bryant claims that Honore was immediately hostile to her, and that he suggested Bryant move to another position or quit her job. At about the same time, a director from the Farmers home office ordered Bryant to hire 35-year-old Steve Nagle to run the Liability department of RPI. According to other employees in RPI, however, the understanding was that Nagle was brought in to replace Bryant. Nagle allegedly admitted to another employee that he had been picked to come to Kansas City to replace Bryant. In fact, Nagle eventually replaced Bryant as director of claims for RPI after Bryant’s termination. Honore gave Bryant a negative performance evaluation for 1998, citing her failure to meet required Balanced Scorecard standards. 3. June 1999 Audit Despite Bryant’s negative performance evaluation, RPI’s numbers did not increase significantly. During the June 1999 Audit, RPI scored 88.46% for compliance and 7.88% in total leakage. Both parties agree that the “target” for compliance under the June 1999 audit was 90 percent. However, the record is not clear as to what the target leakage score was. Thus, we cannot say for certain whether RPI “failed” the June 1999 audit, but what is certain is that it did not 7 meet the compliance goals for the audit. After the audit, Honore informed Bryant in a formal written warning that if her department’s numbers did not improve, he would take corrective action (including the possibility of termination). Honore also told Bryant that he would personally conduct an audit of RPI’s files over the next 90 days. 4. September 1999 Audit Honore’s September 1999 Audit was really a combination of two “mini- audits.” RPI achieved a compliance score of 87.36% and an overall leakage score of 8.01%. Honore also broke down the score in terms of the two component departments of RPI: Liability and APD. APD had a 92.59% compliance score and a 1.53% leakage score. Liability, on the other hand, had a 76.90% compliance score and a 15.4% leakage score. For this audit, RPI’s “target” appears to be a 90% compliance score, a 2% APD leakage, and a 2.5% liability leakage. These numbers indicate that RPI’s APD division was performing up to standards but that its Liability division, headed by Steve Nagle (Bryant’s eventual replacement), was causing RPI to fail the audits. In January 2000, Honore again gave Bryant a negative performance evaluation, stating: [Bryant] has not demonstrated imagination in identifying problems or developing solutions to solve the problems. The lack of leadership and her apparent management style of complete delegation have led to morale problems particularly with other members of the management team. [Bryant] is reactive rather than proactive. It was very apparent after the September 1998 audit that much had not been 8 done to improve the results since the Jan. 98 audit. Again in June of 1999 most of the same problems were still there. While there has been some improvement since June 1999, the results are still unacceptable. Honore placed Bryant on probation and stated that he would return to the office in thirty days to conduct a final audit. He warned Bryant that if RPI’s results were not up to standards, she would be terminated. 5. March 2000 Audit As promised, Honore conducted another audit in March 2000. RPI scored 87.71% in overall compliance and 4.92% in overall leakage. As for the specific divisions, APD scored 88.78% compliance and 3.96% leakage. Liability scored 85.41% compliance and 5.79% leakage. Like the September 1999 Audit, RPI needed an overall compliance score of 90% or more, a liability leakage of less than 2.5%, and an APD leakage of less than 2% to pass the audit. Figure 1: Summary of RPI Audit Results Audit Compliance Compliance Leakage Leakage Score Goal Goal January 85.16% (O) 91% (O) 0 % (H) 7.5% (O) 1998 97.9% (S) September 84.88% (O) 91% (O) 5.18% (O) 7.5% (O) 1998 June 1999 88.46% (O) 90% (O) 7.88 (O) unknown 9 Audit Compliance Compliance Leakage Leakage Score Goal Goal September 87.36% (O) 90% (O) 8.01% (O) 1999 92.59% (A) 1.53% (A) 2% (A) 76.90% (L) 15.4% (L) 2.5% (L) March 2000 87.71% (O) 90% (O) 4.92% (O) 88.78% (A) 3.96% (A) 2% (A) 85.41% (L) 5.79% (L) 2.5% (L) O = Overall result H = Hard leakage A = Result for APD division S = Soft Leakage L = Result for Liability division Although the myriad facts and figures described above can be somewhat confusing, what is clear is that in the five audits prior to Bryant’s termination, RPI failed to meet the stated expectations for either compliance or leakage each time. Two days after the completion of the final March 2000 audit, Farmers terminated Bryant for poor performance. C. Procedural history On March 13, 2002, Bryant filed suit in the district court alleging federal age and gender discrimination claims under the Age Discrimination in Employment Act (“ADEA”), 29 U.S.C. § 621 et. seq., and Title VII of the Civil Rights Act of 1964 (“Title VII”), 42 U.S.C. § 2000e et seq. In addition, Bryant 10 brought age and gender discrimination claims under Kansas state law. After extensive discovery, Farmers moved for summary judgment on all claims. In response, Bryant provided an affidavit stating that she had personally reviewed 103 audit reports from several other Farmers claims offices throughout the country. These audits, conducted in 2000 and 2001, listed each office’s compliance and leakage scores. Bryant compiled this data into a spreadsheet and calculated the average compliance and leakage scores. Both the spreadsheet and the audit data were attached to Bryant’s affidavit. Bryant’s analysis revealed the following: (1) Compliance scores for the 103 offices ranged from 72.17% to 96.23%, with an average score of 87.73%; (2) Leakage scores for the 103 offices ranged from 0.1% to 24.0%, with an average score of 3.39%. Bryant then calculated the average audit score for RPI from June 1999 to March 2000 and came up with 88.70% compliance and 5.36% leakage. 4 Bryant noted that RPI had a higher-than-average compliance score and that 47 offices scored lower than RPI in compliance. As for leakage, Bryant stated that most of 4 Bryant’s calculation of RPI’s “average” compliance and leakage scores is suspect. Bryant calculated the average compliance and leakage scores for all RPI audits conducted between June 1999 and March 2000. However, Bryant averages four audit scores when in fact there were only three. As noted earlier, the September 1999 audit was actually a combination of two “mini-audits,” but Bryant counts the August audit separately because it artificially boosts RPI’s scores. In addition, Bryant omitted from her calculation the results from the September 1998 audit, which would have likely lowered the average scores for both compliance and leakage. 11 her department’s higher-than-average leakage scores were attributable to the poor performance of the Liability department, headed by Nagle (Bryant’s eventual replacement). Removing those scores, RPI would have had a lower-than-average leakage score, and 54 other offices would have scored lower than RPI in leakage. In addition to the analysis of other offices’ audit scores, Bryant submitted several other pieces of evidence for the court’s consideration. For example, Jeff Salsbury, a former RPI employee, testified in an affidavit that Nagle and Honore conspired to manipulate the audit data by artificially inflating the Liability department’s compliance and leakage figures. This was done by removing files which would pass the audit, thus leaving a skewed sample for the auditor. In resolving the summary judgment motion, the district court applied the burden-shifting framework of McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802-04 (1973). The court first ruled that Bryant had established a prima facie case of age and gender discrimination. It then concluded that Farmers had proffered a legitimate, non-discriminatory reason for Bryant’s termination: failure to meet established performance criteria. Finally, the court concluded that Bryant had not established a genuine issue of material fact as to whether this legitimate reason was merely a pretext for discrimination. In discussing pretext, the court excluded large portions of Bryant’s affidavit as inadmissable under the Federal 12 Rules of Civil Procedure and the Federal Rules of Evidence. Judgment was entered against Bryant, and the instant appeal followed. DISCUSSION I. Issue 1: Whether the district court properly excluded portions of Bryant’s affidavit when considering Farmers’ motion for summary judgment. In its order granting summary judgment, the district court, acting sua sponte, excluded as inadmissable the portions of Bryant’s affidavit (1) alleging that certain audit results were attributable to Nagle’s performance; (2) alleging that the audit performance standards were not intended to be mandatory; and (3) containing the statistical analysis of the 103 audit reports from other claims offices. The court concluded that these portions of Bryant’s affidavit were not based upon her personal knowledge and constituted improper opinion testimony. A. Standard of review We review a district court’s ruling on the admissibility of evidence for an abuse of discretion. Wilson v. Merrell Dow Pharm., Inc., 160 F.3d 625, 629 (10th Cir. 1998). Under this standard, a trial court’s decision will not be reversed unless “the appellate court has a definite and firm conviction that the lower court made a clear error of judgment or exceeded the bounds of permissible choice in the circumstances.” Moothart v. Bell, 21 F.3d 1499, 1504 (10th Cir. 1994) (quoting McEwen v. City of Norman, 926 F.2d 1539, 1553-54 (10th Cir. 1991)). 13 B. Legal standards for admissibility At the summary judgment stage, the parties need not submit evidence in a form admissible at trial; however, the content or the substance of the evidence must be admissible. Hardy v. S.F. Phosphates Ltd., 185 F.3d 1076, 1082 n.5 (10th Cir. 1999). For instance, a witness to a car accident could not submit his testimony at trial via affidavit because that statement would be hearsay. However, at the summary judgment stage, the affidavit is proper because its content—the eyewitness account of the affiant—is admissible. Fed. R. Civ. P. 56(e) further governs the admissibility of affidavits at the summary judgment stage: Supporting and opposing affidavits shall be made on personal knowledge, shall set forth such facts as would be admissible in evidence, and shall show affirmatively that the affiant is competent to testify to the matters stated therein. Sworn or certified copies of all papers . . . referred to in an affidavit shall be attached thereto or served therewith. Thus, it is clear that (1) the content of summary judgment evidence must be generally admissible and (2) if that evidence is presented in the form of an affidavit, the Rules of Civil Procedure specifically require a certain type of admissibility, i.e., the evidence must be based on personal knowledge. A summary judgment affidavit may not contain expert testimony unless the affiant has first been designated an expert witness under Fed. R. Civ. P. 26(a)(2). Parker v. Cent. Kan. Med. Ctr., 178 F. Supp. 2d 1205, 1210 (D. Kan. 2001), aff’d, 14 57 Fed. Appx. 401, 404 (10th Cir. 2003). Otherwise, any non-expert testimony in the form of opinions or inferences must be “(a) rationally based on the perception of the witness, (b) helpful to a clear understanding of the witness’s testimony or the determination of the fact in issue, and (c) not based on scientific, technical, or other specialized knowledge.” Fed. R. Evid. 701. C. Analysis Based on the foregoing, there are two potential evidentiary problems with Bryant’s declaration: (1) it might not be based upon personal knowledge; and (2) it might be improper opinion testimony. An individual analysis of each objection to the instant facts is warranted. 1. Lack of personal knowledge In holding Bryant’s declaration inadmissible, the district court stated, “[p]laintiff has not shown that she has personal knowledge of the matters she alleges: that certain audit results should be attributed to Mr. Nagle, that the standards set forth by defendant were not intended to be mandatory, or of the data upon which her statistics are based.” As to the allegations about Mr. Nagle and the mandatory standards, the district court is incorrect. Although Bryant does not give the source for her allegations, as supervisor of the RPI department, Bryant would have had personal knowledge of the fact that the audits were never intended to be used to dismiss 15 employees and the fact that the poor audit results should be attributed to Nagle’s performance. Similarly, the audit data should not have been excluded on the basis of lack of personal knowledge by Bryant. Bryant states in her declaration that she reviewed all of the records of audits performed in 2000 and 2001, and she also states that the audits were prepared in a format familiar to her. Thus, insofar as her declaration repeats the reported results of those audits as she reviewed them, it cannot be said that she lacked personal knowledge of those reports. In exploring this point, it is helpful to look at Fed. R. Evid. 602, which requires a witness’ testimony to be based upon personal knowledge. The advisory committee notes to that rule state: “This rule does not govern the situation of a witness who testifies to a hearsay statement as such, if he has personal knowledge of the making of the statement. . . . This rule, would, however, prevent him from testifying to the subject matter of the hearsay statement, as he had no personal knowledge of it.” The case at bar presents a similar situation. Here, Bryant attached a spread sheet with results from 103 audits. If she were to take the stand and individually read the results, the objection would not be that she lacked personal knowledge of the reports. Since she personally examined these audit reports, she had personal knowledge of their content. The proper objection would be hearsay, as the 16 documents are out-of-court statements offered for the truth of the matter asserted. See Fed. R. Evid. 801(c). However, if offered at trial, these documents would be admissible under either Rule 801(d)(2)(A) (statement by a party offered against that party) or Rule 803(6) (business records exception). Both the district court and Farmers complain that Bryant lacks personal knowledge because she does not know how these audits were performed, who performed the audits, or the methodology used in conducting these audits. However, these objections go to the weight of her testimony, not its admissibility. There is no dispute that the reviewed documents are authentic, a point bolstered by the fact that the documents were prepared on company letterhead and produced during discovery. See Denison v. Swaco Geolograph Co., 941 F.2d 1416, 1423 (10th Cir. 1991) (finding, in context of plain-error review, document to be authentic under similar circumstances). There is also no dispute that Bryant actually looked at the documents. As a result, the portions of her declaration which repeat the results of the audits are based on Bryant’s personal knowledge. 2. Opinion testimony As noted above, an affidavit may not contain expert testimony unless the affiant has first been designated as an expert under Fed. R. Civ. P. 26(a)(2). In this case, no such designation was made. Thus, we must determine whether the conclusions and opinions offered in Bryant’s declaration are properly 17 characterized as expert opinions under Fed. R. Evid. 702 or lay opinions under Rule 701. Bryant maintains that the mere calculation of an average of 103 numbers is not the sort of statistical determination which requires the special “knowledge, skill, experience, training, or education” prescribed by Rule 702. We agree. The Federal Rules of Evidence clearly permit the contents of voluminous writings to be presented in the form of a “chart, summary, or calculation.” Fed. R. Evid. 1006 (emphasis added). Taking a simple average of 103 numbers, though technically a statistical determination, is not so complex a task that litigants need to hire experts in order to deem the evidence trustworthy. A mathematical calculation well within the ability of anyone with a grade-school education is, in our opinion, more aptly characterized as a lay opinion under Fed. R. Evid. 701. Accordingly, we conclude that Bryant’s declaration was improperly excluded from the summary judgment evidence before the court. II. Whether Bryant submitted sufficient evidence of pretext to create a genuine issue of material fact for trial. A. Standard of Review We review the district court’s grant of summary judgment de novo and must apply the same legal standard used by the district court. Simms v. Okla. ex rel. Dep’t of Mental Health & Substance Abuse Servs., 165 F.3d 1321, 1326 (10th Cir. 1999). Under Fed. R. Civ. P. 56(c), summary judgment is only appropriate if 18 the pleadings and admissible evidence produced during discovery, together with any affidavits, show that “there is no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law.” When applying this standard, we must view the evidence and draw reasonable inferences therefrom in the light most favorable to the nonmoving party. Simms, 165 F.3d at 1326. B. Applicable law Where, as here, a plaintiff puts forth circumstantial evidence of age or gender discrimination, we evaluate the case using the burden-shifting framework of McDonnell Douglas, 411 U.S. at 802-04. See also MacDonald v. Delta Air Lines, Inc., 94 F.3d 1437, 1441 (10th Cir. 1996) (applying McDonnell Douglas in the ADEA context). 5 Under McDonnell, the plaintiff must first establish a prima facie case of discrimination. 411 U.S. at 802. The burden of production then shifts to the employer to articulate some legitimate, nondiscriminatory reason for the adverse employment action. Id. Once this is done, the presumption of discrimination established by the prima facie showing “simply drops out of the picture.” St. Mary’s Honor Ctr. v. Hicks, 509 U.S. 502, 511 (1993). The plaintiff then carries the full burden of persuasion to show that the defendant discriminated Here, the district court ruled there did not exist a genuine issue of material 5 fact as to discriminatory intent, and Bryant has not appealed that ruling. Thus, we limit our analysis only to Bryant’s indirect evidence of discrimination. 19 on the illegal basis of age or gender. See Ingels v. Thiokol Corp., 42 F.3d 616, 621 (10th Cir. 1994), abrogated on other grounds, Martinez v. Potter, 347 F.3d 1208, 1210 (10th Cir. 2003). The plaintiff may do so by showing that the proffered reason is pretextual. Ingels, 42 F.3d at 621. Here, neither party disputes the district court’s ruling that Bryant has established a prima facie case of both age and gender discrimination, and we see no need to disturb that conclusion. Further, Farmers met its burden by asserting that it terminated Bryant because she failed to meet the performance criteria established for her position. Poor performance is a quintessentially legitimate and nondiscriminatory reason for termination. See, e.g., Garrett v. Hewlett Packard Co., 305 F.3d 1210, 1218 (10th Cir. 2002). Thus, this case hinges upon whether Bryant has put forth evidence establishing a dispute of fact as to whether Farmers’ stated reason for terminating her was pretextual. Under Tenth Circuit precedent, pretext may be shown by “such weaknesses, implausibilities, inconsistencies, incoherencies, or contradictions in the employer’s proffered legitimate reasons for its action that a reasonable factfinder could rationally find them unworthy of credence and hence infer that the employer did not act for the asserted non-discriminatory reasons.” Morgan v. Hilti, Inc., 108 F.3d 1319, 1323 (10th Cir. 1997) (quoting Olson v. Gen. Elec. Astrospace, 101 F.3d 947, 951-52 (3d Cir. 1996)). A plaintiff can 20 make a showing of pretext with evidence that the defendant’s stated reason for termination was false. Kendrick v. Penske Transp. Servs., Inc., 220 F.3d 1220, 1230 (10th Cir. 2000). Our precedent does not require a plaintiff to offer any evidence of actual discrimination when attempting to show pretext. See Ingels, 42 F.3d at 621. Evidence tending to show pretext permits an inference that the employer acted for discriminatory reasons. Morgan, 108 F.3d at 1323. “Thus, a factfinder may, but is not required to, find discrimination when a plaintiff presents evidence that the defendant’s proffered reasons are unworthy of credence.” Ingels, 42 F.3d at 621- 22. But, as in Ingels, this case arrives to us at the summary judgment stage. See id. at 622. At that stage, if a plaintiff advances evidence establishing a prima facie case and evidence upon which a factfinder could conclude that the defendant’s alleged nondiscriminatory reasons for the employment decisions are pretextual, the case should go to the factfinder. Id. In short, at the summary judgment stage, the inference of discrimination permitted by evidence of pretext must be resolved in favor of the plaintiff. See Simms, 165 F.3d at 1326. Thus, once a plaintiff presents evidence sufficient to create a genuine factual dispute regarding the veracity of a defendant’s nondiscriminatory reason, we presume the jury could infer that the employer acted 21 for a discriminatory reason and must deny summary judgment. See Morgan, 108 F.3d at 1323; Ingels, 42 F.3d at 621-22. C. Analysis After reviewing Bryant’s summary judgment evidence, we are convinced that she has established a genuine factual dispute regarding pretext. Bryant’s affidavit and accompanying statistical summary of audit results indicate that RPI’s audit scores were not substantially deficient when compared to other claims offices. This evidence also indicates that her department’s poor performance was due—in large part—to the performance of the very person who replaced her, Nagle. In addition, Bryant raises doubts as to the accuracy and reliability of the audit data. Nine months before Bryant’s termination, Nagle allegedly admitted to Jeff Salsbury, a claims representative in the liability department, that he was brought in to take Bryant’s place as claims director of RPI. Salsbury also testified that: I personally observed some manipulation of the audit files before the files were audited. Steven Nagle came to others and me with a list of files which were to be audited and instructed us to pull the files, which would pass. Then he took the files we selected as ones which would pass. I don’t know what he did with the files, but I thought it was strange that we were directed to pick-out the “good” files from the list of files to be audited. Handpicking the files made me feel very uneasy. In evaluating this evidence, the district court found that any allegation of a conspiracy remained “highly speculative” because Salsbury could not state with 22 certainty what happened to the pulled files. Certainly it is possible that Nagle could have asked for the good files to be pulled so that he could “boost” the results of the audits by making sure they were included. Another, equally credible possibility is that he removed the files in order to artificially lower the scores. It is not possible from the record to tell which of these two scenarios is the truth. But a plaintiff facing summary judgment does not have to conclusively establish the truth. She must only establish that there is a genuine factual dispute with regard to the truth, and in this she has succeeded. Choosing between these two possibilities is exactly the type of determination a jury is empowered to make. When viewed in a light most favorable to Bryant, Salsbury’s personal observations of file manipulation, combined with Nagle’s alleged admissions and the undisputed fact that Honore selected Nagle to replace Bryant, raise a genuine issue of material fact as to whether the audit results accurately reflected Bryant’s performance. As such, Farmers’ reliance—in part—on those audit results as grounds for termination may have been pretextual. It is true that the poor audit scores were not the sole reason for Bryant’s termination. In his final letter to Bryant, Honore cited a number of other performance deficiencies, including failure to completely and successfully implement a new company program, failure to ensure that subordinate employees 23 were properly licensed, and failure to complete certain course work. Bryant has not offered any evidence tending to show that these other deficiencies in her performance were false in any way. As a general rule, an employee must proffer evidence that shows each of the employer’s justifications is pretextual. Tyler v. RE/MAX Mountain States, Inc., 232 F.3d 808, 814 (10th Cir. 2000). However, we recognize that “when the plaintiff casts substantial doubt on many of the employer’s multiple reasons, the jury could reasonably find the employer lacks credibility. Under those circumstances, the jury need not believe the employer’s remaining reasons.” Id. (citations omitted); accord Fuentes v. Perskie, 32 F.3d 759, 764 n.7 (3d Cir. 1994). It is true that in the instant case, Bryant has only cast doubt on one of the proffered reasons for her termination: failure to meet internal audit standards. But it is also clear that Bryant’s failure to meet audit standards was the dominant reason put forth by Farmers to justify her termination. Bryant’s failure to address the other, less consequential reasons for her termination does not entitle Farmers to summary judgment. It is not simply a question of how many of the defendant’s reasons a plaintiff has refuted, but rather a question of whether casting doubt on a particular justification necessarily calls into doubt the other justifications. See Tyler, 232 F.3d at 814. Where, as here, one of the stated reasons for termination predominates over the others, demonstrating that reason to be pretextual is 24 enough to avoid summary judgment. 6 CONCLUSION Because Bryant has raised a genuine factual issue regarding whether Farmers’ stated reason for her termination was false, summary judgment was inappropriate. Accordingly, we REVERSE the judgment of the district court. 6 This court’s recent decision in Jaramillo v. Colorado Judicial Dep’t, 427 F.3d 1303 (10th Cir. 2005), further supports our conclusion. In Jaramillo, this court held that “[i]n some cases, . . . a successful attack on part of the employer’s legitimate, non-discriminatory explanation is enough to survive summary judgment even if one or more of the proffered reasons has not been discredited.” Id. at 1310. This court then listed five circumstances when “[s]omething less than total failure of the employer’s defense is sufficient to create a genuine issue of fact.” Id. This case falls directly into the third circumstance listed in Jaramillo, where “the employer offers a plethora of reasons, and the plaintiff raises substantial doubt about a number of them.” Id. (citing Tyler, 232 F.3d at 814). As we have noted, Farmers asserted several reasons why it fired Bryant. The predominant reason, however, was that she had failed to meet internal audit standards. Bryant has asserted sufficient evidence indicating that that dominant reason was pretextual to submit that question to a jury. That is sufficient in this case for Bryant to survive summary judgment. The pretext inquiry about whether a plaintiff raises “substantial doubt about a number of [the employer’s reasons]” is not limited to just a numeric quantitative assessment of the proffered reasons, but also includes a qualitative assessment that takes into account which reasons dominated the employer’s decisionmaking process. Here, the failure to meet the internal audit was the dominant reason advanced, and hence putting on evidence that it was pretextual satisfies the third circumstance listed in Jaramillo. 25