Koehler v. Infosys: Rule 702 Requires Proven Reliability for Name-Matching Demographic Proof, and Courts May Exclude It Sua Sponte

1. Introduction

In Brenda Koehler v. Infosys Technologies Limited, Inc., four non–South Asian former employees/applicants (Gregory Handloser, Brenda Koehler, Kelly Parker, and Layla Bolten) alleged that Infosys discriminated against non–South Asians in hiring and employment. Their class-based theory depended heavily on statistical proof offered by labor economist David Neumark, who attempted to infer whether employees/applicants were “South Asian” through a last-name “name-matching” methodology.

After the district court excluded Neumark under Federal Rule of Evidence 702, it denied class certification, denied a late request to supplement summary judgment briefing with alternative demographic data (the “PeopleFluent data”), and granted summary judgment for Infosys on all claims. The Seventh Circuit affirmed across the board.

The key issues on appeal were: (1) whether Neumark’s testimony satisfied Rule 702 (including whether exclusion could extend beyond the precise scope of Infosys’s motion), (2) whether class certification could stand without that expert proof, (3) whether plaintiffs could reopen or supplement the summary judgment record years later, and (4) whether plaintiffs had triable disparate-treatment, disparate-impact, or “pattern and practice” claims.

2. Summary of the Opinion

  • Rule 702 / Daubert: The court held the district court acted within its discretion in excluding Neumark because he lacked qualifications to identify South Asian names and plaintiffs presented no reliability foundation (testing, error rate, peer review, standards, acceptance) for his name-matching method.
  • Sua sponte exclusion permitted: Even if Infosys’s motion emphasized applicants, the district court could exclude the testimony more broadly; a district court may rule on expert admissibility sua sponte so long as parties have an opportunity to be heard and the court applies Rule 702.
  • Class certification: Plaintiffs conceded they could not satisfy Federal Rule of Civil Procedure 23 without Neumark; denial was affirmed.
  • No late supplementation: Denial of leave to supplement summary judgment briefing with PeopleFluent data was affirmed; plaintiffs had the data for years and declined earlier invitations to supplement. Their belated “excusable neglect” theory under Rule 6(b) was waived because they did not raise it below.
  • Summary judgment on merits: “Pattern and practice” claims under International Brotherhood of Teamsters v. United States could not proceed absent class certification; individual disparate-treatment claims failed for lack of pretext (or for constructive discharge, lack of intolerable conditions); disparate-impact claims failed without statistical proof.

3. Analysis

3.1. Precedents Cited

A. Expert admissibility, Rule 702, and the district court’s gatekeeping role

  • Daubert v. Merrell Dow Pharmaceuticals, Inc. and Kumho Tire Co. v. Carmichael supply the foundational “gatekeeping” framework requiring reliability and relevance for expert testimony. The Seventh Circuit treated this case as a straightforward application: a methodology that purports to classify protected-group status must be grounded in reliable principles and reliably applied.
  • Artis v. Santos framed the standard of review (abuse of discretion) and reinforced that appellate review is deferential when the district court applies the correct legal framework.
  • Gopalratnam v. Hewlett-Packard Co. and Anderson v. Raymond Corp. supplied the Seventh Circuit’s familiar three-part Rule 702 inquiry (qualifications, reliability, relevance) and emphasized its case-specific nature.
  • Kirk v. Clark Equip. Co., Deputy v. Lehman Bros., Inc., Smith v. Ford Motor Co., and Schultz v. Akzo Nobel Paints, LLC were used to describe (and contextualize as flexible) the reliability factors (testability, peer review, error rate, standards, general acceptance) and to underscore that the focus is on methodological rigor, not whether the expert is “ultimately correct.”

B. District court authority to exclude expert testimony beyond a party’s precise motion

  • Perez v. K & B Transp., Inc. provided the key rule: a district court may rule on expert admissibility sua sponte if the parties had an opportunity to be heard and the court applies Rule 702.
  • Kirstein v. Parks Corp. and Lewis v. CITGO Petroleum Corp. supported the Seventh Circuit’s longstanding view that the Daubert inquiry has no required form and may include sua sponte scrutiny.

C. Expert “qualification” is question-specific

  • Gayton v. McCoy, Carroll v. Otis Elevator Co., United States v. Truitt, Ollison v. Gossett, and Hall v. Flannery collectively grounded the court’s conclusion that qualifications must match the specific task—here, inferring South Asian status from surnames.
  • The court’s discussion (including examples citing De Souza v. I.N.S., D'Souza v. Holder, D'Souza v. U.S. Att'y Gen., United States v. Dave, Dave v. Ashcroft, Kondamudi v. Garland, Joseph v. Gonzales, and Joseph v. McDonough) illustrated a practical point that became legally dispositive: “Western-sounding” names can overlap with South Asian names, making “common sense” name judgments precisely the kind of inference that demands specialized competence and a validated method.

D. Classwide “pattern and practice” discrimination and Teamsters

  • The court treated International Brotherhood of Teamsters v. United States as class-action architecture: once class certification was denied, a Teamsters “standard operating procedure” claim could not survive as a standalone vehicle.
  • Franks v. Bowman Transp. Co., Cooper v. Fed. Rsrv. Bank of Richmond, Gilty v. Vill. of Oak Park, Williams v. Boorstin, and Matthews v. Waukesha County reinforced the point that pattern-and-practice proof is, at most, collateral in an individual case and is conventionally tied to classwide adjudication.

E. Disparate treatment (pretext) and constructive discharge

  • McDonnell Douglas Corp. v. Green framed the burden-shifting approach (though the court proceeded directly to pretext under Barnes v. Bd. of Trs. of Univ. of Ill. once Infosys offered nondiscriminatory reasons).
  • Ortiz v. Werner Enters., Inc. supplied the “evidence as a whole” framing, even when the court uses McDonnell Douglas as an analytic tool.
  • Argyropoulos v. City of Alton, Anderson v. Mott St., Robertson v. Dep't of Health Servs., Cunningham v. Austin, and Brooks v. Avancez supported the “honest belief” conception of pretext: plaintiffs must show a lie, not merely an unfair or mistaken decision.
  • For constructive discharge, the court relied on Patterson v. Ind. Newspapers, Inc., Fischer v. Avanade, Inc., Chapin v. Fort-Rohr Motors, Inc., Beverly v. Abbott Lab'ys, Wince v. CBRE, Inc., and Fields v. Bd. of Educ. of City of Chicago to underscore the “pervasive and extreme” requirement and to distinguish rude or offensive episodes from “intolerable” conditions.

F. Disparate impact and the need for statistical proof

  • Downing v. Abbott Lab'ys, Puffer v. Allstate Ins. Co., and Gilty v. Vill. of Oak Park were used to reiterate that disparate impact claims typically require adequate statistical evidence and that small/weak samples can be fatal.

G. Record management, supplementation, and waiver

  • On late supplementation, the court relied on Ewing v. 1645 W. Farragut LLC, Rainey v. Taylor, Perez v. Staples Cont. & Com. LLC, and Moro v. Shell Oil Co. to affirm broad district-court discretion to manage the record and deny post hoc attempts to “do over” summary judgment submissions when the information was available earlier.
  • Plaintiffs’ invocation of Federal Rule of Civil Procedure 6(b) and Pioneer Investment Services v. Brunswick Associates Ltd. failed because the argument was not made in the district court; the Seventh Circuit treated this as waiver.
  • The panel also highlighted systemic delay and the obligation of timely case management by referencing Fed. R. Civ. P. 1 and Simstad v. Scheub.

3.2. Legal Reasoning

A. The “sua sponte” scope holding: notice and opportunity to be heard matter more than motion phrasing

Plaintiffs tried to narrow the exclusion to applicant-related opinions on the theory that Infosys’s motion targeted applicants. The Seventh Circuit rejected this as inconsistent with Perez v. K & B Transp., Inc.: courts may assess admissibility sua sponte, and what matters is whether the parties had a fair chance to litigate the core Rule 702 problems. Here, Infosys challenged the name-matching methodology and Neumark’s qualifications generally; the district court held a hearing; and plaintiffs had ample opportunity to defend the method. The court’s rationale also avoided an artificial severability: if the identified methodological flaws go to the core of the technique, they do not become “reliable” simply because the same technique is used in a different subset of the analysis.

B. Qualifications: being an economist is not expertise in ethnographic surname classification

The opinion’s sharpest move is its insistence on question-specific expertise. Neumark’s field (labor economics) did not supply, without more, specialized knowledge about identifying whether last names indicate South Asian identity. The court treated his deposition admissions—no training or experience in name matching or identifying South Asian names—as dispositive. Plaintiffs’ fallback (“he only removed clearly non-South Asian ‘western’ names”) failed because the categories can overlap: a person can have a “western-sounding” surname and still be South Asian. That overlap made his “common sense” filtering precisely the step that required demonstrable expertise.

C. Reliability: the proponent must actually build the Rule 702 record

The court’s reliability reasoning is procedural as much as substantive. Even if name-matching can sometimes be done reliably, plaintiffs bore the burden (more likely than not) to show the method’s reliability under Rule 702. They offered no evidence of testing, peer review, error rate, standards, or acceptance—nor did Neumark have experience with the technique. On appeal, plaintiffs attempted to supply reliability support, but the Seventh Circuit effectively treated this as too late: abuse-of-discretion review asks whether the district court acted reasonably on the record presented, not on new materials assembled years later.

D. Case-management discipline: no late “do-over” after declining earlier opportunities

On the PeopleFluent data, the court emphasized timeline and choice. Plaintiffs learned of the data in 2016 and obtained it in 2016–2017. A magistrate judge repeatedly offered plaintiffs an avenue to supplement their filings then; they did not do so. Only after losing the expert did they seek to add new material (in late 2022). Under Perez v. Staples Cont. & Com. LLC and Moro v. Shell Oil Co., a district court may deny supplementation when the information was available and no adequate explanation exists for the failure to submit it. The Seventh Circuit also rejected the attempt to reframe the issue under Rule 6(b) and Pioneer Investment Services v. Brunswick Associates Ltd. because plaintiffs did not raise that legal standard below—hence waiver.

E. Merits: Teamsters needs a class; individuals must prove causation, not general atmosphere

On the merits, the court’s logic cascades from the procedural rulings:

  • Teamsters “pattern and practice”: Once class certification was denied, Teamsters could not serve as an independent classwide burden-shifting mechanism. At most, pattern evidence could be “collateral” to individual claims, consistent with Gilty v. Vill. of Oak Park.
  • Individual disparate treatment: Under the “honest belief” pretext cases, plaintiffs had to show Infosys’s reasons were lies. Handloser could not rebut that he received a “four” in the relevant review cycle and was terminated in a reduction in force targeting that rating. Parker and Koehler could not rebut Infosys’s stated skills/qualification reasons. The court cited Johnson v. Nordstrom, Inc. and Ennin v. CNH Indus. Am., LLC to reject subjective belief and speculation as substitutes for evidence of pretext.
  • Constructive discharge: Bolten’s evidence described offensive comments and workplace slights, but the court applied the high threshold from Patterson v. Ind. Newspapers, Inc. and later cases to hold these facts did not rise to “intolerable” conditions.
  • Disparate impact: With Neumark excluded and supplementation denied, plaintiffs lacked the statistical proof typically required under Puffer v. Allstate Ins. Co. and Downing v. Abbott Lab'ys.

3.3. Impact

  • Heightened scrutiny of “name-based” demographic inference: The decision signals skepticism toward using surnames as a proxy for race/ethnicity/national origin absent (i) an expert qualified in that precise inferential task and (ii) a documented, validated methodology with known error properties. Plaintiffs and defendants should expect courts to demand a concrete reliability showing—not assurances.
  • Record-building is outcome-determinative under deferential review: The opinion is a caution that appellate courts will not rescue a party who fails to present reliability materials at the Rule 702 stage and then seeks to relitigate with new support on appeal.
  • Procedural sequencing in employment class actions: Where class certification and dispositive motions hinge on a statistical expert, exclusion of that expert can functionally end the case. This increases the strategic importance of early Daubert preparation and alternative proof.
  • Case-management consequences for late-disclosed evidence: Even where a defendant failed to timely produce requested data, a plaintiff who later obtains it must act promptly and preserve requests to supplement; declining magistrate-judge-offered remedies can foreclose later attempts.
  • Reinforcement (and practical narrowing) of Teamsters: By tying the viability of a Teamsters “pattern and practice” theory to class certification, the opinion limits plaintiffs’ ability to use Teamsters as a substitute for individualized causation proof after a class is denied.

4. Complex Concepts Simplified

  • Rule 702 / Daubert: Before an expert can testify, the judge must ensure the expert is properly qualified for the specific task and that the method is reliable (not just plausible). Courts look for markers like testing, peer review, standards, and known error rates.
  • “Sua sponte” exclusion: A judge can exclude expert testimony even without a perfectly matching party request, so long as the parties had a fair chance to argue admissibility and the judge uses the correct Rule 702 framework.
  • Class certification (Rule 23): To proceed as a class, plaintiffs must meet strict requirements (common issues that can be resolved together, adequate representatives, etc.). If the evidence meant to prove classwide discrimination is excluded, certification often fails.
  • Teamsters “pattern and practice”: A framework typically used in classwide cases where plaintiffs attempt to show discrimination was the employer’s regular policy. Without a class, the framework generally does not operate as the main engine of liability.
  • Disparate treatment vs. disparate impact: Disparate treatment is intentional discrimination (you were treated worse because of protected status). Disparate impact targets neutral practices that disproportionately harm a protected group—usually requiring statistical proof.
  • Pretext: Not whether the employer was wrong or unfair, but whether the stated reason was a lie masking discrimination.
  • Constructive discharge: Quitting can be treated like being fired only if conditions were so intolerable that a reasonable person would feel forced to resign.
  • Waiver: If you do not raise a legal argument in the district court, you generally cannot raise it later on appeal.

5. Conclusion

Koehler v. Infosys is a procedural-and-proof-centered employment discrimination decision with a clear lesson: when a case depends on statistical inference—especially inference built on surname-based demographic classification—the proponent must (1) present an expert qualified for that exact inferential task and (2) build a concrete reliability record under Federal Rule of Evidence 702. The Seventh Circuit also reaffirmed that district courts may police expert admissibility sua sponte, that litigants are bound by strategic choices made in managing late-produced evidence, and that Teamsters “pattern and practice” theories do not survive the denial of class certification as a substitute for individualized proof.