Proving discrimination when no one admits it: comparisons, pretext, and “culture fit”
Key takeaways
- No confession is required. Discrimination is proven by inference: you're in a protected class, you could do the job, something adverse happened — fired, demoted, passed over — and the circumstances suggest the class played a substantial role.
- California rejected the “stray remarks” doctrine: comments by coworkers and non-decisionmakers can't be ignored as a category — courts weigh the whole record (Reid v. Google, 2010).
- Comparators carry cases: colleagues outside your class, under the same rules and the same decisionmaker, treated better for the same conduct or record.
- “Culture fit” and other unfalsifiable criteria receive scrutiny, not deference — vagueness plus a skewed pattern is itself evidence.
- Other employees' accounts of similar treatment — “me too” evidence — are admissible to show motive and pattern.
Nobody says the quiet part anymore, and nobody writes it down. The email that spells out the real reason does not exist, and waiting to find one is not a legal strategy. Discrimination cases are built instead by inference — a method the law spent decades sharpening precisely because admissions vanished. The materials are ordinary: who was treated better, what words people used, whether the stated criteria hold up, what happened to others like you. Assembled in order, they answer the only question the statute asks: did your protected characteristic play a substantial role in what was done to you.
The part you actually have to prove
Start with what you don't have to prove, because it's the shorter list, and the better news. You don't have to prove your manager is a bigot. You have to prove the protected characteristic was a substantial motivating reason for what was done to you — whether or not the manager would fail some test of the heart. Those are different jobs. One requires reading a mind; the other requires reading a record. The jury gets to read between the lines.
You belong to a protected class — race, sex, age over forty, disability, religion, national origin, sexual orientation, and the rest of FEHA's list. You were doing the job. Something adverse happened: you were fired, demoted, rejected, passed over. And the circumstances support an inference that the protected characteristic was a substantial motivating reason — the causation standard explained in the framework guide. Everything below is about that last element, because it is where every real case is decided.
Brian Reid and the words people used
Brian Reid was hired by Google at fifty-two — a Ph.D., a former Stanford professor — and his one formal review said he consistently met expectations. Around him, the record filled with the other vocabulary: he was "an old man," "an old fuddy-duddy"; his ideas were "obsolete," "too old to matter." Less than two years after he arrived he was out, deemed not a "cultural fit."
Google's argument reached the California Supreme Court: those remarks were "stray" — made by non-decisionmakers, or outside the decision process — and should be shut out of the case entirely, so no jury would ever hear them. The Court refused (Reid v. Google (2010)). California courts may not discard such remarks as a category. The evidence is weighed as a whole, because a comment in the hallway can reveal the atmosphere the decision was made in, and a non-decisionmaker's bias can travel upward into the decision itself. The Court kept the balance honest in both directions: one remark, standing alone, rarely carries a case. Woven together with a skewed pattern and a reason that doesn't hold, the same remark becomes part of a picture that does.
Comparators: the employer's own choices, side by side
The workhorse of discrimination proof is comparison. A comparator is someone outside your protected class, held to the same standards, answering to the same decisionmaker, whose situation matched yours where it mattered — and whose outcome didn't. The salesperson with the same missed quota who kept his territory. The younger manager with the thinner file who survived the "restructuring." The colleague whose identical conduct drew coaching where yours drew termination. Comparator evidence carries cases because nobody has to speculate about what was in anyone's heart: the employer's own choices, laid side by side, do the arguing. Building it means noticing specifics while you still can — names, roles, dates, outcomes — and keeping the papers that show the common rules: the policy, the quota memo, the review rubric everyone was supposedly scored against.
"Culture fit" and other reasons no one can fail
Some stated reasons can be checked — a sales number, an error rate, an attendance record. Others are built to be uncheckable: fit, polish, executive presence, "not quite the right energy." The law does not hand subjective criteria a pass. It examines them harder, because vagueness is where preference hides. A criterion no one defined, applied by deciders who all resemble each other, producing a pattern that tracks a protected line — the older candidates who never quite fit, the women who somehow lacked presence — is not a neutral judgment shielded from review. It is a fact pattern courts recognize, and the very failure to say what "fit" meant becomes part of the proof. Reid's case carried that phrase to the Supreme Court for a reason: culture fit is what the quiet part sounds like now.
"Me too" — the others he did it to
The answer is "me too" evidence. California courts admit testimony from other employees treated the same way, especially at the same hands, because what a decisionmaker did to others is evidence of what moved him here. A pattern several people witnessed stops being a perception problem and becomes a practice. Statistics can serve the same role at a larger scale, with one honest caveat: small numbers prove little, and three departures don't make a dataset.
Testing their story
Every inference case eventually collides with the employer's stated reason, and they're going to come back with a tidy one: performance, restructuring, fit. The tools for testing it — explanations that shift between forums, the personnel file the statute lets you demand, discipline steps skipped, documents that should exist and don't, what discovery pries loose — have their own complete guide. The short version: a stated reason is evidence to be tested, not a verdict to be accepted — and the testing runs on the employer's own paper.
The limits, priced straight
The same doctrine that builds these cases marks their edges, and none of these cases are slam dunks. Being in a protected class plus a bad outcome is not, alone, a case — the inference needs something to stand on. A single remark with nothing around it usually fails. And one recurring pattern cuts for employers: when the same person who hired you, knowing your age or race or sex, fires you a short time later, courts entertain an inference against discrimination — the reasoning being that a biased decisionmaker wouldn't have hired you in the first place (Horn v. Cushman & Wakefield (1999)). That inference can be answered: circumstances change, and bias can attach to what you became — older, pregnant, disabled, outspoken. Comparators, patterns, shifting reasons, and me-too voices overcome each of these limits regularly. But they mark the line between a grievance and a claim, and it's better to price that line honestly in the first conversation than to learn it from a judge.
Common questions
I have no direct evidence — no slurs, no smoking gun. Can a case still be made?
Possibly — most winning cases have neither. Discrimination is proven by inference: comparators treated better under the same rules, criteria too vague to check, a stated reason that shifts or has no paper behind it, remarks that reveal the atmosphere, and accounts from others who lived the same pattern. What matters is whether the pieces together support the inference — not whether any single piece confesses.
Is it legal to fire someone for “culture fit”?
The phrase itself isn't unlawful — but it isn't a shield either. Subjective criteria get scrutiny, not deference. If no one can define the fit, the deciders all resemble each other, and the people who never fit share a protected characteristic, the vagueness becomes evidence rather than explanation.
A coworker — not my boss — made comments about my age. Does that matter legally?
It can. California rejected the rule that would automatically toss remarks by non-decisionmakers: courts must weigh all the evidence together, because bias in the room can travel into the decision (Reid v. Google, 2010). A remark alone rarely wins a case; a remark alongside a pattern is another matter.
Other people at my company went through the same thing. Can their experiences help my case?
Yes. “Me too” evidence — similar treatment of other employees, especially by the same manager — is admissible in California to show motive, intent, and pattern. Several voices describing one practice is among the strongest answers to “you're imagining it.”
The person who hired me is the one who fired me. Is that a problem?
It's a consideration, not a verdict. Courts sometimes draw an inference against discrimination when the same person hires and fires within a short period — the logic being that bias would have blocked the hire. The inference is rebuttable — it can be answered: circumstances change, bias can attach to what you became (older, pregnant, disabled, outspoken), and the rest of the evidence still gets weighed.
Article history
July 31, 2026 — Published and reviewed by Bruce Weisenberg.