Age Discrimination in Resume Screening: What the Data Shows and What to Do About It

I've spent most of this column on how software reads your resume: keywords, parsing logic, match scores. This one starts somewhere older, because it turns out the bias came first, and the software mostly learned it from us. If you've ever wondered whether something as specific as a graduation year or a decades-old job title could quietly cost you an interview, the honest answer is yes, and there's now a fairly large body of evidence on exactly how much.

The experiment that measured it directly

The clearest evidence doesn't come from a survey asking people whether they feel discriminated against, it comes from a field experiment that removed the guesswork entirely. Researchers built matched sets of fictitious resumes representing three age bands (roughly 29-31, 49-51, and 64-66), gave each set identical qualifications, and sent them to more than 13,000 real job openings across 12 cities and 11 states, over 40,000 applications in total. It's one of the largest hiring audits of its kind ever run.

The results were consistent and blunt. The callback rate began falling substantially once applicants moved into their 40s, and kept falling from there. Applicants aged 49 to 51 received about 29% fewer callbacks than applicants aged 29 to 31 with the exact same resume otherwise. The effect was worse for women: applicants aged 64 to 66 received roughly 47% fewer callbacks than younger women, a considerably larger gap than the one measured for men in the same older bracket. ([Federal Reserve Bank of San Francisco, "Age Discrimination and Hiring of Older Workers"](https://www.frbsf.org/research-and-insights/publications/economic-letter/2017/02/age-discrimination-and-hiring-older-workers/); [Center for Retirement Research at Boston College](https://crr.bc.edu/age-discrimination-still-a-major-obstacle-for-older-workers/))

Why this counts as discrimination under the law, not just a rough patch

The Age Discrimination in Employment Act of 1967 (ADEA) protects applicants and employees who are 40 or older from discrimination in hiring, firing, promotion, and other terms of employment, at private employers with 20 or more employees, state and local governments, and the federal government. ([U.S. Department of Labor](https://www.dol.gov/general/topic/discrimination/agedisc); [U.S. Equal Employment Opportunity Commission](https://www.eeoc.gov/age-discrimination))

Importantly, the law doesn't require proving anyone intended to discriminate, which matters, because almost no recruiter or hiring algorithm is consciously doing so. It relies on the same disparate-impact standard I've written about in the context of AI screening tools: if a hiring practice produces a significantly worse outcome for a protected group, that can be unlawful regardless of intent, measured against the federal "four-fifths rule" (a selection rate for one group below 80% of the highest-passing group's rate is treated as a red flag worth investigating). A 29% or 47% gap in callbacks, if it held up as a selection-rate comparison rather than just a callback-rate one, would sit well past that threshold.

The newest wrinkle: this bias is now showing up inside AI screening tools

For decades, this kind of discrimination lived in the judgment of individual recruiters and hiring managers, hard to prove in any single case, easy to demonstrate only in aggregate, the way the field experiment above did. That's changing, because hiring decisions increasingly run through software first.

Mobley v. Workday, the nationwide collective-action lawsuit I covered in detail in my breakdown of the case, alleges race, disability, and age discrimination against Workday's AI-powered applicant screening tools. The age claim is the one this article is about, and it's not a footnote: the case is proceeding under the ADEA specifically, with a court-authorized notice process that opened in January 2026 for applicants 40 and older who were screened through Workday's tools since September 2020. ([SHRM, "The Workday AI Lawsuit Is a Wake-Up Call for HR"](https://www.shrm.org/topics-tools/news/technology/workday-ai-lawsuit-wake-up-call-hr); [Forbes, "A Federal Judge, A 1967 Law And A Billion Rejected Job Applications"](https://www.forbes.com/sites/sheilacallaham/2026/05/29/a-federal-judge-a-1967-law-and-a-billion-rejected-job-applications/))

What makes the case worth reading alongside the field-experiment data above is how familiar the pattern is. The lead plaintiff describes rejections arriving within minutes of applying, sometimes overnight, a pattern that looks less like a recruiter reading a resume and more like a filter sorting by a proxy for age before anyone does. The lawsuit doesn't allege the software was told to reject older applicants. It alleges the software's screening criteria, whatever exactly they are (the algorithms themselves aren't public), reproduce the same disparity researchers had already measured in human-reviewed hiring years earlier. That's the through-line: this isn't a new bias AI invented, it's an old, well-documented one that automated screening now risks running at a scale no single recruiter ever could.

What actually signals age on a resume, to a person or a parser

An ATS doesn't have a field for "age," and it isn't inferring your birthdate. What it, and any human reader, picks up on are proxies, details that correlate strongly with age even though none of them technically state it:

• ▸A graduation year on your degree or certifications, doing simple arithmetic for the reader
• ▸Work history extending back 20, 25, or 30+ years, especially with an early role's start date visible
• ▸Phrasing like "20+ years of experience" in a summary, which reads as a number even when meant as a strength
• ▸Mentions of retired or superseded tools (a specific old software version, a discontinued platform, a defunct employer) that date the era of the experience rather than its relevance
• ▸An email address on a legacy provider associated with an earlier internet era
• ▸A resume opening with an "Objective" statement, a convention that mostly fell out of style in professional resume writing over a decade ago, and reads as dated formatting even before the content is considered
• ▸The line "References available upon request," another convention that's mostly disappeared from current resume-writing guidance

None of these individually proves anything to a screening tool or a person. Together, they're the exact kind of pattern the field experiment above measured the effect of, not an explicit "age" input, but a cluster of correlated signals.

The fixes with real evidence behind them

• ▸Drop the graduation year once your degree is roughly 15 or more years old. This is one of the most consistently recommended, and least controversial, of all age-related resume edits. ([CNBC, "This is the age when you should remove your graduation year from your resume"](https://www.cnbc.com/2018/07/06/heres-when-you-should-remove-your-graduation-year-from-your-resume.html))
• ▸Trim your listed work history to roughly the last 10 to 15 years, covering your two to five most relevant roles, which also happens to be the length guidance most resume advice already converges on for reasons that have nothing to do with age.
• ▸Remove other decade-dating details: certification years outside that same window, references to discontinued tools or platforms, and any employer names that have since been acquired, renamed, or closed, unless the context specifically requires them.
• ▸Modernize your skills section to reflect the tools and platforms you use now, not a historical inventory of everything you've ever touched.
• ▸Update outdated formatting conventions: an objective statement, a centered header, a two-space-after-period typing convention, or a "references available upon request" line all read as dated before a single word of content is judged. ([The Muse, "4 Smart Moves to Age-Proof Your Resume as an Older Worker"](https://www.themuse.com/advice/smart-moves-age-proof-resume-for-older-workers))

What this doesn't mean

None of this is a case for concealing your experience, and it isn't a case for dishonesty about dates, which can backfire badly during a background check or reference call and is a different problem entirely from the one this article is about. Many senior, leadership, and specialist roles explicitly value long tenure and want to see it. The evidence above supports a narrower claim: that a handful of specific, incidental details, a graduation year, a stray decade of extra work history, a dated formatting convention, correlate with worse callback rates for reasons that have nothing to do with a candidate's actual fit for the role. Removing those specific signals isn't about hiding who you are. It's about making sure your actual qualifications, not a handful of proxies for your age, are what gets read first.

Resume Leap's resume audit already checks formatting and length against current conventions before you apply, including flagging an outdated header style, a stale "Objective" section, or a work history that runs well past the length recruiters typically read closely. It's built to catch exactly the kind of incidental, easily-fixed signal this article is about, before an ATS or a recruiter ever gets the chance to weigh it.