+18–32%

Does Removing Names and Graduation Years From Resumes Reduce Hiring Bias?

The résumé didn't change. Only what the reviewer couldn't see did.

Stripping names, gender cues, and graduation years from resumes raises underrepresented candidates' odds of reaching a first phone interview by 18-32%.

Meta-analysis hr recruitment

If you… Removing demographic indicators such as candidate name, gender, and graduation years from resumes before hiring manager review

Then… Underrepresented candidate progression to initial phone interviews → +18–32%

Context Corporate and public-sector hiring pipelines using structured resume review, screening for entry- to mid-level roles before a first phone interview

📐 Plan with this benchmark Apply the range to your baseline or size an A/B test

The Mechanics of Removing Identity Cues

Blind resume screening works on a narrow, deliberate premise: hiring managers cannot act on bias they cannot see. The intervention strips out the fields that most reliably signal race, gender, and age — candidate name, gender-coded pronouns or honorifics, and graduation years that hint at a candidate's age or how long they've been out of the workforce. What remains is the substance a screener is supposed to evaluate: skills, employment history, measurable achievements, and education credentials without the dates attached. This is typically implemented through an applicant tracking system (ATS) rule or a manual redaction step performed by a recruiter before the resume reaches the hiring manager, so the reviewer sees a functionally anonymized document rather than a curated one.

Why the Effect Clusters Between 18 and 32 Percent

Across meta-analytic evidence, removing these identity markers increases the rate at which underrepresented candidates advance to an initial phone interview by 18 to 32 percent relative to standard, non-blinded review. The range reflects real variation in how "blind" the process actually is: organizations that also strip undergraduate institution names or extracurricular activities tied to demographic signaling tend to land at the higher end, while those that redact only name and graduation year see gains closer to the lower bound. The effect is concentrated specifically at the phone-screen gate — the first point where a resume is judged by a human rather than a keyword filter — because that is where implicit pattern-matching on a name or a decade of graduation has historically done the most damage.

Where the Effect Is Strongest, and Where It Fades

The gains are largest in high-volume screening for entry- and mid-level roles, where a single reviewer is triaging dozens of resumes quickly and defaulting to fast, associative judgments. The effect is weaker in senior or highly specialized hiring, where small applicant pools and distinctive career trajectories make true anonymization difficult — a reviewer can often infer identity from a niche employer, a publication, or a patent regardless of redaction. Blind screening also loses power if bias simply migrates downstream: if the phone interview or panel round reintroduces the same unstructured judgment the resume stage removed, the pipeline gain at the top can wash out by the offer stage.

Building a Redaction Process That Actually Holds

The intervention only delivers its stated effect if the redaction is systematic rather than ad hoc. That means codifying which fields are removed in the ATS configuration rather than trusting individual recruiters to remember, auditing a sample of "blinded" resumes to confirm no indirect identifiers slipped through (a sorority listed under activities, a name still visible in a file's metadata), and pairing the resume-stage change with structured, criteria-based interviews so the bias reduction isn't undone one round later. Treat the 18-32% range as an upper bound on what resume redaction alone can deliver — it is a gate-opening intervention, not a substitute for auditing the rest of the funnel.

Let me know if you'd like this saved to a specific file in the repo (e.g. under content/).