Do names still change job callback rates?

Yes. A name can still change whether an otherwise equivalent job application gets a response.

But the useful answer is no longer one famous number from 2004.

A U.S. synthesis found no decline in discrimination against African-American applicants through 2015. A later audit sent more than 83,000 applications and showed that the average gap was concentrated among particular employers. Newer experiments then found that the penalty could be much larger for applicants who had not yet had the chance to gain experience. In Finland, a repeated study found real improvement for some origin signals between 2016 and 2024—without reaching equal treatment.

Names still matter. How much they matter depends on the employer, the market, the signal and the point in a career.

The experiment changes the name, not the qualifications

The strongest evidence comes from résumé correspondence experiments.

Researchers create realistic applications, keep qualifications comparable and randomly assign names across them. The applications go to real vacancies. Researchers then record whether an employer calls or emails about an interview.

Random assignment makes the average comparison causal within the experiment: differences in response rates between otherwise comparable applications can be attributed to the randomized name signal.

That last word matters. A name is not race or ethnicity itself. It can prompt assumptions about race, national origin, gender, social class, generation or familiarity at the same time. The experiment measures the treatment of that perceived signal—not an essential quality of the applicant.

And a callback is only the first gate. It is not a job offer, salary or career.

The famous 50% result was a 3.2-point gap

In the study that still dominates search results, Marianne Bertrand and Sendhil Mullainathan sent 4,870 fictitious résumés to job ads in Boston and Chicago.

Applications with White-sounding names received callbacks 9.65% of the time. Equivalent applications with African-American-sounding names received them 6.45% of the time.

Both ways of describing the result are correct:

• the absolute difference was 3.20 percentage points;
• the White-name callback rate was 50% higher in relative terms.

The difference is not cosmetic. Saying only 50% can sound as if half of all applications were affected or as if the gap were 50 percentage points. The actual rates translated to roughly one callback per ten applications versus one per fifteen.

The experiment also varied résumé quality. Stronger credentials generated more callbacks for White-signaled applications but did not close the racial gap. The study did not follow candidates into offers or wages, and its newspaper-ad channel does not represent every modern hiring system.

It remains a landmark result. It should not remain the entire answer.

Across U.S. field experiments, the gap did not fade

Lincoln Quillian and colleagues assembled 28 U.S. field experiments with 55,842 applications to 26,326 positions.

Their central time analysis covered 24 studies conducted from 1989 through 2015. Across that period, White applicants received on average 36% more callbacks than equally qualified African-American applicants. The estimate did not decline over time.

For Latino applicants, the average White callback advantage was 24%. There was some evidence of improvement, but it was less stable across analytical choices and based on only nine experiments.

The meta-analysis rejects a comforting interpretation of the 2004 study: the effect was not simply an artifact of one city pair, one research team or one moment.

It also prevents another mistake. An average ratio across studies is not a law of hiring. Baseline callback rates and absolute gaps vary. So do employers.

The average hides which employers create the loss

Patrick Kline, Evan Rose and Christopher Walters sent more than 83,000 applications to 108 of the largest U.S. employers.

Distinctively Black names reduced the probability of employer contact by 2.1 percentage points on average relative to distinctively White names.

That is smaller than the 3.2-point gap in the 2004 experiment. It is not a clean story of progress.

The gap varied sharply between companies. The between-company standard deviation was 1.9 points—almost as large as the average effect itself. Employers in the top quintile of measured discrimination accounted for nearly half of the contacts lost by Black applicants in the experiment.

After controlling the false-discovery rate at 5%, the researchers identified 23 companies as discriminating against Black applicants. Company-level gaps were persistent across time and geography inside the experiment, while broad industry differences explained about half of the variation between firms.

One organization-wide average can therefore be reassuring and incomplete at the same time. A company can look close to balanced overall while one job family, location or screening process produces a concentrated loss of access.

The first rejection can shape the résumé used later

A callback gap does not end at one application. It can change who gets the experience that the next employer rewards.

A 2025 experiment tested that mechanism in entry-level automotive jobs. The researchers independently randomized a White- or Black-signaling name and whether the résumé listed relevant prior experience across 1,001 applications.

The overall callback rates were 28.3% for White-signaled applications and 20.0% for Black-signaled applications: an 8.3-point gap.

The split by experience was more revealing.

The 12.4-point gap among inexperienced applicants was statistically significant. The 4.3-point gap among experienced applicants was not (p = .20). Across all applicants, relevant experience raised callbacks by 9.5 points, or 49% relative to the inexperienced rate.

In the authors’ model, these estimates can create a loop: discrimination at the experience-entry gate reduces modeled access to experience, which the next round rewards.

The researchers combined the callback experiment, a survey of 107 hiring managers and local vacancy counts. If applicants applied to all eight openings in the model’s median market, White applicants were about 7.5 percentage points more likely than Black applicants to be hired in round two. Unequal first-round access to experience accounted for about 44%—nearly half—of that modeled gap.

That number is a model-based decomposition, not a directly observed lifetime career. Its value is structural: an apparently smaller penalty later in a career can coexist with a larger total effect because the earlier screen has already changed who arrives with experience.

Finland shows that change can be real and incomplete

The newest evidence asks a different question: what happens when the same field design returns eight years later?

Akhlaq Ahmad compared two experiments conducted in Finland in 2016 and 2024. Each sent five matched applications to 1,000 vacancies. Across both periods, 10,000 applications went to 2,000 jobs in restaurant and catering, retail, cleaning, clerical and customer-service work.

The applications signaled Finnish, English, Russian, Iraqi or Somali origin. This was not a name-only test: cover letters also stated a corresponding mother tongue. Qualifications, Finnish education and work experience were held equivalent.

For Somali-signaled applicants, the relative callback rate fell from 3.94 in 2016 to 2.48 in 2024. In other words, the estimated application burden needed to match the Finnish-signaled callback count declined from almost four times to about two and a half times. For Iraqi-signaled applicants, it fell from 2.91 to 2.06. Both reductions in the study’s net discrimination rate were statistically significant.

That is evidence of improvement.

It is also evidence of persistence. In 2024, Somali- and Iraqi-signaled applications still needed more than twice as many attempts to produce a comparable number of callbacks. English-signaled applications changed little. The Russian-signaled gap increased.

One country, five origin signals and five entry occupations cannot establish a global trend. The repeated design does establish something more useful: hiring discrimination can change without disappearing, and different groups can move in different directions in the same labor market.

Five studies, five parts of the answer

Changing a name is not a fairness system

Search results often turn this evidence into a question for applicants: should you abbreviate, anglicize or replace your name?

The experiments do not support a guarantee. They report average differences in specific markets. A changed name can alter one signal while an address, education, language or interview introduces another. It can also impose an identity cost on the person facing unequal treatment.

An applicant may make that choice. It should not become the solution demanded by the evidence.

The controlled variable sits on the résumé. The causal response occurs in the hiring process.

Original sources

• Bertrand and Mullainathan, 2004
• Quillian et al., 2017
• Kline, Rose and Walters, 2022
• Bohren, Hull and Imas, 2025
• Ahmad, 2026

Independent editorial summary. The authors, journals, universities and employers are not affiliated with LifeScore.