How can my race affect my chances of getting hired?

Race can change how employers respond to otherwise comparable applicants. The strongest evidence comes from experiments that vary racial or ethnic cues while holding qualifications constant: large studies generally find disadvantages for Black and other minority applicants, but advantages and null results also occur in particular settings. Race is not an individual hiring-odds score, and the evidence does not support a single ranking that applies to every job. [1–5]

Reviewed November 8, 2026 (America/Chicago). This is a general evaluation of hiring research, not a recommendation to change your identity or a prediction for a particular application.

Overall

The broad pattern is persistent but uneven. Quillian and colleagues’ 2017 US meta-analysis included 28 experiments, 55,842 applications, and 26,326 positions; its core trend analysis used 24 studies since 1989. White applicants received 36% more callbacks than Black applicants and 24% more than Latino applicants on average. Black–White discrimination did not measurably decline over the core period; evidence of improvement for Latinos was modest. [1]

Kline, Rose and Walters’ experiment, released in 2021 and published in 2022, sent more than 83,000 applications to 108 large US employers. Black-sounding names reduced employer contact by 2.1 percentage points. Gaps differed substantially across firms; firms in the highest-discrimination fifth accounted for nearly half of lost contacts to Black applicants. Their 2024 report card graded race and gender gaps for 97 employers from this experiment, rather than conducting an independent 2024 hiring census. [2,6]

International evidence is not a worldwide estimate. Quillian and colleagues’ 2019 meta-analysis covered 97 experiments and over 200,000 applications across nine European/North American countries, finding discrimination against nonwhite natives in every included country and smaller disadvantages for White immigrants. Quillian and Lee’s 2023 six-country synthesis covered 90 studies and roughly 174,000 applications; the institutional summary identifies observations through 2019. Overall racial/ethnic discrimination persisted, although France improved from high initial levels and the Netherlands worsened. These syntheses overlap with earlier studies; they must not be counted as independent replications of every observation. [7,8]

Read the units: “36% more callbacks” is a relative rate, not a 36-percentage-point increase. “2.1 percentage points fewer” is an absolute probability gap. Callback ratios, odds ratios, invitations, any positive contact, and completed hires are not interchangeable. The table preserves the source’s measure and comparison rather than forcing unlike outcomes onto one scale. [1–5,9]

Results by race

One group can have several rows because different experiments ask different questions. “Name” means perceived race/ethnicity, not a verified identity. “Photo” means appearance was manipulated; these experiments are particularly relevant to photo privacy. Values are source-reported unless explicitly identified as rounded source summaries.

Group	Measured effect (callback/hire gap, with number)	Comparison group	Study + year + country	Sample/setting	Signal type	Notes
White	50% more interview callbacks	Black-sounding names	Bertrand & Mullainathan, 2004, US [3]	Boston and Chicago advertised vacancies	Name	Older benchmark, not a current nationwide or tech estimate.
White	36% more callbacks	Black applicants	Quillian et al., 2017, US [1]	28 experiments overall; core analysis since 1989	Mixed across constituent audits	Pooled estimate; not White-men-only.
White	24% more callbacks	Hispanic/Latino applicants	Quillian et al., 2017, US [1]	Same meta-analysis	Mixed	Do not relabel the reciprocal as “24% fewer” for Latinos.
Black	2.1 percentage points fewer employer contacts	White-sounding names	Kline, Rose & Walters, 2021/2022, US [2]	>83,000 applications; 108 large employers	Name	Contact, not completed hiring; substantial firm variation.
Black, elite-college applicants	About 13% response rate, versus nearly 18%	White elite-college applicants	Gaddis, 2015, US [10]	1,008 vacancies; college-selectivity audit	Name	Rounded rates from institutional summary; elite credentials helped both groups but did not erase the gap.
Black men, fully race-neutralized résumé	25.5% callback rate, versus 10.0%	Black-male résumé with explicit racial cues	Kang et al., 2016, US; fieldwork 2015 [11]	1,600 applications; 16 metros; 200 per race × presentation condition	Name + experience wording	Both name and experience changed. This is not a Black-versus-White population estimate; full-whitening gap 15.5 percentage points.
Hispanic/Latino	White comparison group received 24% more callbacks	White applicants	Quillian et al., 2017, US [1]	Pooled US audits	Mixed	Source reports advantage in White-relative-to-Latino direction; no invented inverse percentage.
Asian: Indian, Pakistani and Chinese names pooled	28% fewer callbacks; 4.4 percentage points lower from 15.7% baseline	English-sounding names with Canadian credentials	Oreopoulos, 2011, Canada; fieldwork 2008–2009 [9]	12,910 résumés; 3,225 postings	Name	Pooled across origins; not an Indian-only estimate or current US tech result.
Asian men, Chinese-name profile with race-neutralized résumé	21.0% callback rate, versus 11.5%	Same Asian profile with explicit cues	Kang et al., 2016, US; fieldwork 2015 [11]	Same 1,600-application audit	Name + experience wording	Chinese name Lei Zhang became Luke Zhang; 9.5-percentage-point gap. Do not generalize it to every Asian subgroup.
Black men, junior software applicants	33.5% fewer callbacks; modeled 10.2% vs 15.4%	White men seeking junior roles laterally	Weisshaar et al., 2024, US; fieldwork 2020–2021 [5]	11,190 applications overall; 40 largest metros	Name	Significant disadvantage; does not describe all job levels.
Black women, junior software applicants	25.9% fewer callbacks; modeled 11.4% vs 15.4%	White men seeking junior roles laterally	Weisshaar et al., 2024, US [5]	Same software audit	Name	Significant disadvantage.
White women, junior software applicants	16.8% fewer callbacks; modeled 12.8% vs 15.4%	White men seeking junior roles laterally	Weisshaar et al., 2024, US [5]	Same software audit	Name	Gender-within-race comparison, included to avoid treating all White applicants alike.
White women, senior software applicants	24.0% more callbacks; modeled 30.2% vs 24.4%	White men seeking senior roles laterally	Weisshaar et al., 2024, US [5]	Same software audit	Name	Significant White-female advantage/White-male relative disadvantage; not a pure race effect.
Black men, senior software applicants	14.3% fewer callbacks; modeled 20.9% vs 24.4%	White men seeking senior roles laterally	Weisshaar et al., 2024, US [5]	Same software audit	Name	Not statistically significant at 5%; do not present as a confirmed disadvantage or exact equality.
Black women, senior software applicants	4.7% fewer callbacks; modeled 23.3% vs 24.4%	White men seeking senior roles laterally	Weisshaar et al., 2024, US [5]	Same software audit	Name	Not statistically significant at 5%.
Black appearance	Approximately 20% lower likelihood of employer interest	White appearance, separating region-of-ancestry effects	Polavieja et al., 2023, Germany/Netherlands/Spain [12]	Almost 13,000 applications; young country nationals; varied occupations	Photo, ancestry mainly by name	Rounded institutional-summary estimate. Appearance effect clearer in Germany/Netherlands; Spain more conditional. Not a photo-versus-no-photo experiment.
Asian/Indigenous American appearance (study’s combined category)	Approximately 20% lower likelihood of employer interest	White appearance, separating ancestry effects	Polavieja et al., 2023, three European countries [12]	Same photo experiment	Photo	Appearance category, not an Indian-specific, Asian-American, or US Indigenous hiring estimate.
Dark-skinned Caucasian appearance (study’s terminology)	Approximately 10% lower likelihood of employer interest	White appearance, separating ancestry effects	Polavieja et al., 2023, three European countries [12]	Same photo experiment	Photo	Approximate institutional summary; category often associated with North African appearance. It is not equivalent to every MENA identity.
Turkish-background women, no headscarf	13.5% positive responses, versus 18.8%	German-name women with same bareheaded model	Weichselbaumer, online 2019/issue 2020, Germany [13]	1,474 applications; secretaries/accountants/chief accountants	Name; same model photograph	Ethnicity signal primarily name; both groups had photos.
Turkish-background women, headscarf	4.2% positive responses, versus 13.5% without headscarf and 18.8% German-name baseline	Turkish-name bareheaded and German-name bareheaded women	Weichselbaumer, 2019/2020, Germany [13]	Same office-job audit	Photo attire + name	Intersection of ethnicity/religion; not a clean skin-color-only effect.
White European-origin minorities	20.73% positive contacts, versus 24.55%; majority/minority callback ratio 1.18	White British majority	Di Stasio & Heath, 2021, Britain; fieldwork 2016–2017 [14]	3,195 jobs in full audit; 1,111 in European/majority analytic sample	Name + origin/language wording	Demonstrates White ethnic-minority disadvantage; aggregate rates not all individually significant subgroup results.
Black African/Caribbean, Chinese and South Asian minorities pooled	Majority/minority positive-contact ratio 1.91; interview-only ratio 1.77	White British majority	Di Stasio & Heath, 2021, Britain [14]	Additional nonwhite groups in same full audit	Name + background cues	Pooled reference comparison; do not assign these ratios separately to every group.
Black men, new college graduates	2.1 percentage points fewer callbacks overall; 5.1 points fewer in management	White men	Braun et al., 2026 preprint, US; fieldwork 2016–2017 [15]	36,880 applications; 9,220 ads; 4,969 firms	Name	Recent analysis, old observations; not a 2026 hiring snapshot.
Black women, new college graduates	1.4 percentage points fewer callbacks overall	White men	Braun et al., 2026 preprint, US [15]	Same entry-level audit	Name	Significant pooled callback disadvantage; not a tech-specific estimate.

The photo rows above use the university’s explicitly approximate probability summaries, not odds ratios. They must not be mixed with the exact modeled marginal effects in the journal article. No numerical effect for an unstudied race is implied. [12]

In tech

There are credible findings in both directions for White men. In the 2024 software audit, White men were favored over Black men, Black women, and White women in junior lateral applications. For senior lateral applications, White women received 24% more callbacks than White men. When junior applicants applied upward to senior jobs, modeled rates were 21.1% for White women versus 16.6% for White men; the paper reports a significant 4.5-percentage-point marginal-effect difference. Black applicants’ senior-role differences from White men were not significant at 5%. The latter is not evidence of a general Black preference or proof of equality. [5]

The experiment compared Black and White names; it did not estimate Asian/Indian or Hispanic callbacks. It did not randomize DEI policies or photos. Interviews supported the authors’ interpretation that diversity value influenced screening, but a screening advantage for White women cannot establish a racial penalty against White people as a whole. Pandemic-era data from 2020–2021 do not settle the late-2026 market. [5]

Cross-national evidence adds context rather than a universal answer. The GEMM project harmonized 19,181 applications across 53 ethnic groups in five European countries; its British audit included software developers. Di Stasio and Heath found the pooled EU-background disadvantage concentrated in non-graduate occupations; the software-developer majority/minority callback ratio was 0.92, descriptively favoring EU-background applicants, but the high-skilled advantage was not statistically significant. Small occupation-specific subgroups require caution. [14,16]

Each bar is a relative difference in callbacks reported by one study, against the comparison group named on the left. Bars are not averaged across studies.

Limits

Screening is not hiring. These numbers mainly measure employer contact or interview invitations, not offers or completed hires; no conversion to a final hiring probability is warranted. [2,3,5,9,11,14,15]
Signals differ. Names may signal race, ancestry, class, or familiarity. Photographs can separately signal appearance; photo-attire studies can also signal religion. These are not interchangeable treatments or direct tests of a confidential demographic-form answer. [3,9,11–14]
No group ranking. “White,” “Asian,” “Black,” and “Hispanic” include many populations. Country-of-origin groups, perceived appearance, race, sex and religion overlap but are not the same variable. No verified standalone American Indian/Alaska Native or Pacific Islander estimate was established in this review; the European study’s combined appearance category does not fill that gap. [8,9,12,14]
Firm and role heterogeneity matter. A large average gap does not mean every employer discriminates, and an insignificant estimate does not prove zero discrimination. [2,5]
DEI labels are not outcomes. Kang and colleagues found no evidence that pro-diversity job-ad language reduced the disadvantage of unwhitened résumés in their audit. This is not evidence that all diversity programs cause discrimination, nor that modern DEI policies explain every callback gap. [11]
Recency is not fieldwork date. The 2026 preprint reanalyzes 2016–2017 applications; the 2024 software paper uses 2020–2021 applications. Neither is a real-time late-2026 census. The preprint is supplementary evidence, not equivalent in publication status to the peer-reviewed audits and syntheses. [5,15]
Chart contract: race-chart.json contains only source-stated relative callback percentage differences, positive for more and negative for fewer, against the stated comparator. It deliberately excludes percentage-point gaps, raw callback levels, ratios, approximate press-summary photo values, and non-significant point estimates. No reciprocal percentage has been calculated. Not all table rows are chartable on that scale. Facet by study/setting; do not average the chart values into racial scores. The White-women senior advantage is a sex-within-race comparison, not a racial effect. [1,3,5,9]

Sources

All URLs returned readable content through WebFetch during this November 8, 2026 research session. Readable institutional summaries replace publisher pages that timed out; no failed or 404 URL is relied on.

Quillian, Pager, Hexel & Midtbøen (2017), US meta-analysis, publisher abstract reproduced by Russell Sage Foundation: https://www.russellsage.org/research/outputs/meta-analysis-field-experiments-shows-no-change-racial-discrimination-hiring-over
Kline, Rose & Walters (NBER 2021; QJE 2022), Systemic Discrimination Among Large U.S. Employers: https://www.nber.org/papers/w29053
Bertrand & Mullainathan (NBER 2003; AER 2004), Are Emily and Greg More Employable than Lakisha and Jamal?: https://www.nber.org/papers/w9873
Oreopoulos (2011), institutional summary of Canadian audit: https://www.povertyactionlab.org/evaluation/discrimination-against-skilled-immigrants-canadian-labor-market
Weisshaar, Chavez & Hutt (2024), Hiring Discrimination Under Pressures to Diversify, full text and Table 2: https://journals.sagepub.com/doi/10.1177/00031224241245706
Kline, Rose & Walters (2024), A Discrimination Report Card: https://www.nber.org/papers/w32313
Quillian et al. (2019), nine-country meta-analysis: https://sociologicalscience.com/articles-v6-18-467/
Quillian & Lee (2023), six-country trends; Northwestern institutional summary: https://news.northwestern.edu/stories/2023/01/racial-discrimination-in-hiring-remains-a-persistent-problem-northwestern-study?fj=1
Oreopoulos (2011), Canadian audit, J-PAL summary (same source as [4]): https://www.povertyactionlab.org/evaluation/discrimination-against-skilled-immigrants-canadian-labor-market
Gaddis (2015), author abstract and University of Michigan numerical summary: https://stevenmichaelgaddis.com/pubs/discrimination-in-the-credential-society-an-audit-study-of-race-and-college-selectivity-in-the-labor-market/ ; https://news.umich.edu/african-americans-with-elite-college-degrees-have-little-advantage-in-job-market/
Kang, DeCelles, Tilcsik & Jun (2016), Whitened Résumés, author-hosted full text, especially Figures 1–2: https://www.sonia-kang.com/pdfs/Whitened%20resumes:%20Race%20and%20self-%20presentation%20in%20the%20labor%20market.pdf
Polavieja, Lancee, Ramos, Veit & Yemane (2023), In Your Face, UC3M institutional summary of photo experiment: https://www.uc3m.es/ss/Satellite/UC3MInstitucional/en/Detalle/Comunicacion_C/1371356488122/1371216001705/A_study_analyses_racial_discrimination_in_job_recruitment_in_Europe
Weichselbaumer (online 2019; issue 2020), Multiple Discrimination against Female Immigrants Wearing Headscarves, full text, Tables 2–5: https://journals.sagepub.com/doi/10.1177/0019793919875707
Di Stasio & Heath (2021), London Calls?, full text and Tables 1–3: https://www.frontiersin.org/journals/sociology/articles/10.3389/fsoc.2021.737857/full
Braun et al. (2026), Hiring Discrimination and the Task Content of Jobs, preprint, version dated August 24, 2026; fieldwork 2016–2017: https://arxiv.org/html/2604.01933
GEMM project documentation and publication list: https://www.bramlancee.eu/gemm-study/

White	50%	Black-sounding names	Bertrand & Mullainathan 2004, US
White	36%	Black applicants	Quillian et al. 2017, US meta-analysis
White	24%	Hispanic/Latino applicants	Quillian et al. 2017, US meta-analysis
Asian: Indian/Pakistani/Chinese names pooled	-28%	English-sounding names with Canadian credentials	Oreopoulos 2011, Canada
Black men	-33.5%	White men	Weisshaar et al. 2024, US, junior-to-junior software roles
Black women	-25.9%	White men	Weisshaar et al. 2024, US, junior-to-junior software roles
White women	-16.8%	White men	Weisshaar et al. 2024, US, junior-to-junior software roles
White women	24%	White men	Weisshaar et al. 2024, US, senior-to-senior software roles