{ "researchDate": "2026-10-07", "revision": "draft-3", "locale": "en", "presentation": "overview", "assessment": "Qualified explanation · research draft", "reviewNeeds": "Source verification, subject-matter and editorial approval pending.", "updateTriggers": "New studies, corrected sources or evidence changing the stated finding.", "related": [ { "slug": "misinformation-disinformation", "label": "Checking information and intent separately" } ], "id": "dei-qualifications", "slug": "dei", "topic": "Hiring", "tag": "DEI & qualifications", "title": "DEI does not mean unqualified", "signLines": [ "DEI does not mean", "unqualified" ], "question": "What can a DEI label tell us about someone’s qualifications?", "summary": "Qualified applicants can be overlooked for reasons that have little to do with their skills. DEI efforts aim to reduce those barriers and give people a fair chance. A DEI label does not tell you whether someone can do the job.", "scope": "DEI means diversity, equity and inclusion. These efforts take many forms, and they do not all work equally well.", "geographicScope": { "kind": "jurisdiction", "countryCodes": [ "US", "FR" ], "jurisdictionCodes": [ "US", "FR" ], "note": "Primarily US hiring; the anonymous-résumé experiment is French. Results are bounded by study dates, employers and outcomes." }, "evidence": [ { "heading": "Why these efforts exist", "text": "Imagine two applicants who meet a job’s requirements. If one is passed over because of their name or race, their skills may never get a fair hearing. Hiring studies show that equally qualified applicants can receive different treatment. Reducing those barriers is one reason for DEI efforts.", "sources": [ 1 ], "sourceIds": [ "quillian-2017" ] }, { "heading": "A fairer process can still have high standards", "text": "A job still needs people with the right skills. Employers can reach a wider range of applicants, ask everyone comparable interview questions and judge answers against the same job requirements. Research supports carefully planned interviews as a useful way to assess people; it does not show that every interview removes bias.", "sources": [ 2, 3, 7 ], "sourceIds": [ "opm-structured-interviews", "sackett-2022", "sackett-2023" ] }, { "heading": "What hiring tests found", "text": "A review of US hiring experiments compared employers’ replies to applications with matching qualifications. In studies conducted since 1989, white applicants received 36% more positive replies, such as interview invitations, than Black applicants. For illustration, 36% more than 100 replies would be 136; those are not the studies’ actual counts. The research measured employer replies, not performance at work.", "sources": [ 1 ], "sourceIds": [ "quillian-2017" ] }, { "heading": "Programs need to be judged by their results", "text": "Different efforts have different results. A study of 3,016 employees found that a short diversity training changed some attitudes but had mostly limited effects on behavior. In a French experiment, hiding applicants’ names and other identifying information reduced minority applicants’ chances of interviews and hiring at the firms taking part. Neither result tells us how every DEI program works.", "sources": [ 4, 5 ], "sourceIds": [ "chang-2019", "behaghel-2015" ] }, { "heading": "Qualifications and job performance are not the same thing", "text": "An older study in four US metropolitan areas found lower educational qualifications among some people hired through affirmative-action programs. Most groups did not show worse measured job performance. That finding does not prove equal performance everywhere, but it shows why a degree or test score is not the whole picture.", "sources": [ 6 ], "sourceIds": [ "holzer-neumark-1999" ] } ], "takeaway": "Judge someone by the job’s requirements and the evidence about their skills. A DEI label cannot replace that assessment.", "limits": [ "Evidence of unfair hiring does not prove that every effort to fix it works.", "An invitation to interview, a degree and performance at work tell us different things.", "These studies cannot tell us whether a particular employee was qualified, favored or chosen over someone more capable." ], "contextTitle": "What does “qualified” mean?", "contextIntro": "Meeting a job’s requirements is different from getting the highest score on one test.", "detail": "To judge a particular hiring decision, you need to know what the job requires, how the candidates were assessed and how consistently the rules were applied. The mix of people hired cannot answer those questions on its own. These studies cover particular employers and years, not every employer today. The hiring review’s estimate of 36% more replies had a 95% confidence interval of 25–47%, showing uncertainty about the size of the difference.", "sources": [ { "id": "quillian-2017", "revision": "research-2026-10-07", "title": "Meta-analysis of field experiments shows no change in racial discrimination in hiring over time", "publisher": "PNAS, 2017", "url": "https://www.russellsage.org/sites/default/files/10870.full_.pdf", "type": "Meta-analysis of hiring experiments", "access": "Full funder-hosted paper inspected in supplied dossier", "limit": "28 US experiments overall; main post-1989 analysis 24 studies. Callback outcomes, not employee performance.", "retrievedAt": "2026-10-06", "locale": "en", "geographicCoverage": { "kind": "undetermined", "countryCodes": [], "jurisdictionCodes": [], "note": "28 US experiments overall; main post-1989 analysis 24 studies. Callback outcomes, not employee performance." }, "readerTitle": "Hiring tests with equally qualified applicants", "readerAccess": "We could read the full paper.", "readerLimit": "The review covers 28 US experiments; its main comparison uses 24 studies conducted since 1989. It measures employers’ replies, not performance at work or how DEI programs work today.", "supports": [ "Why these efforts exist", "What hiring tests found" ], "limitationsInternal": [ "Full funder-hosted paper inspected in supplied dossier", "28 US experiments overall; main post-1989 analysis 24 studies. Callback outcomes, not employee performance." ], "limitationsPublic": [ "The review covers 28 US experiments; its main comparison uses 24 studies conducted since 1989. It measures employers’ replies, not performance at work or how DEI programs work today." ] }, { "id": "opm-structured-interviews", "revision": "research-2026-10-07", "title": "Structured Interviews", "publisher": "US Office of Personnel Management", "url": "https://www.opm.gov/policy-data-oversight/assessment-and-selection/other-assessment-methods/structured-interviews/", "type": "Assessment guidance", "access": "Full official page inspected in supplied dossier", "limit": "Operational guidance, not a new randomized trial or proof of eliminating all bias.", "retrievedAt": "2026-10-06", "locale": "en", "geographicCoverage": { "kind": "undetermined", "countryCodes": [], "jurisdictionCodes": [], "note": "Operational guidance, not a new randomized trial or proof of eliminating all bias." }, "readerTitle": "Asking comparable job interview questions", "readerAccess": "We could read the full guidance page.", "readerLimit": "This explains a way to assess candidates. It is not a new experiment or proof that every interview removes bias.", "supports": [ "A fairer process can still have high standards" ], "limitationsInternal": [ "Full official page inspected in supplied dossier", "Operational guidance, not a new randomized trial or proof of eliminating all bias." ], "limitationsPublic": [ "This explains a way to assess candidates. It is not a new experiment or proof that every interview removes bias." ] }, { "id": "sackett-2022", "revision": "research-2026-10-07", "title": "Revisiting meta-analytic estimates of validity in personnel selection", "publisher": "Journal of Applied Psychology, 2022", "url": "https://pubmed.ncbi.nlm.nih.gov/34968080/", "type": "Selection-validity reassessment", "access": "2022 abstract and 2023 author follow-up full text inspected in dossier", "limit": "Predictive validity is distinct from causal effects on discrimination; effects vary by setting.", "retrievedAt": "2026-10-06", "locale": "en", "geographicCoverage": { "kind": "undetermined", "countryCodes": [], "jurisdictionCodes": [], "note": "Predictive validity is distinct from causal effects on discrimination; effects vary by setting." }, "readerTitle": "Which hiring assessments predict job performance?", "readerAccess": "We could read the 2022 summary and a full follow-up published in 2023.", "readerLimit": "Predicting performance is different from showing that a process reduces discrimination. Results vary between jobs and employers.", "publicationTitle": "Revisiting meta-analytic estimates of validity in personnel selection: Addressing systematic overcorrection for restriction of range.", "publicationDoi": "10.1037/apl0000994", "bibliographyCheckedAt": "2026-10-07", "readerBlurb": "Reassesses how well hiring assessments predict work performance. Predicting performance is different from showing that hiring becomes less discriminatory.", "supports": [ "A fairer process can still have high standards" ], "limitationsInternal": [ "2022 abstract and 2023 author follow-up full text inspected in dossier", "Predictive validity is distinct from causal effects on discrimination; effects vary by setting." ], "limitationsPublic": [ "Reassesses how well hiring assessments predict work performance. Predicting performance is different from showing that hiring becomes less discriminatory." ] }, { "id": "chang-2019", "revision": "research-2026-10-07", "title": "The mixed effects of online diversity training", "publisher": "PNAS, 2019", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC6475398/", "type": "Preregistered randomized experiment", "access": "Full text inspected in supplied dossier", "limit": "One organization; voluntary participants; behavior followed up to 20 weeks. No demonstrated backlash in this trial.", "retrievedAt": "2026-10-06", "locale": "en", "geographicCoverage": { "kind": "undetermined", "countryCodes": [], "jurisdictionCodes": [], "note": "One organization; voluntary participants; behavior followed up to 20 weeks. No demonstrated backlash in this trial." }, "readerTitle": "A test of online diversity training", "readerAccess": "We could read the full study.", "readerLimit": "Employees at one organization volunteered for the study and were followed for up to 20 weeks. Most measured behavior did not change. The study did not find that training caused a backlash.", "supports": [ "Programs need to be judged by their results" ], "limitationsInternal": [ "Full text inspected in supplied dossier", "One organization; voluntary participants; behavior followed up to 20 weeks. No demonstrated backlash in this trial." ], "limitationsPublic": [ "Employees at one organization volunteered for the study and were followed for up to 20 weeks. Most measured behavior did not change. The study did not find that training caused a backlash." ] }, { "id": "behaghel-2015", "revision": "research-2026-10-07", "title": "Unintended Effects of Anonymous Résumés", "publisher": "AEJ: Applied Economics, 2015", "url": "https://doi.org/10.1257/app.20140185", "type": "French randomized hiring experiment", "access": "Published abstract and full 2014 author manuscript inspected in dossier; final methods not separately compared line by line", "limit": "Participating firms self-selected. Adverse findings do not generalize to all anonymous recruitment.", "retrievedAt": "2026-10-06", "locale": "en", "geographicCoverage": { "kind": "undetermined", "countryCodes": [], "jurisdictionCodes": [], "note": "Participating firms self-selected. Adverse findings do not generalize to all anonymous recruitment." }, "readerTitle": "What happened when French firms hid names on résumés?", "readerAccess": "We could read the published summary and an earlier full author version. We did not compare every method with the final paper.", "readerLimit": "The firms volunteered to take part. The result does not show how anonymous recruitment works at every employer.", "supports": [ "Programs need to be judged by their results" ], "limitationsInternal": [ "Published abstract and full 2014 author manuscript inspected in dossier; final methods not separately compared line by line", "Participating firms self-selected. Adverse findings do not generalize to all anonymous recruitment." ], "limitationsPublic": [ "The firms volunteered to take part. The result does not show how anonymous recruitment works at every employer." ] }, { "id": "holzer-neumark-1999", "revision": "research-2026-10-07", "title": "Are Affirmative Action Hires Less Qualified? Evidence from Employer–Employee Data on New Hires", "publisher": "Journal of Labor Economics, 1999", "url": "https://www.journals.uchicago.edu/doi/10.1086/209930", "type": "Observational employer–employee study", "access": "Published abstract only; manuscript retrieval failed", "limit": "Old four-metro-area observational evidence. No detected difference is not equivalence proof.", "retrievedAt": "2026-10-06", "locale": "en", "geographicCoverage": { "kind": "undetermined", "countryCodes": [], "jurisdictionCodes": [], "note": "Old four-metro-area observational evidence. No detected difference is not equivalence proof." }, "readerTitle": "Qualifications and performance among affirmative-action hires", "readerAccess": "We could read only the published study summary, not the full paper.", "readerLimit": "This is an older study of four US metropolitan areas. Not finding a performance difference does not prove the groups were exactly the same.", "supports": [ "Qualifications and job performance are not the same thing" ], "limitationsInternal": [ "Published abstract only; manuscript retrieval failed", "Old four-metro-area observational evidence. No detected difference is not equivalence proof." ], "limitationsPublic": [ "This is an older study of four US metropolitan areas. Not finding a performance difference does not prove the groups were exactly the same." ] }, { "id": "sackett-2023", "locale": "en", "geographicCoverage": { "kind": "undetermined", "countryCodes": [], "jurisdictionCodes": [], "note": "Predictive validity is distinct from causal effects on discrimination; effects vary by setting." }, "revision": "2026-10-07", "title": "Revisiting the design of selection systems in light of new findings regarding the validity of widely used predictors", "publisher": "Industrial and Organizational Psychology, 2023", "url": "https://doi.org/10.1017/iop.2023.24", "type": "Author follow-up to the selection-validity reassessment", "access": "Full 2023 author follow-up inspected in original supplied dossier; publisher bibliographic metadata and abstract verified 2026-10-07", "limit": "An interpretation of the earlier reassessment, not a separate discrimination experiment. Validity varies across settings.", "retrievedAt": "2026-10-07", "readerTitle": "Follow-up about using the revised hiring-assessment findings", "readerAccess": "Publisher metadata and summary verified; earlier full text retained in the original research dossier.", "readerLimit": "Explains why results vary across settings. It follows the 2022 reassessment rather than providing a separate hiring-discrimination experiment.", "readerBlurb": "Explains how to use the revised findings and why results vary across settings. It is a follow-up to the 2022 paper, not another hiring-discrimination experiment.", "supports": [ "A fairer process can still have high standards" ], "limitationsInternal": [ "Full 2023 author follow-up inspected in original supplied dossier; publisher bibliographic metadata and abstract verified 2026-10-07", "An interpretation of the earlier reassessment, not a separate discrimination experiment. Validity varies across settings." ], "limitationsPublic": [ "Explains how to use the revised findings and why results vary across settings. It is a follow-up to the 2022 paper, not another hiring-discrimination experiment." ] } ], "dossier": "dei_qualifications_evidence.md", "shareContext": "DEI efforts aim to reduce unfair barriers. A person’s skills and a program’s results still need to be assessed.", "socialContext": "A DEI label cannot tell you someone’s skills. Different programs have different results.", "copyReviewedAt": "2026-10-07", "readerCopyEdition": "plain-language-1", "citationDesignEdition": "publication-first-1", "assetDesignEdition": "space-first-1", "headline": { "text": "DEI does not mean unqualified", "revision": 1, "status": "draft", "sourceMessageId": null, "approvedAt": null, "authority": "Existing owner decisions recorded in AGENTS/D39; unavailable message IDs are not invented." }, "language": "en", "jurisdiction": { "kind": "jurisdiction", "countryCodes": [ "US", "FR" ], "jurisdictionCodes": [ "US", "FR" ], "note": "Primarily US hiring; the anonymous-résumé experiment is French. Results are bounded by study dates, employers and outcomes." }, "claimScope": "DEI means diversity, equity and inclusion. These efforts take many forms, and they do not all work equally well.", "research": { "dossierRef": "docs/research/dei_qualifications_evidence.md", "reviewedAt": "2026-10-07", "reviewDue": "New studies, corrected sources or evidence changing the stated finding.", "contraryEvidenceConsidered": [], "status": "research draft; editorial/context review separate" }, "status": "researched draft preview", "realWorldClaimReviewed": false, "clinicalEditorialApproved": false }