Gender Bias in Large Language Model Brand Recommendations: A Three-Study Analysis of Prompt-Induced Disparities Across Seasonal and Recipient Contexts

Gender Bias in Large Language Model Brand Recommendations: A Three-Study Analysis of Prompt-Induced Disparities Across Seasonal and Recipient Contexts | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Gender Bias in Large Language Model Brand Recommendations: A Three-Study Analysis of Prompt-Induced Disparities Across Seasonal and Recipient Contexts Dmitrij Żatuchin This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8883056/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Large language models (LLMs) are increasingly used by consumers for product recommendations, yet their potential for perpetuating gender bias in commercial contexts remains understudied. We present a three-study analysis of gender-based disparities in brand recommendations across three major LLMs: Google Gemini 3 Flash, OpenAI GPT-5.2, and xAI Grok-4-1. Study 1 ( \((n = 299)\) queries, December 2025) examined adult gift recipients (husband/wife/partner); Study 2 ( \((n = 480)\) queries, January 2026) examined child recipients (son/daughter/child); Study 3 ( \((n = 500)\) queries, February 2026) examined romantic-occasion recipients (husband/wife/boyfriend/girlfriend/partner) in a Valentine's Day context. Combined analysis of 1,279 queries reveals that gender bias is not only systematic but context-dependent : Christmas-framed queries produce 28--61% fewer brands for female-targeted prompts, while Valentine's Day framing reverses this pattern, with female-framed queries receiving 8.8% more brands from Gemini. Chi-square analysis confirms systematic gender-category associations across all three studies (Study 1: \((\chi^2 = 137.32)\) ; Study 2: \((\chi^2 = 524.32)\) ; both \((p < 0.001)\) ; Cram\'{e}r's \((V = 0.23)\) -- \((0.38)\) ). We identify ``category gatekeeping'' as the underlying bias mechanism, document 69 gender-locked brands across all studies, and introduce the Prompt-Adjusted Share of Recommendation (PASOR) metric with empirical validation. Cross-model agreement remains consistently low (Jaccard \((0.00)\) -- \((0.68)\) ). These findings establish that gender bias in LLM recommendations is systematic, platform-dependent, and modulated by seasonal context, with implications for AI fairness, open-world machine learning, and data-driven consumer decision-making. Large language models Gender bias Brand recommendations AI fairness Open-world learning Consumer AI Data-driven decision-making Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8883056","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":592070174,"identity":"ce01ac6c-d3e1-4fcc-8c9c-17aac620d0f6","order_by":0,"name":"Dmitrij Żatuchin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA90lEQVRIiWNgGAWjYDCCA4wNDAwGbBAOY4MEAz8DAzMhLY0NEC3MEC2SDQS1MICsYYBpAdp4gIAWvmuH2x9XFPAxyLefPybxc4dFnvHx5sMGDBU2OLVI3k5sbDwDdBhjTzKbZO8ZiWKzM8eSExjOpOHUYgDS0gDUwsyQzCbN2CaRuO1GjvEBxrbDhLWw8T+GaNk8//3nA4z//hPWwiMBtWWDBA9zAmPDAbx+mQnUwiMh8djYsheoZcaZNGODhGPJOLXw3U5/8LHhzzE5+f7Ehzd+ttUl9rcffizxocYOpxYoOMaDyk8gpIGBoYawklEwCkbBKBi5AAASpFQDP8oRowAAAABJRU5ErkJggg==","orcid":"","institution":"Estonian Entrepreneurship University of Applied Sciences (EUAS)","correspondingAuthor":true,"prefix":"","firstName":"Dmitrij","middleName":"","lastName":"Żatuchin","suffix":""}],"badges":[],"createdAt":"2026-02-15 01:23:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8883056/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8883056/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104767447,"identity":"77935410-5134-4092-b9f4-ecd7b32a5e40","added_by":"auto","created_at":"2026-03-17 03:56:05","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":320727,"visible":true,"origin":"","legend":"","description":"","filename":"genderbiasllmrecommendations2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8883056/v1_covered_e61dac95-e4ff-4c94-98ae-c481d967bf47.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Gender Bias in Large Language Model Brand Recommendations: A Three-Study Analysis of Prompt-Induced Disparities Across Seasonal and Recipient Contexts","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Large language models, Gender bias, Brand recommendations, AI fairness, Open-world learning, Consumer AI, Data-driven decision-making","lastPublishedDoi":"10.21203/rs.3.rs-8883056/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8883056/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eLarge language models (LLMs) are increasingly used by consumers for product recommendations, yet their potential for perpetuating gender bias in commercial contexts remains understudied. We present a three-study analysis of gender-based disparities in brand recommendations across three major LLMs: Google Gemini 3 Flash, OpenAI GPT-5.2, and xAI Grok-4-1. Study 1 (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((n = 299)\\)\u003c/span\u003e\u003c/span\u003e queries, December 2025) examined adult gift recipients (husband/wife/partner); Study 2 (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((n = 480)\\)\u003c/span\u003e\u003c/span\u003e queries, January 2026) examined child recipients (son/daughter/child); Study 3 (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((n = 500)\\)\u003c/span\u003e\u003c/span\u003e queries, February 2026) examined romantic-occasion recipients (husband/wife/boyfriend/girlfriend/partner) in a Valentine's Day context. Combined analysis of 1,279 queries reveals that gender bias is not only systematic but \u003cem\u003econtext-dependent\u003c/em\u003e: Christmas-framed queries produce 28--61% fewer brands for female-targeted prompts, while Valentine's Day framing \u003cem\u003ereverses\u003c/em\u003e this pattern, with female-framed queries receiving 8.8% more brands from Gemini. Chi-square analysis confirms systematic gender-category associations across all three studies (Study 1: \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((\\chi^2 = 137.32)\\)\u003c/span\u003e\u003c/span\u003e; Study 2: \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((\\chi^2 = 524.32)\\)\u003c/span\u003e\u003c/span\u003e; both \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((p \u0026lt; 0.001)\\)\u003c/span\u003e\u003c/span\u003e; Cram\\'{e}r's \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((V = 0.23)\\)\u003c/span\u003e\u003c/span\u003e--\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((0.38)\\)\u003c/span\u003e\u003c/span\u003e). We identify ``category gatekeeping'' as the underlying bias mechanism, document 69 gender-locked brands across all studies, and introduce the Prompt-Adjusted Share of Recommendation (PASOR) metric with empirical validation. Cross-model agreement remains consistently low (Jaccard \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((0.00)\\)\u003c/span\u003e\u003c/span\u003e--\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((0.68)\\)\u003c/span\u003e\u003c/span\u003e). These findings establish that gender bias in LLM recommendations is systematic, platform-dependent, and modulated by seasonal context, with implications for AI fairness, open-world machine learning, and data-driven consumer decision-making.\u003c/p\u003e","manuscriptTitle":"Gender Bias in Large Language Model Brand Recommendations: A Three-Study Analysis of Prompt-Induced Disparities Across Seasonal and Recipient Contexts","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-17 07:28:17","doi":"10.21203/rs.3.rs-8883056/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"cd629ee5-1cd4-403a-8ad0-45ba72e849a3","owner":[],"postedDate":"February 17th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-17T03:54:35+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-17 07:28:17","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8883056","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8883056","identity":"rs-8883056","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}