[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84432-en":3,"doc-seo-84432-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},84432,1099513958607,"Jiven","https://ap-avatar.wpscdn.com/avatar/100002390cf8733938c?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778829742770036399",8,"Research & Report","LLM-Based Social Simulations Require a Boundary","This position paper argues that LLM-based social simulations need clear methodological boundaries to produce meaningful contributions to social science. Although large language models can simulate human behavior, their tendency toward homogeneous “average persona” outputs restricts behavioral diversity, which is essential for capturing complex social dynamics. The paper links mean alignment to behavioral variance and reviews representative studies, finding validation practices often under-measure variance. It recommends matching validation depth to heterogeneity needs, reporting variance, and limiting claims to collective qualitative patterns when variance is insufficient.","LLM-Based Social Simulations Require a Boundary  \nZengqing Wu 1 Run Peng 2 Takayuki Ito 3 Makoto Onizuka 1 Chuan Xiao 1 4  \narXiv :2506 . 19806v3 [ cs .CY] 12 Jul 2026  \nAbstract  \nThis position paper argues that LLM-based social simulations require clear boundaries to make meaningful contributions to social science. While Large Language Models (LLMs) offer promising capabilities for simulating human behavior, their tendency to produce homogeneous outputs, acting as an “average persona”, fundamentally limits their ability to capture the behavioral diversity essential for complex social dynamics. We examine why heterogeneity matters for social simulations and how current LLMs fall short, analyzing the relationship between mean alignment and variance in LLM-generated behaviors. Through a systematic review of representative studies, we find that validation practices often fail to match the heterogeneity requirements of research questions: while most papers include ground truth comparisons, fewer than half explicitly assess behavioral variance, and most that do report lower variance than human populations. We propose that researchers should: (1) match validation depth to the heterogeneity demands of their research questions,(2) explicitly report variance alongside mean alignment, and (3) constrain claims to collective-level qualitative patterns when variance is insufficient. Rather than dismissing LLM-based  \nsimulations, we advocate for a boundary-aware approach that ensures these methods contribute genuine insights to social science.  \n1. Introduction  \nSocial simulation is a modeling tool that employs computational methods to understand social phenomena. Computational methods, particularly those modeling interactions between individuals, demonstrate advantagesin capturing the complex and nonlinear behaviors typically inherent in social phenomena (Eidelson, 1997 ; Remondino et al., 2010 ; San Miguel et al., 2012) . Among these,  \n1University of Osaka 2University of Michigan 3 Kyoto  \nUniversity 4Nagoya University. Correspondence to:  \nChuan Xiao \u003C[chuanx@ist.osaka-u.ac.jp](chuanx@ist.osaka-u.ac.jp) >, Zengqing Wu \u003C[zengqing.wu@ist.osaka-u.ac.jp](zengqing.wu@ist.osaka-u.ac.jp) >.  \nPreprint. July 14, 2026.  \nAgent-Based Modeling (ABM) is a widely used technique in this area, simulating how individual behaviors and local rules give rise to macro-level patterns (Bonabeau, 2002 ; Epstein, 1999 ; Schelling, 1971). ABM offers a bottom-up modeling approach, supports heterogeneity among agents, allows for the exploration of emergent phenomena, and provides researchers with interpretable mechanisms linking microand macro-level behaviors (Jackson et al., 2017 ; Page, 2012 ; Reeves et al., 2022) . Meanwhile, it is controversial due to its reliance on simplification (Edmonds & Moss, 2004), limited adaptability (Wu et al., 2023), sensitivity to initial conditions (Manzo & Matthews, 2014), and challenges in representing subjective or human-like behaviors (Maet al., 2024 ; Puig et al., 2021), diminishing the contribution of social simulation methods to social science (Reeves et al., 2022) .  \nRecently, LLM agents and social simulations have attracted growing attention. Existing studies have applied LLM agents to domains such as economics (Han et al., 2023 ; Li et al., 2024), education (Zhang et al., 2024c), game theory (Sreedhar & Chilton, 2024), and social networks (Wang et al., 2023 ; Yang et al., 2024c ; Zhang et al., 2025a), with claimed advantages like handling natural language, enabling flexible behaviors, and showing human-like reasoning. However, concerns have also been raised: LLMs may carry social and cognitive biases (Mohammadi, 2024 ; Navigli et al., 2023), lack behavioral diversity (Ma et al., 2025), and are hard to validate or explain (Larooij & Trnberg, 2025 ; Ma et al., 2024) . Whether or not using LLMs is a good protocol for social simulations remains an open question—or may not even be the central question to ask. Many existi","cbCaiaF8GyRkv6wA","https://ap.wps.com/l/cbCaiaF8GyRkv6wA","pdf",392039,2,1,24,"English","en",105,"# Introduction\n## Agent-Based Modeling and Prior Critiques\n## Emerging Role of LLM Agents in Social Simulation\n## Defining “Boundary” and Paper Claims","[{\"question\":\"Why do LLM-based social simulations require a “boundary”?\",\"answer\":\"Because current LLMs often generate behaviorally homogeneous outputs that limit behavioral diversity, so only certain research claims can be reliably supported given model limitations.\"},{\"question\":\"How does the paper connect mean alignment with behavioral variance?\",\"answer\":\"It analyzes how validation should consider not only whether simulated behavior matches the mean of real data, but also whether the variance and diversity of behaviors align with the heterogeneity required by the research question.\"},{\"question\":\"What validation shortcomings does the paper identify in existing studies?\",\"answer\":\"Most studies include ground-truth comparisons for mean alignment, but fewer than half explicitly assess behavioral variance, and reported variances frequently fall below those observed in human populations.\"}]",1784195606,60,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"llm-based-social-simulations-require-a-boundary","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":20},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/llm-based-social-simulations-require-a-boundary/84432/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-22","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why do LLM-based social simulations require a “boundary”?","Question",{"text":75,"@type":76},"Because current LLMs often generate behaviorally homogeneous outputs that limit behavioral diversity, so only certain research claims can be reliably supported given model limitations.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the paper connect mean alignment with behavioral variance?",{"text":80,"@type":76},"It analyzes how validation should consider not only whether simulated behavior matches the mean of real data, but also whether the variance and diversity of behaviors align with the heterogeneity required by the research question.",{"name":82,"@type":73,"acceptedAnswer":83},"What validation shortcomings does the paper identify in existing studies?",{"text":84,"@type":76},"Most studies include ground-truth comparisons for mean alignment, but fewer than half explicitly assess behavioral variance, and reported variances frequently fall below those observed in human populations.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,109,114,119,122,127,130,134],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":29,"slug":108},5,"Comic","comic",{"id":110,"doc_module":4,"doc_module_name":46,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]