·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
A Diversity Diet for a Healthier Model: A Case Study of French ModernBERT3h◆A Cross-lingual Comparison of Human and Classification Model Entrainment Behavior in Code-switched Speech Settings3h◆Romanized Arabic Across Dialects: Views, Usage Patterns, and Linguistic Variation3h◆When Humans Leave Context Unsaid: Implicit Context as a Risk for Large Language Models3h◆Beyond Static and Linear: What Attention Constraints Best Fit Human Reading Times?3h◆Tradeoff-Aware Territory Design: A Multi-Objective Decision System for Enterprise Customer-to-Seller Allocation3h◆Phronix: The Capability-Centric Learning Paradigm for Agents3h◆Reproducible Evaluation as the Substrate for Safe Prompt Compression in Production Agents3h◆The Helpful Agent Problem: When Emergent Cooperation Among LLM Agents Bypasses Intended Workflows3h◆CEMIR vs. FISER: a comparative study of reasoning-based and enrichment-based approaches to pragmatic intent recognition3h◆DELA: Learning Controllable Edits for Diffusion Language Models3h◆Learning Nonlinear Waveform Fusion Across Scales for Integrated Sensing and Communication3h◆PersonaScienceBench: Evaluating Scientific Agents with Diverse Simulated Researchers3h◆Financial Context-Rich Strategy for Actor-Critic based Agents: A Deep Reinforcement Learning Approach for Stock Trading3h◆RSIHub: A Composable Framework for Agent-Evolution Methods3h◆Closing the Modality Gap via Simplex-Constrained Representations3h◆SemantEq: Faithful Bidirectional Equivalence of Formal Statements3h◆RubricCraft: Where Solutions Diverge, Criteria Emerge3h◆SWE-Clar: Can Coding Agents Clarify Incomplete Client Requirements and Deliver Greenfield Software?3h◆Machine-Readable Operator Manuals over the ISO 11783 Network for Offline, Multilingual LLM Assistance3h◆A Diversity Diet for a Healthier Model: A Case Study of French ModernBERT3h◆A Cross-lingual Comparison of Human and Classification Model Entrainment Behavior in Code-switched Speech Settings3h◆Romanized Arabic Across Dialects: Views, Usage Patterns, and Linguistic Variation3h◆When Humans Leave Context Unsaid: Implicit Context as a Risk for Large Language Models3h◆Beyond Static and Linear: What Attention Constraints Best Fit Human Reading Times?3h◆Tradeoff-Aware Territory Design: A Multi-Objective Decision System for Enterprise Customer-to-Seller Allocation3h◆Phronix: The Capability-Centric Learning Paradigm for Agents3h◆Reproducible Evaluation as the Substrate for Safe Prompt Compression in Production Agents3h◆The Helpful Agent Problem: When Emergent Cooperation Among LLM Agents Bypasses Intended Workflows3h◆CEMIR vs. FISER: a comparative study of reasoning-based and enrichment-based approaches to pragmatic intent recognition3h◆DELA: Learning Controllable Edits for Diffusion Language Models3h◆Learning Nonlinear Waveform Fusion Across Scales for Integrated Sensing and Communication3h◆PersonaScienceBench: Evaluating Scientific Agents with Diverse Simulated Researchers3h◆Financial Context-Rich Strategy for Actor-Critic based Agents: A Deep Reinforcement Learning Approach for Stock Trading3h◆RSIHub: A Composable Framework for Agent-Evolution Methods3h◆Closing the Modality Gap via Simplex-Constrained Representations3h◆SemantEq: Faithful Bidirectional Equivalence of Formal Statements3h◆RubricCraft: Where Solutions Diverge, Criteria Emerge3h◆SWE-Clar: Can Coding Agents Clarify Incomplete Client Requirements and Deliver Greenfield Software?3h◆Machine-Readable Operator Manuals over the ISO 11783 Network for Offline, Multilingual LLM Assistance3h◆
News/Two Wrongs, No Right: Auditing Social-Desirability Bias in LLM Annotators for Computational Social Science
arxiv
PublishedJune 12, 2026 at 4:00 AM
▼bearish

Two Wrongs, No Right: Auditing Social-Desirability Bias in LLM Annotators for Computational Social Science

Source
arxiv.orgfull article ↗
Read on arxiv→
Publisher summary· verbatim

arXiv:2606.12426v1 Announce Type: cross Abstract: LLM annotators are increasingly used in computational social science (CSS), but it is unclear whether their alignment-shaped errors preserve the empirical conclusions a researcher would report. We audit three open-source 7B instruction-tuned models (

Stay posted· Newsletter

Three short emails a week: Monday, Wednesday and Friday. Top movers, new models and the most-referenced story.

// no spam · unsubscribe one-click · free forever

Your email is used only to send The Bubble Brief. Privacy policy

Discussion
Mentioned models
03
  • 01
    Zephyr
  • 02
    Mistral
  • 03
    Qwen
Source
↗
arxiv
Read original ↗All from arxiv →
Tags
04
#bias#computational social science#language models#validation

No replies yet. Be first.

Mentioned models
03
  • 01
    Zephyr
  • 02
    Mistral
  • 03
    Qwen
Source
↗
arxiv
Read original ↗All from arxiv →
Tags
04
#bias#computational social science#language models#validation

Related coverage

More from ARXIV
arxivA Diversity Diet for a Healthier Model: A Case Study of French ModernBERT3harxivA Cross-lingual Comparison of Human and Classification Model Entrainment Behavior in Code-switched Speech Settings3harxivRomanized Arabic Across Dialects: Views, Usage Patterns, and Linguistic Variation3harxivWhen Humans Leave Context Unsaid: Implicit Context as a Risk for Large Language Models3h
The Bubble Brief
MON · WED · FRI

Read bias insights every Monday, Wednesday and Friday — top movers, new models, story of the week.

// no spam · unsubscribe one-click · free forever

Your email is used only to send The Bubble Brief. Privacy policy

Originally published on arxiv ↗
PrivacyTermsRefundsLegal notice|AboutMethodologyContact

© 2026 Databubble

Built by Marouane Gazouzi · Readyo, AI interview coach

HomeModelsNews