Human Psychometric Questionnaires Mischaracterize LLM Behavior

Hugging Face Daily Papers Papers

Summary

This paper finds that human psychometric questionnaires fail to reliably predict LLM behavior in real-world interactions, and proposes generation-based profiling as a more accurate alternative.

We examine whether human psychometric questionnaires can serve as reliable tools for characterizing and predicting LLM behavior in everyday user interactions. We analyze eight open-source LLMs by comparing their value and personality profiles derived from two different methods: Likert self-reports on established questionnaires (PVQ-40/21 and BFI-44/10) and generation probabilities over value-laden responses to everyday user queries. The two profiles diverge substantially. Within-construct item consistency, often cited as evidence of stable LLM dispositions, disappears in generation probabilities. We attribute this gap to the fact that explicit lexical cues in established questionnaire items allow models to recognize the target construct and respond in alignment-consistent, socially desirable ways, whereas realistic user queries provide no such cues. In addition, demographic persona prompts shift models' responses to human questionnaires in ways consistent with real human patterns, but no such shifts appear in the generation probabilities of responses to realistic user queries, showing their limited ability to simulate the behaviors of target demographics in real-world user interactions. Overall, our study shows that human psychometric questionnaires are insufficient tools for predicting LLM behavior and suggests generation-based profiling as a more accurate measure.
Original Article
View Cached Full Text

Cached at: 06/09/26, 08:43 AM

Paper page - Human Psychometric Questionnaires Mischaracterize LLM Behavior

Source: https://huggingface.co/papers/2509.10078

Abstract

Human psychometric questionnaires fail to reliably predict LLM behavior in real-world interactions, while generation-based profiling offers superior accuracy for understanding model responses to everyday user queries.

We examine whether humanpsychometric questionnairescan serve as reliable tools for characterizing and predicting LLM behavior in everyday user interactions. We analyze eight open-sourceLLMsby comparing their value andpersonality profilesderived from two different methods:Likert self-reportson established questionnaires (PVQ-40/21andBFI-44/10) andgeneration probabilitiesovervalue-laden responsesto everyday user queries. The two profiles diverge substantially. Within-construct item consistency, often cited as evidence of stable LLM dispositions, disappears ingeneration probabilities. We attribute this gap to the fact that explicit lexical cues in established questionnaire items allow models to recognize the target construct and respond in alignment-consistent, socially desirable ways, whereas realistic user queries provide no such cues. In addition,demographic persona promptsshift models’ responses to human questionnaires in ways consistent with real human patterns, but no such shifts appear in thegeneration probabilitiesof responses to realistic user queries, showing their limited ability to simulate the behaviors of target demographics in real-world user interactions. Overall, our study shows that humanpsychometric questionnairesare insufficient tools for predicting LLM behavior and suggests generation-based profiling as a more accurate measure.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2509\.10078

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2509.10078 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2509.10078 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2509.10078 in a Space README.md to link it from this page.

Collections including this paper1

Similar Articles

Evaluating LLMs as Human Surrogates in Controlled Experiments

arXiv cs.CL

This paper evaluates whether off-the-shelf LLMs can reliably simulate human responses in controlled behavioral experiments by comparing LLM-generated data with human survey responses on accuracy perception. The findings show that while LLMs capture directional effects and aggregate belief-updating patterns, they do not consistently match human-scale effect magnitudes, clarifying when synthetic LLM data can serve as behavioral proxies.

Evaluation Drift in LLM Personality Induction: Are We Moving the Goalpost?

arXiv cs.CL

This paper investigates whether fine-tuning LLMs on long-form essays with associated Big Five personality profiles stabilizes questionnaire responses and can induce target profiles, finding that while variance reduces, accuracy on the full five-dimensional profile remains near chance.