Saved in:
Bibliographic Details
Main Author: Cacioli, Jon-Paul
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.17707
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918455117086720
author Cacioli, Jon-Paul
author_facet Cacioli, Jon-Paul
contents Clinical personality assessment screens response validity before interpreting substantive scales. LLM evaluation does not. We apply the validity scaling framework from the PAI and MMPI-3 to metacognitive probe data from 20 frontier models across 524 items. Six validity indices are operationalised: L (maintaining confidence on errors), K (betting on errors), F (withdrawing consensus-endorsed items), Fp (withdrawing correct answers), RBS (inverted monitoring), and TRIN (fixed responding). A tiered classification system identifies four models as construct-level invalid and two as elevated. Valid-profile models produce item-sensitive confidence (mean r = .18, 14 of 16 significant). Invalid-profile models do not (mean r = -.20, d = 2.17, p = .001). Chain-of-thought training produces two opposite response distortions. Two latent dimensions account for 94.6% of index variance. Companion papers extract a portable screening protocol (Cacioli, 2026e) and validate it against selective prediction (Cacioli, 2026f). All data and code: https://github.com/synthiumjp/validity-scaling-llm
format Preprint
id arxiv_https___arxiv_org_abs_2604_17707
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Before You Interpret the Profile: Validity Scaling for LLM Metacognitive Self-Report
Cacioli, Jon-Paul
Computation and Language
Artificial Intelligence
Clinical personality assessment screens response validity before interpreting substantive scales. LLM evaluation does not. We apply the validity scaling framework from the PAI and MMPI-3 to metacognitive probe data from 20 frontier models across 524 items. Six validity indices are operationalised: L (maintaining confidence on errors), K (betting on errors), F (withdrawing consensus-endorsed items), Fp (withdrawing correct answers), RBS (inverted monitoring), and TRIN (fixed responding). A tiered classification system identifies four models as construct-level invalid and two as elevated. Valid-profile models produce item-sensitive confidence (mean r = .18, 14 of 16 significant). Invalid-profile models do not (mean r = -.20, d = 2.17, p = .001). Chain-of-thought training produces two opposite response distortions. Two latent dimensions account for 94.6% of index variance. Companion papers extract a portable screening protocol (Cacioli, 2026e) and validate it against selective prediction (Cacioli, 2026f). All data and code: https://github.com/synthiumjp/validity-scaling-llm
title Before You Interpret the Profile: Validity Scaling for LLM Metacognitive Self-Report
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2604.17707