Observer-rated behavioral expression of high- and moderate-intensity ADHD personas declines progressively across turns in unscripted LLM conversations¶

Claim · All claims
Evidence · 1 study · causal r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? In unscripted education conversations, observer-rated ADHD scores showed a highly significant, substantial decline per three-turn assessment interval, while self-reports stayed stable. → Jana Gonnermann-Müller 2026

Evidence¶

Jana Gonnermann-Müller 2026¶

Jana Gonnermann-Müller, Jennifer Haase, Nicolas Leins, Thomas Kosch, Sebastian Pokutta. (2026). LLM-Based Educational Simulation: Evaluating Temporal Student Persona Stability Across ADHD-Related Persona Specifications. https://arxiv.org/abs/2605.06307 doi:10.48550/arxiv.2605.06307

q2 · i? · causal · r2

LMM analysis of Experiment II unscripted education conversations with observer ratings at turns 3, 6, and 9. Descriptively, high-intensity personas declined by ∆ = −4.0 points (11.1%) and moderate by ∆ = −2.82 (7.8%); the article reports the turn effect as "highly significant and substantial (β = −1.13...)". No standardized effect size is printed.

"For unscripted education observer ratings, the turn effect was highly significant and substantial (β = −1.13, SE = 0.05, t = −20.80, p<. 001). Post-hoc pairwise comparisons confirmed significant differences between all turn pairs (allp < .001)."

Discussion¶

Related Claims¶

• Default personas show a self-report–behavior gap: they endorse moderate-range ADHD characteristics on self-report while displaying only weakly elevated ADHD-like behavior to observers — related
• Between-conversation persona stability is intensity-dependent: high-intensity ADHD personas cluster tightly while moderate-intensity personas show the highest aggregate variability — related
• LLM-simulated students' self-reported ADHD persona scores remain stable throughout extended multi-turn conversations — related
• Model family and prompt format moderate the level of simulated ADHD persona expression — related
• Scripted conversations with recurring task-relevant prompts nearly eliminate observer-rated persona drift, up to a 97% reduction for high-intensity education personas — related
• LLM observer ratings converge excellently with human expert judgments in applying the CAARS observer instrument to generated narratives — related
• RPLA failures are temporally distributed: ethical violations are rare in the first three turns and become significantly more frequent after turn 6 — related