Interventions in LLM-simulated user experiments induce distribution shifts in latent attributes that create confounding bias, diagnosable with negative control outcomes and partially mitigated by adding setting-relevant persona details.
Ringlein, Trang Q
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3verdicts
UNVERDICTED 3representative citing papers
Introduces extended bridge functions and derives identification results for joint interventional distributions retaining proxy variables in proximal causal inference.
Causal effects are identifiable from a single proxy of the unobserved confounder under the SPICE completeness assumption, supported by a neural estimation framework.
citing papers explorer
-
The Illusion of Intervention: Your LLM-Simulated Experiment is an Observational Study
Interventions in LLM-simulated user experiments induce distribution shifts in latent attributes that create confounding bias, diagnosable with negative control outcomes and partially mitigated by adding setting-relevant persona details.
-
Identifying Interventional Joint Distributions via Extended Bridge Functions
Introduces extended bridge functions and derives identification results for joint interventional distributions retaining proxy variables in proximal causal inference.
-
Identifying Causal Effects Using a Single Proxy Variable
Causal effects are identifiable from a single proxy of the unobserved confounder under the SPICE completeness assumption, supported by a neural estimation framework.