Direction-flipped influence audits show contextual cues shift LLM moral choices by 12-18 points on average across multiple benchmarks, revealing asymmetries, backfires, and inconsistencies in 40% of conditions.
Categorical Data Analysis
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
years
2026 4representative citing papers
A new fully conjugate Bayesian variable selection method for GLMs with posterior consistency guarantees for inclusion indicators and coefficients, implemented via Gibbs sampling.
Refines area-based and other DIF effect-size measures, identifies flaws in existing ones via simulation, and proposes unified cut-offs across Mantel-Haenszel, SIBTEST, and model-based methods with R implementation.
A marginalized transition model with Markov dependence and category-specific changepoint specification is developed for detecting shifts in serially correlated categorical time series, demonstrated on Canadian cloud cover observations.
citing papers explorer
-
Direction-Flipped Influence Audits Reveal Hidden Structure in Moral Choices of LLMs
Direction-flipped influence audits show contextual cues shift LLM moral choices by 12-18 points on average across multiple benchmarks, revealing asymmetries, backfires, and inconsistencies in 40% of conditions.
-
Bayesian Variable Selection in Generalized Linear Models
A new fully conjugate Bayesian variable selection method for GLMs with posterior consistency guarantees for inclusion indicators and coefficients, implemented via Gibbs sampling.
-
Refining Effect-Size Measures and Classification for Differential Item Functioning: Toward Unified Guidelines Across Methods
Refines area-based and other DIF effect-size measures, identifies flaws in existing ones via simulation, and proposes unified cut-offs across Mantel-Haenszel, SIBTEST, and model-based methods with R implementation.
-
Changepoint Detection in Categorical Time Series with Application to Daily Total Cloud Cover in Canada
A marginalized transition model with Markov dependence and category-specific changepoint specification is developed for detecting shifts in serially correlated categorical time series, demonstrated on Canadian cloud cover observations.