Partial Identification

Follow this topic Ask about this topic

Partial identification asks which values of a causal effect, policy value, or other target remain compatible with observed data when point identification fails. Researchers derive sharp bounds or identified sets using linear programming, monotonicity and sensitivity assumptions, instrumental variables, missing-data models, and causal graphs, then build estimators and confidence intervals for those sets. Applications include treatment effects under hidden confounding, selective labels and missing outcomes, counterfactual harms, off-policy evaluation, and limits of machine-learning capability or preference measurements. Current work seeks tighter bounds and useful decisions or experiments without relying on implausible assumptions, while confronting a central limit: more observations can reduce sampling error but cannot resolve ambiguity built into the data or assumptions.

Latest papers

31 papers7 papers in the last four weeks, up 17% on the four weeks before. 0.0% of all new papers.

Certifying Distributionally Robust Decisions from Projection Observations
Oct 25, 2026Pouya Ahadi, Kamran PaynabarDecision-Making under Uncertainty Distributionally Robust Optimization
H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA.
Unified Sharp Bounds on Probabilities of Causation
Oct 19, 2026Xin Shu, Ang LiCausal Inference Partial Identification
Department of Computer Science, Florida State University
Estimating Causal Bounds from Observational Data With Neural Generative Models
Oct 18, 2026Max Zhu, Srivathsan Amruth, Sina Akbari +1Causal Inference Causal Effect Estimation
Department of Pure Mathematics and Mathematical Statistics University of Cambridge
Waiting Buys Exactly Two Moments: Endogenous Identification in Prediction-Driven Allocation
Oct 16, 2026Rama AlHamidi, Rasul Khanbayov, Erchin Serpedin +1Sequential Decision Making Predict-Then-Optimize
Department of Computer and Electrical Engineering, Texas A&M University at Qatar, Doha, Qatar · College of Science and Engineering, Hamad Bin Khalifa University, Doha, Qatar · Department of Computer and Electrical Engineering, Texas A&M University, College Station, TX, USA
What Does Disagreement Mean? A Semantic Identifiability Frontier for Pairwise Preference Learning
Oct 14, 2026Manoj Saravanan, Rohit Kumar Salla, Shrikar Reddy KotaPairwise Preference Learning Annotator Disagreement
Virginia Tech, Blacksburg, VA, USA.
What a Data Release Must Preserve for Off-Policy Evaluation
Oct 11, 2026Kyungbok Lee, Yating Zou, Michael R. KosorokOff-Policy Evaluation Partial Identification
Department of Biostatistics, Gillings School of Global Public Health University of North Carolina at Chapel Hill
Same Predictions, Different Harms: Causal Auditing of Patient World Models
Oct 4, 2026Yicheng Qi, Xiyi XiongStructural Causal Models Causal World Models
Imperial College London, London, United Kingdom
Path-specific harm decomposition: A partial identification framework
Sep 24, 2026Ruizi Yan, Dennis Frauen, Maresa Schröder +1Causal Effect Estimation Causal Mediation Analysis
Department of Statistics, University of Oxford · LMU Munich · Munich Center for Machine Learning (MCML)
The Delegation Blind Spot: Auditing Product Decisions from Agent Choices
Sep 22, 2026Shivam GuptaAI Agent Auditing Partial Identification
Independent Research
Information Set Emulation: Causal Certificates for AI Derived EHR Features
Sep 15, 2026Takes Fujita, Nobutaka HattoriCausal Effect Estimation Partial Identification
VRI · Department of Neurology, Juntendo University School of Medicine
The Anatomy and Boundary of Adaptation under Temporal Tabular Shift
Sep 10, 2026Tianyu Wang, Xi Vincent Wang, Lihui Wang +2Tabular Foundation Models Distribution Shift
Department of Production Engineering, KTH Royal Institute of Technology, Stockholm, Sweden · Global Institute of Future Technology, Shanghai Jiao Tong University, Shanghai, China
What Fixed-Rollout pass@k Evaluations Can Identify
Sep 8, 2026Pranav Singh, Prashant SinghConfidence Region Estimation Partial Identification
Department of Mathematics Indian Institute of Technology Ropar Ropar, Punjab 140001, India
Resolution-Aware Experimental Design under Partial Identifiability
Sep 3, 2026Sofianos Panagiotis FotiasPartial Identification Bayesian Experimental Design
National Technical University of Athens
Partial Identification under Causal Orders by Linear Programming
Aug 25, 2026Eric Rossetto, Alessandro AntonucciPartial Identification Causal Inference
Istituto Dalle Molle di Studi sull’Intelligenza Artificiale (IDSIA) Scuola Universitaria Professionale della Svizzera Italiana (SUPSI) Lugano, Switzerland
Agent Behavioral Contracts II: Certifying Compositional Reliability Without Assuming Independence
Aug 13, 2026Varun Pratap Bhardwaj, Garima Singh, Arun Pratap BhardwajAgent Failure Analysis AI Agent Reliability
Qualixar / Independent Researcher, India · Independent Researcher, India
General Probabilities of Causation with Causal Knowledge
Aug 12, 2026Xin Shu, Zhen Lei, Ang LiPartial Identification Causal Identifiability
Department of Computer Science, Florida State University
Bounding the Causal Impact of ML-assisted Decision-Making via Counterfactual Correctness
Jul 23, 2026Jonathan Zhang, Erik Skalnes, Jacob Chen +1AI-Assisted Decision Making Causal Effect Estimation
Computer Science Dept., Johns Hopkins University, Baltimore, Maryland, USA
Partial Identification with Multiple Nonlinear Measurements of a Latent Regressor
Jul 13, 2026Burhan Ogut, Michelle YinPartial Identification Linear Regression
American Institutes for Research. · Northwestern University, School of Education and Social Policy.
NL-PAC: Specification Ambiguity and Certified Minimax Risk Floors in LLM-Mediated Supervision
Jul 9, 2026Berkay AnahtarciLLM Evaluation Partial Identification
Department of Mathematical Engineering, Özyeğin University, Istanbul, Türkiye
Root Cause Analysis with Latent Confounders using Partial Ancestral Graphs
Jun 18, 2026Henrique O. Caetano, Rafael Arone, Carlos Dias MacielCausal Inference with Unobserved Confounding Partial Identification
Signal Processing Laboratory, São Carlos School of Engineering, University of São Paulo, São Paulo, Brazil · Guaratinguetá School of Engineering, São Paulo State University, Guaratinguetá, Brazil
Identification and Inference for Algorithmic Frontiers with Selective Labels
Jun 12, 2026Yiqi Liu, Francesca Molinari, Amilcar VelezDebiased ML Parameter Identifiability
Department of Economics, Cornell University
Privacy-Robust Incrementality Measurement for Advertising Systems under Signal Loss
Jun 2, 2026Prashant Shekhar, Caroline HowardOnline Advertising Partial Identification
Department of Mathematics Embry-Riddle Aeronautical University Daytona Beach, FL, USA
IV-ICL: Bounding Causal Effects with Instrumental Variables via In-Context Learning
May 13, 2026Vahid Balazadeh, Hamidreza Kamkari, Medha Barath +2Causal Effect Estimation Parameter Identifiability
University of Toronto · Vector Institute · MIT CSAIL +1
Decision Support for Marketplace Policies under Incomplete Evidence: From Replay to Launch Readiness
May 13, 2026Prashant Shekhar, Caroline HowardAI-Assisted Decision Making Off-Policy Evaluation
Department of Mathematics Embry-Riddle Aeronautical University, Daytona Beach, FL, USA
Identified-Set Geometry of Distributional Model Extraction under Top- $K$ Censored API Access
May 11, 2026Wenhua Nie, ZiCheng Zhu, Jianan Wu +3Parameter Identifiability Partial Identification
National Taiwan University
Optimal Experiments for Partial Causal Effect Identification
May 7, 2026Tobias Maringgele, Jalal EtesamiCausal Effect Estimation Partial Identification
Technical University of Munich
How Wrong Can Your Counterfactual Be? Quantifying Confounding Bias for Continuous Treatments without a Control Group
Mar 8, 2026Yu Wang, Xiangchen Liu, Siguang LiCausal Effect Estimation Causal Inference
Department of Economics, Cornell University. · Department of Family and Consumer Sciences, California State University, Long Beach. · Society Hub, Hong Kong University of Science and Technology (Guangzhou).
Causal Identification from Counterfactual Data: Completeness and Bounding Results
Feb 26, 2026Arvind Raghavan, Elias BareinboimCausal Inference Causal Identifiability
Causal Artificial Intelligence Lab, Department of Computer Science, Columbia University
Bounds and Identification of Joint Probabilities of Potential Outcomes and Observed Variables under Monotonicity Assumptions
Feb 21, 2026Naoya Hashimoto, Yuta Kawakami, Jin TianMonotonicity Constraints Causal Identifiability
Mohamed bin Zayed University of Artificial Intelligence
AI-Generated Measurements for Identification and Inference with Missing Data: A Weak Shadow Variable Approach
Feb 17, 2026Hongyu Chen, David Simchi-Levi, Ruoxuan XiongLearning with Missing Data Partial Identification
Massachusetts Institute of Technology, Cambridge, MA 02139 · Emory University, Atlanta, GA 30322

Oct 25, 2026J/K move · Enter open · S save

Certifying Distributionally Robust Decisions from Projection Observations

Pouya Ahadi, Kamran Paynabar

H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA.

Decision-Making under Uncertainty Distributionally Robust Optimization Partial Identification

Open paper AskPDF arXiv

A scalar measurement can determine a robust decision without identifying the latent covariance or every decision risk. We study finite Gaussian projection experiments with a fixed mean-plus-standard-deviation target. The observation laws define a set of indistinguishable moment models. We exhibit decisions that agree throughout that set despite unidentified risk values, and give a confidence-set certificate that returns a decision or an unresolved outcome. We prove soundness, necessary abstention for indistinguishable models with incompatible optima, and a fixed-checkpoint certification guarantee. A scalar profile reduces nonlinear risk comparisons to convex inner problems. Across seven synthetic instances, jointly calibrated moment regions give sufficient budgets 21 to 52 times explicit expected-sample information lower bounds. Separately measured stopping times are lower, with zero decision errors in 1,900 simulated streams. These comparisons do not constitute matching bounds, and the guarantees require Gaussian replication.