Ethos | Bio | Papers | Visuals | Students

At UT Austin, we train students to become full-stack researchers—and increasingly, designers of the systems that do research. Our students learn to carry projects end-to-end: from idea generation and theory to data creation, analysis, and iterative refinement across diverse subfields. Using modern AI (including agentic workflows) and scalable computation, students build reproducible pipelines that can ingest and update planetary-scale data—like satellite imagery and other high-dimensional sources. But the goal isn’t tool use for its own sake: students learn to set the objectives, constraints, and evaluation standards that guide these systems through large spaces of hypotheses, while grounding results in causal inference and careful measurement. The outcome is scholarship that can rigorously test policy counterfactuals and translate evidence into durable, responsible improvements in societal well-being.

We welcome students at every stage to engage with projects—from motivated high-schoolers to undergraduates, graduate students, and those from highly non-traditional backgrounds. See also [Jobs] for information about open positions.

Present:
[1.] Assistant Professor in the Department of Government at the University of Texas at Austin.
[2.] Consultant, Institute for Health Metrics & Evaluation (IHME), University of Washington.

Past:
[1.] Visiting Assistant Professor in the Department of Government at Harvard University (2024).
[2.] Postdoc, AI & Global Development Lab (2021-2022).

Methodological work: AI and global development, EO for causal inference, adversarial dynamics, computational text analysis.

Substantive work: Political economy, social movements, descriptive representation.

[CV] [Homepage] [.bib] [logs] [arXiv]

[Team] [Students] [Google Scholar] [UT Profile]

[PlanetaryCausalInference.org] [AI & Global Dev Lab GitHub]

[YouTube Tutorials] [Data Assets]

• Nicolas Audinet de Pieuchon presents: Benchmarking Debiasing Methods for LLM-based Parameter Estimates
• Nicolas Audinet de Pieuchon presents: Can Large Language Models (or Humans) Disentangle Text?
• Adel Daoud presents: A First Course in Planetary Causal Inference: Confounding (@IC2S2 2025)
• Adel Daoud presents: Planetary Causal Inference: Overview (@Yale)
• Connor Jerzak presents: Seeing Like a Satellite While Learning Across Scales: Remote Audits + Multi-Scale Optimization for Heterogeneity (@Columbia)
• Connor Jerzak presents: Selecting Optimal Candidate Profiles in Adversarial Environments (@UT Dallas & National Chung Hsing University)
• Richard Johansson presents: Conceptualizing Treatment Leakage in Text-based Causal Inference (@NAACL)
• Satiyabooshan Murugaboopathy presents: Platonic Representations for Poverty Mapping: Unified Vision-Language Codes or Agent-Induced Novelty?
• Kazuki Sakamoto presents: A Scoping Review of Earth Observation and Machine Learning for Causal Inference
• Fucheng Warren Zhu presents: Optimizing Multi-Scale Representations to Detect Effect Heterogeneity Using EO and Computer Vision

• Benchmarking Debiasing Methods for LLM-based Parameter Estimates (EMNLP 2025) – Nicolas Audinet de Pieuchon, Adel Daoud, Connor T. Jerzak, Moa Johansson, Richard Johansson. This paper benchmarks different techniques for removing bias from LLM-generated labels, showing that combining large-scale LLM annotations with a modest number of expert labels can reduce bias and improve parameter estimation. [PDF] [.bib] [Video] [Slides] [Data]

• Can Large Language Models (or Humans) Disentangle Text? (NLP+CSS 2024) – Nicolas Audinet de Pieuchon, Adel Daoud, Connor T. Jerzak, Moa Johansson, Richard Johansson. Investigates whether LLMs or humans can separate intertwined textual features and highlights the limitations of current models. [PDF] [.bib] [Code]

• Conceptualizing Treatment Leakage in Text-Based Causal Inference (NAACL 2022) – Adel Daoud, Connor T. Jerzak, Richard Johansson. Introduces the notion of treatment leakage when causal treatments are encoded in text, offering guidance on designing text-based experiments to minimize leakage. [PDF] [.bib] [Video]

• Linking Datasets on Organizations Using Half a Billion Open-Collaborated Records (PSMR 2024) – Brian Libgober, Connor T. Jerzak. Describes methods for linking disparate organizational data sets using over 500 million open-collaborated records. [PDF] [.bib] [Code]

• An Improved Method of Automated Non-Parametric Content Analysis for Social Science (Political Analysis 2023) – Connor T. Jerzak, Gary King, Anton Strezhnev. Presents an enhanced non-parametric content-analysis approach that automates the extraction of substantive information from text. [PDF] [.bib] [Code]

• Planetary Causal Inference: Understanding the Environment, Society, and Economy through Earth Observation and AI Systems (Cambridge University Press 2026+) – Connor T. Jerzak, Adel Daoud. Book project on Planetary Causal Inference, capturing efforts in an emerging field to combine satellite imagery and planetary-scale data sources with localized studies, especially RCTs, to derive insights about causality. [More]

• Chinese vs. World Bank Development Projects: Insights from Earth Observation and Computer Vision on Wealth Gains in Africa, 2002-2013 (World Development 2026) – Adel Daoud, Cindy Conlin, Connor T. Jerzak. Satellite-imagery–augmented causal estimates compare sector-specific wealth gains from Chinese vs World Bank projects across 9,899 African neighborhoods, finding positive effects for both and generally larger gains for China. [PDF] [.bib] [Data]

• Debiasing Machine Learning Predictions for Causal Inference Without Additional Ground Truth Data: “One Map, Many Trials” in Satellite-Driven Poverty Analysis (AAAI 2026) – Markus Pettersson, Connor T. Jerzak, Adel Daoud. Proposes methods to debias ML predictions for causal inference in poverty mapping. [PDF] [.bib] [Code]

• Optimizing Multi-Scale Representations to Detect Effect Heterogeneity Using Earth Observation and Computer Vision: Applications to Two Anti-Poverty RCTs (CLeaR 2025) – Warren Zhu Fucheng, Connor T. Jerzak, Adel Daoud. Develops multi-scale image representations to discover treatment effect heterogeneity in randomized controlled trials. [PDF] [Video] [.bib]

• A Scoping Review of Earth Observation and Machine Learning for Causal Inference: Implications for the Geography of Poverty (2025) – Kazuki Sakamoto, Connor T. Jerzak, Adel Daoud. Reviews the literature on combining earth-observation data with machine learning for causal inference. [PDF] [.bib] [Data]

• Effect Heterogeneity with Earth Observation in Randomized Controlled Trials: Exploring the Role of Data, Model, and Evaluation Metric Choice – Connor T. Jerzak, Ritwik Vashistha, Adel Daoud. Examines how historical data, model selection and evaluation metrics affect detection of effect heterogeneity when using earth-observation images. [PDF] [.bib]

• Image-based Treatment Effect Heterogeneity (CLeaR 2023) – Connor T. Jerzak, Fredrik Johansson, Adel Daoud. Introduces methods for estimating heterogenous treatment effects directly from images. [PDF] [.bib] [Code]

• Integrating Earth Observation Data into Causal Inference: Challenges and Opportunities – Connor T. Jerzak, Fredrik Johansson, Adel Daoud. Outlines challenges and strategies for incorporating satellite imagery into causal inference and shows that high-resolution satellite data can help adjust for confounders. [PDF] [.bib] [Code]

• Platonic Representations for Poverty Mapping: Unified Vision-Language Codes or Agent-Induced Novelty? – Satiyabooshan Murugaboopathy, Connor T. Jerzak, Adel Daoud. Investigates whether unified vision–language models can represent poverty or if generative agents create novel representations. [PDF] [.bib] [Data]

• FastRerandomize: Fast Rerandomization Using Accelerated Computing (SoftwareX 2026) – Connor T. Jerzak, Rebecca Goldstein, Aniket Kamat, Fucheng Warren Zhu. Presents an efficient algorithm for rerandomization of experiments, leveraging accelerated computing. [PDF] [.bib] [Code]

• Selecting Optimal Candidate Profiles in Adversarial Environments Using Conjoint Analysis and Machine Learning – Connor T. Jerzak, Priyanshi Chandra, Rishi Hazra. Develops a framework to identify candidate profiles that remain robust when voters evaluate profiles strategically. [PDF] [.bib]

• Attenuation Bias with Latent Predictors – Connor T. Jerzak, Stephen Jessee. Explores how measurement error in latent predictors can attenuate causal estimates and proposes corrections. [PDF] [.bib]

• CausalImages: An R Package for Causal Inference with Earth Observation, Biomedical and Social Science Images – Connor T. Jerzak, Adel Daoud. Introduces an R package for performing causal inference directly on image and image-sequence data. [PDF] [.bib] [Code]

• Electoral Rules and Descriptive Representation: A Comprehensive View Across Multiple Identities (Under contract, Cambridge University Press) – John Gerring, Alan Hicken, Connor T. Jerzak, Robert Moser, Erzen Öncel. Studies how electoral rules affect representation. [PDF] [.bib]

• The Composition of Descriptive Representation (APSR 2024) – John Gerring, Connor T. Jerzak, Erzen Öncel. Analyzes how descriptive representation is composed and which demographic attributes drive voters’ preferences. [PDF] [.bib] [Code]

• The Impact of a Transportation Intervention on Electoral Politics: Evidence from E-ZPass (Research in Transportation Economics 2020) – Connor T. Jerzak, Brian Libgober. Assesses how the introduction of the E-ZPass toll system influenced housing values and partisan voting patterns. [PDF] [.bib]

• Football fandom in Egypt (Routledge Handbook of Sport in the Middle East 2022) – Connor T. Jerzak. Examines the intersection of football fandom and social identity in Egypt. [PDF] [.bib]

Planetary Causal Inference Workflow	Institutional Analysis
Fast Rerandomization with Accelerated Computing	Effect Heterogeneity with Image Sequences
PSRM 2024	ACL Anthology
PCI Book Launch

"Nullius in verba" — motto of the Royal Society (adopted 1662); meaning: "take nobody’s word for it" or "take nothing on authority": test claims by experiment, evidence, and first-principles reasoning.