ADAdel Daoud

Adel Daoud

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Adel Daoud has a total of 56 co-authors. Here are the top co-authors by number of shared publications:

  • Connor Thomas Jerzak

    University of Texas at Austin USA

    6 shared publications
  • Devdatt Dubhashi

    The Division of Data Science and Artificial Intelligence of the Department of Computer Science and Engineering, Chalmers University of Technology, Gothenburg, Sweden

    3 shared publications
  • Fredrik D. Johansson

    RISE Research Institutes Chalmers University of Technology University of Gothenburg Sweden KTH Royal Institute of Technology Chalmers University of Technology University of Gothenburg

    3 shared publications
  • Fredrik Johansson

    The Division of Data Science and Artificial Intelligence of the Department of Computer Science and Engineering, Chalmers University of Technology, Gothenburg, Sweden

    3 shared publications
  • Makkunda Sharma

    Wadhwani AI, Mumbai, India

    3 shared publications
  • Aki Yazawa

    Department of Primary Care and Medical Education Institute of Medicine, University of Tsukuba Ibaraki Japan

    2 shared publications

Latest Publications

  • Open Access

    Economic horizontal inequality is linked to conflict persistence, not onset: High-resolution evidence from satellite-derived wealth measures in Africa

    by Connor T. Jerzak, Adel Daoud, Johannes Fagerlind

    Conflict research links horizontal inequality—systematic inequalities between identity groups—to armed violence, but it remains unclear whether economic inequality predicts conflict outbreak or instead shapes conflict dynamics after violence becomes plausible. We revisit this question in Africa using high-resolution satellite-derived International Wealth Index estimates to measure economic horizontal inequality among politically relevant ethnic groups and local wealth shocks. In group-year models, economic horizontal inequality is not statistically associated with conflict onset, while political marginalization is a stronger predictor. By contrast, economic horizontal inequality is associated with active-rebel group-years and higher casualty intensity among active-conflict group-years. A spatiotemporal counterfactual analysis finds that dense local wealth-shock scenarios are not associated with more conflict events on average, but estimated conditional contrasts become positive in areas with large absolute economic horizontal inequality. Economic horizontal inequality may therefore matter less as a general trigger than as a condition shaping conflict persistence, severity, and shock vulnerability.

  • Open Access

    To What Extent Large-Language Models Understand Causal Reasoning in Sociological Texts: A Pilot Study

    by Cheng Lin, Adel Daoud

    Causal reasoning is central to social sciences, supporting both theory development and testing. Large language models (LLMs) are increasingly used to analyze texts across social sciences, yet their abilities to interpret causal statements and reconstruct causal structures remain uncertain in social science texts. This study examines to what extent LLMs can (a) identify causal sentences and (b) recover causal edges to reconstruct directed acyclic graphs (DAGs). This examination relies on two text-based simulations and one experiment using real sociological texts. In simulated texts, models reliably detected explicit causal statements but often misclassified non-causal ones, particularly when probabilistic language was used. For causal graph recovery, clarity of expression strongly influenced accuracy. Across 500 replications with real texts, performance ranged from moderate to near perfect, revealing both model heterogeneity and inconsistency. These results indicate that LLMs can assist in extracting causal knowledge from social science texts, yet accurate inference of complex causal mechanisms still requires careful theoretical reasoning. Future applications may benefit from targeted prompting and domain-specific fine-tuning, but their contribution to sociological theorizing should be carefully assessed against traditional methodological standards.

  • Open Access

    To what extent do disadvantaged neighbourhoods mediate social assistance dependency? Evidence from Sweden

    by Cheng Lin, Adel Daoud, Maria Brandén

    This article investigates social assistance dependency and its relation to neighbourhood disadvantage in Sweden. We combine Swedish register data, tracking and analysing a cohort from 1998–2017, with the help of causal mediation, our analysis identifies the impact of early-adulthood social assistance on mid-adulthood social assistance. More specifically, we examine the mediating roles of neighbourhood conditions and compare this effect to the well-known mediating effect of unstable work experiences. Our findings suggest a differential effect for individuals with a high versus low probability of receiving social assistance in early adulthood. For individuals with a baseline high probability of receiving early-adulthood social assistance, the total estimated effect of early-adulthood social assistance on mid-adulthood social assistance recipiency is over 15 per cent points. Neighbourhood disadvantage only has a minor mediating effect on average, however, for individuals with a high risk of early-adulthood social assistance, the effect is substantial, over 5 per cent points, even more than the mediating effect from unstable work. The findings suggest that for high-risk individuals, social assistance recipiency in young adulthood is linked to subsequent entrenchment in disadvantaged areas and unstable employment, reinforcing a cycle of poverty. Our findings contribute to understanding the complex interactions between policy, socioeconomic status, and environmental factors in perpetuating social assistance dependency.


  • Open Access

    A Scoping Review of Earth Observation and Machine Learning for Causal Inference: Implications for the Geography of Poverty

    by Kazuki Sakamoto, Connor T. Jerzak, Adel Daoud

    Earth observation (EO) data such as satellite imagery can have far-reaching impacts on our understanding of the geography of poverty, especially when coupled with machine learning (ML) and computer vision. Early research used computer vision to predict living conditions in areas with limited data, but recent studies increasingly focus on causal analysis. Despite this shift, the use of EO-ML methods for causal inference lacks thorough documentation, and best practices are still developing. Through a comprehensive scoping review, we catalog the current literature on EO-ML methods in causal analysis. We synthesize five principal approaches to incorporating EO data in causal workflows: (1) outcome imputation for downstream causal analysis, (2) EO image deconfounding, (3) EO-based treatment effect heterogeneity, (4) EO-based transportability analysis, and (5) image-informed causal discovery. Building on these findings, we provide a detailed protocol guiding researchers in integrating EO data into causal analysis—covering data requirements, computer vision model selection, and evaluation metrics. While our focus centers on health and living conditions outcomes, our protocol is adaptable to other sustainable development domains utilizing EO data.

  • Deep Learning With DAGs

    by Sourabh Balgi, Adel Daoud, Jose M. Peña, Geoffrey T. Wodtke, Jesse Zhou

    Social science theories often postulate systems of causal relationships among variables, which are commonly represented using directed acyclic graphs (DAGs). As non-parametric causal models, DAGs require no assumptions about the functional form of the hypothesized relationships. Nevertheless, to simplify empirical evaluation, researchers typically invoke such assumptions anyway, even though they are often arbitrary and do not reflect any theoretical content or prior knowledge. Moreover, functional form assumptions can engender bias, whenever they fail to accurately capture the true complexity of the system. In this article, we introduce causal-graphical normalizing flows (cGNFs), a novel approach to causal inference that leverages deep neural networks to empirically evaluate theories represented as DAGs. Unlike conventional methods, cGNFs model the full joint distribution of the data using a DAG specified by the analyst, without relying on stringent assumptions about functional form. This enables flexible, non-parametric estimation of any causal estimand identified from the DAG, including total effects, direct and indirect effects, and path-specific effects. We illustrate the method with a reanalysis of Blau and Duncan’s ( 1967 ) model of status attainment and Zhou’s ( 2019 ) model of controlled mobility. The article concludes with a discussion of current limitations and directions for future development.


  • Increasing the Confidence of Predictive Uncertainty: Earth Observations and Deep Learning for Poverty Estimation

    by Mohammad Kakooei, Adel Daoud

    Reducing global poverty, particularly in low- and middle-income countries, is a critical objective of the Sustainable Development Goals (SDGs). To track progress towards these goals, high-frequency, granular geo-temporal data that captures changes at the neighborhood level is essential for researchers and policymakers. Recent advancements in methodology have combined machine learning (ML) and earth observations (EO) for poverty estimation, thereby addressing significant data gaps. However, a notable limitation of these EO-ML methods is their frequent deployment without a robust mechanism to quantify predictive uncertainty. Understanding this uncertainty is crucial for making informed decisions, effectively managing risks, and instilling confidence in users and stakeholders regarding the model’s predictions. Although deep learning (DL) offers methods to quantify predictive uncertainties, their reliability is often constrained, failing to accurately reflect the underlying variations in predictions. Our proposed method aims to enhance confidence in predictive uncertainty without sacrificing accuracy. It begins by integrating an external model to explicitly capture data variability. Subsequently, we employ two orthogonal metrics—accuracy and uncertainty—to evaluate the influence of training data, especially in scenarios involving satellite imagery (e.g., selecting a subset of source domain countries for prediction in the target country). By applying these metrics, we formulate criteria to assess the importance of choosing specific countries from the source domain as training data. Our analysis highlights the effectiveness of this methodology in situations where the target country offers high-dimensional data, like satellite images, but faces a shortage of adequate training samples for deep learning models.


  • A theory of famines—A response

    by Adel Daoud