Networks and ideological polarization

Social Networks and Polarization [Job Market Paper]
-Featured in the World Bank’s Development Impact Blog series on 2026 Job Market candidates.
-Shortlisted for the London School of Economics Reform Club Economics and Current Affairs Award.

Abstract: Ideological polarization is increasing across multiple issues and contexts, with potentially serious implications for social cohesion and political stability. A common explanation is that people consume confirmatory information through networks of opinion exchange that create “echo-chambers”, “filter bubbles” or “outrage engagement”. However, social learning models rarely incorporate such mechanisms and typically result in consensus rather than long-term polarization. I build a theoretical social learning framework that can generate consensus, persistent disagreement or even opinion divergence by integrating endogenous pro-attitudinal attention with DeGroot learning. This allows me to explore how different social learning mechanisms interact with network structure to influence the effectiveness of policies aiming to improve social cohesion. Using spectral methods, I derive measures of a network’s polarizability that depend on its structure and how opposing views are processed, offering testable predictions and feasible measures for understanding a given setting’s risk of long-term division. I show that policies thought to reduce polarization, such as increasing inter-group dialogue, can be counterproductive if they fail to consider how backlash-style effects might affect network dynamics. Instead of indiscriminately creating connections between groups, a more targeted approach focusing on critical agreeable relationships is suggested as a safer and more effective route to addressing polarization. Alternative interventions that amplify the voices of moderates (relative to extremists) or widen tolerance towards opposing views are also shown to have less theoretically ambiguous effects. I propose methods for identifying the conditions that favour different approaches in practice.

Counteracting extremism on social media: testing approaches to deter young South African men from violent gender-related ideologies
An RCT testing implications of my job market paper, in partnership with J-PAL Africa.
Funding: The Hub for Equal Representation in the Economy, LSE

The role of trust in vaccine confidence: insights from a longitudinal survey in seven Sub-Saharan African countries with Alex De Figueiredo and Heidi Larson
Bayesian Hierarchical Analysis, testing implications of my Job Market Paper. Data collection in progress.
Project summary, Interim policy report

Networks and labour markets / human capital development

Network effects in labour markets with socially mediated search and human capital
A model of how social networks affect employment and inequality under network assortativity – a common descriptive pattern LMICs of people being socially connected to others with similar employment outcomes. Extends the framework of Calvo-Armengol and Jackson (2004) to include network-mediated human capital acquisition, using insights from the empirical investigations detailed below. 

Making connections: exploring the impact of social network assortativity on the labour market for South African youth
Field experiment in South Africa exploring the impact of strengthening professional networks of entry-level workers. Piloting in progress. Funding: Agency Fund, J-PAL Jobs and Opportunities Initiative, IGC Cities, STICERD, LSE

Measuring the labour market quality of social networks with Raquel Caldeira & Patrizio Piraino
A shorter term study experimentally testing different Aggregate Relational Data (ARD)-based approaches to measuring the Labour Market Networks of African Youth. Data collection in progress.
Incorporated into the above project’s piloting stage, with the same funders: Agency Fund, J-PAL Jobs and Opportunities Initiative, IGC Cities, STICERD, LSE

Shifting expectations: An analysis of youth labour market expectations in the context of a work-based-learning programme with Rocco Zizzamia and Patrizio Piraino
Registered PAP A mixed-methods study of how expectations evolve during entry-level work, including a field experiment testing the impact of a goal-setting intervention. Data collected, analysis in progress.
Funding: IDRC, J-PAL PPE

The Role of Social Identities in Economics Education with Dimitra Petropoulou & Ben Dahmen
Funding: The Hub for Equal Representation in the Economy, LSE

Causal inference with network spillovers

RCT-based causal identification with network spillovers with Sreevidya Ayyar.
A survey of methods for the design and analysis of RCTs for researchers who wish to avoid or quantify network spillovers.

Publications and working papers

Age-Dependent Changes in Intuitive and Deliberative Cooperation in Nature Scientific Reports, 13, 4457 (2023), https://doi.org/10.1038/s41598-023-31691-9 with Elena Nava, Francesco Nava and Francesco Margoni
A lab-in-field RCT investigating how cooperative behaviour evolves with age and decision time.

[Pre-PhD: Case Study of an RCT] Helping young people in South Africa bridge the gap between intention and behaviour in their search for work – Chapter 10 of OECD (2018), Case Studies on Leaving No One Behind: A companion volume to the Development Co-operation Report 2018, OECD Publishing, Paris, https://doi.org/10.1787/9789264309333-en – with Martin Abel, Rulof Burger, Eliana Carranza & Patrizio Piraino.

Other research-related output (lessons from fieldwork)

I had a professional career in research before I started doing my own studies. As the former head of J-PAL Africa’s research vertical, I worked on many RCTs in developing countries, managing over 400 field staff and working together with complex research partner organisations and many principal investigators to produce high quality research amid some serious operational constraints. See my policy page for more.

Spending so much time in the field made me borderline obsessive about getting the measurement of core outcomes right (harder than it looks), minimising attrition and building a culture of pursuing data quality excellence among field staff. Here are some outputs I (co-)created on these topics:

Surveying young work seekers in South Africa: tips from J-PAL Africa’s Research Team, University of Cape Town’s Southern Africa Labour and Development Research Unit (SALDRU) – focused on young work-seekers but lessons can be applied to many phone-surveying settings – with Emmanuel Bakirdjian and Wim Louw.
Featured in World Bank DIME Wiki on CATI, Conducting Phone Surveys in a Pandemic, and J-PAL Best Practices for Conducting Phone Surveys

J-PAL Research Resources
An excellent selection of publicly available resources for designing and running RCTs, used heavily in J-PAL’s field staff training. Here are some that I co-authored or contributed to:
Surveyor Hiring and Training
Field Team Management
Survey Logistics
Measurement and Survey Design

Cape Town Data Quality Workshop: Measurement of Development Indicators talk at conference hosted by DataFirst at the University of Cape Town on lessons improving data quality when using phone surveys in South Africa.

Digital data collection for rigorous, randomised research Training Workshop at the 2017 Global Evidence Summit.