An ERC Consolidator Grant led by Efe Tokdemir at Bilkent University, 2026–2031. Framework, work packages, and research design.
COUNSTER introduces a novel triadic framework to investigate government strategies for combatting radicalization, insurgency, and terrorism by focusing on the role non-state armed actors’ (NSAAs) constituents play in conflict processes.
While the existing literature extensively addresses the military aspects of conflict, how governments strategically engage with NSAAs’ constituents to sever the ties between the two — a strategy this project terms counstering — has been widely neglected despite its prevalence. COUNSTER addresses this gap by arguing that we should turn our focus toward modeling the causes and outcomes of counstering efforts, given that military power is neither the only, nor the most effective, tool in governments’ arsenal.
“The ultimate victory will depend on the hearts and minds of the people who actually live out there.” — Lyndon B. Johnson, 1965
Decades of protracted internal conflict — from long-running insurgencies to post-invasion counterinsurgency campaigns — show that military measures alone rarely resolve root causes; they often generate new grievances instead. COUNSTER therefore asks: why, how, and under what circumstances do governments fight non-state armed actors (NSAAs) by targeting the constituents those actors claim to represent — and why do some of these counstering strategies succeed while others fail?
The first generation of conflict studies (counterterrorism) overemphasised dyadic government–NSAA relations. The second generation (counterinsurgency) brought citizens into the picture, but focused on military-centric, offensive strategies that are repeatedly shown to be counterproductive. COUNSTER proposes a third approach: putting the constituency — the audience whose support both governments and NSAAs compete for — at the centre of the analysis.
Choosing between counterterrorism, counterinsurgency, and constituency-centred strategies — attractive or repressive — to end armed challenges.
Insurgent and terrorist organisations whose survival depends on funding, recruits, and legitimacy provided by their support base.
The communities NSAAs claim to represent — the pivotal audience for whose allegiance both sides compete.
Three work packages run in parallel across the five-year grant, each feeding a different objective and cross-validating the others.
The first open-access, comprehensive and dynamic dataset on Government–Constituency Relations. Combining manual coding of 9,000+ events with machine learning and computational social science tools, GovConReD covers attractive and repressive counstering measures from 1980 to 2030 across 120 countries, 130+ constituencies, and around 350 non-state armed actors.
Extending the PI’s existing collection of 9,000+ events, developing the conceptual framework for repressive tactics, and writing the GovConReD codebook — measure type, initiator, demographic and geographic coverage, institutionalisation, implementation pace, novelty, conditionality, and schedule.
NLP models scan open news sources at scale to extend the time and content coverage of the data. Measures are coded automatically first, then re-coded manually on samples as a robustness check.
The completed dataset is published open-access for scholars and practitioners: 50 years of attractive and repressive counstering measures, coded event by event with their full feature set.
Large-N comparative analysis of the predictors of counstering — which constituencies are targeted with which tactics, and why (Objective 1) — and of the outcomes those strategies produce for conflict trajectories (Objective 2).
Micro-level data on how constituents actually perceive and react to government strategies — qualitative fieldwork in Türkiye, Colombia, India, and the Philippines, followed by two rounds of online conjoint experiments in the same four contexts.
Semi-structured elite interviews and four focus groups in each country, run with local scholars, NGOs and research firms. They establish how and why particular counstering policies were adopted or dropped, what they were meant to achieve, and how different audiences read them.
Recurring themes extracted from the qualitative material map the pathways between a government measure and a constituent’s reaction — and inform the experimental design that follows.
Two rounds of online conjoint experiments in the four contexts, manipulating the attributes of counstering measures — type, timing (novelty, schedule, conditionality), coverage, initiator features, implementation pace, and institutionalisation level.
Average marginal component effects compared across treatments isolate the causal contribution of the strategy itself, and expose the gap between what governments intend and how constituents respond (Objective 3).
A methodological innovation: conflict scholars selected for their field experience and local knowledge in the four case contexts provide expert assessments that triangulate the macro- and micro-level findings, overcoming media bias, government intimidation, and preference falsification in conflict zones.
Building a pool of country experts for Türkiye, Colombia, India, and the Philippines, chosen for direct field experience and local knowledge of the conflicts under study.
Surveying the pool on when, why, and under whose supervision governments initiate concessions; which strategies they use to build constituency relations; what conditions their success; and what causes the gap between expected and actual outcomes.
Expert assessments facilitate micro-level data collection, measure the gap between the macro- and micro-level analyses, and reduce the empirical risks attached to Objective 2.
Upon release, GovConReD will be freely available to scholars and practitioners, enabling comparative, empirical research on strategies against radicalisation, insurgency, and terrorism.
More on the Data pageThese countries host 21 conflicts that have claimed more than 200,000 lives — 10% of all conflict-related casualties worldwide.
Transforms the study of conflict from a dyadic to a triadic framework (Government–NSAA–Constituency), challenging the “one-fits-all” approach and relaxing the unitary-audience assumption.
Combines large-N observational data at the macro level with focus groups and conjoint experiments in four countries to examine how constituencies actually react to counstering strategies.
Helps NGOs, international organisations, and governments design measures against radicalisation, insurgency, and terrorism — contributing to the goals of the EU Strategic Compass for Security and Defence 2030.
Triangulates macro-level observational and micro-level experimental data with expert opinion surveys, addressing data limitations and biases common in conflict-zone research.