Modelling collective action to change social norms around domestic violence: social dilemmas and the role of altruism

The paper addresses how community mobilisation can reduce domestic violence and why such interventions sometimes succeed or fail. Domestic violence, including intimate partner violence affecting an estimated 30% of women globally, imposes severe human and economic costs. While community mobilisation has reduced violence in some low- and middle-income contexts, mechanisms remain unclear and recent trials show mixed impacts. The authors propose using a formal game-theoretic model as a thought experiment to clarify assumptions, generate consistent explanations, and explore conditions under which collective action to shift gendered social norms reduces violence. The central research question is whether and how social norms make violence prevention a social dilemma for women, and what mechanisms (e.g., forms of altruism) can overcome it.

Using an adapted socio-ecological framework, the paper reviews determinants of domestic violence spanning individual, relationship, community, and societal levels, emphasising patriarchal ideology and rigid gender roles. Two feminist theories—Hegemonic Masculinity and Male Peer Support Theory—suggest men are rewarded by peers for enforcing stereotypical gender roles, with peer sanctions/rewards shaping perpetration. Empirical studies show women’s perceived disobedience is a frequent justification for violence; acceptance of wife-beating correlates with victimization and peer acceptance correlates with perpetration. Men engaging in prevention face peer resistance, and individual-focused interventions often fail where norms condone violence. Evidence from community mobilisation highlights diverse, community-driven actions and the role of organised diffusion in changing norms. This synthesis motivates a formal model where men’s attitudes, peer norms, costs of violence, and women’s costly preventive actions interact.

The authors develop a two-stage, multi-player game-theoretic model representing a locality with n couples (women wi and partners hi). Positive assumptions: (I) Men’s attitudes to violence (ai) drive perpetration utility; (II) Social norms: peers impose rewards/sanctions (ηij) on men based on violence; (III) Perpetration incurs costs (ci vi); (IV) Women can take costly action (effort eij) to change men’s attitudes (ai(e) strictly decreasing, convex), with women’s effort cost djej; (V) Women suffer disutility from violence s_i(1−ti)vi with tolerance ti ∈ [0,1). Normative assumption (VI): Assess women’s welfare setting tolerance to zero to address adaptive preferences. Game structure: Stage 1—women choose efforts eij to influence men’s attitudes; Stage 2—men choose violence levels vi. Payoffs are specified for women and men via utility functions; women’s welfare is evaluated with tolerance set to zero. Under regularity conditions, a unique subgame perfect Nash equilibrium exists (Proposition 1) linking optimal vi and eij to marginal effects on attitudes and costs. Comparative statics (Corollary 1) show equilibrium violence generally decreases with stronger social norm responsiveness to effort and increases with tolerance and effort costs, while effects of external punishment costs can be ambiguous. The model identifies positive externalities from women’s efforts on neighbours’ relationships, generating social dilemmas (Theorem 1): a subset of women can improve welfare via coordinated effort beyond equilibrium levels, but unilateral action is not individually rational. The authors simulate scenarios: Scenario 1 (weak individual impact on norms, small community) varies parameters for tolerance, cost/effectiveness of effort, and cost of violence; Scenario 2 increases peer norm strength and community size to show potential backfire when increasing costs of violence; Solutions are modelled by augmenting women’s utility with altruistic motives: process-based altruism (intrinsic benefit b to effort), empathetic altruism (weight η on others’ suffering), and reciprocal altruism (matching others’ effort at rate ρ=p). Propositions 2–4 provide conditions under which increasing altruism improves welfare and discuss equilibrium existence (no equilibrium for high reciprocity p≥1). Scenario 3 explores effects of each altruism type in a homogeneous population; Scenario 4 introduces heterogeneity across couples (parameters varied 1/3× to 3×) to examine distributional effects of altruism.

- Social norms create a social dilemma: It is individually rational for women to under-invest in preventive action, yet all women would be better off if everyone invested more (Theorem 1). Positive externalities from norms change underlie this collective action problem. - Conventional levers may have limited or perverse effects: In Scenario 1, reducing women’s tolerance for violence or increasing the effectiveness of effort had almost no discernible impact on effort, violence, or welfare; reducing effort costs only mattered when costs approached near zero. Increasing costs of perpetration reduced violence but largely independent of women’s effort (effort stayed near constant). - Potential backfire from external punishment: In Scenario 2, raising the cost of perpetration initially increased violence before eventually reducing it after costs more than doubled. Anticipated external deterrence can cause women to reduce their own efforts, offsetting or reversing benefits. - Altruism can overcome the social dilemma: In Scenario 3, modest increases in process-based or empathetic altruism immediately increased women’s efforts and reduced violence, improving welfare. For example, increasing average effort from 0 to 1 unit halved experienced violence; achieving a comparable reduction via unilateral action on one’s own partner would require nearly 100× more effort. Reciprocal altruism had limited effect until reciprocity approached p≈1, after which effort and violence changed rapidly; for p≥0.1 equilibrium approached non-existence boundaries in the simulation setup. - Heterogeneity creates unequal impacts: Scenario 4 showed substantial variation across couples; while median outcomes improved with altruism (e.g., median violence drop of ~10 units and welfare gain of ~5 units for a +1.0 intrinsic benefit), some women could be worse off at high altruism levels even as the median and maximum improved. - Context matters: Social dilemmas are most acute in larger communities where norms are weakly responsive to any one woman’s effort. In settings lacking strong domestic-violence-related social norms (e.g., socially isolated or secrecy cultures), collective action may yield fewer returns, and individual interventions may be more appropriate.

The model clarifies why coordinated collective action is often necessary to reduce domestic violence when social norms reward or punish men’s behavior. It shows that individualistic approaches can be ineffective in the presence of strong norm externalities, and that standard strategies—like lowering tolerance or modestly reducing effort costs—may not shift equilibrium behavior. Importantly, harsher external sanctions can backfire if they lead women to reduce their preventive efforts before sanctions become sufficiently strong to dominate norms. The analysis highlights altruism as a mechanism to internalize externalities: process-based and empathetic altruism effectively increase efforts and reduce violence in homogeneous populations, while reciprocal altruism can trigger sharp changes near critical thresholds but may pose stability issues. The paper also distinguishes contexts: where social norms around domestic violence are weak or information is limited, social dilemmas may not arise, suggesting less need for mobilization and greater efficiency of individual-level interventions (e.g., counselling, economic empowerment). The discussion urges reevaluating cost-effectiveness to account for externalities (spillovers) from participation and to recognize moral dimensions: women may be individually justified in abstaining from action while collectively incurring avoidable harms. The work situates itself relative to prior household-bargaining models by incorporating community-level norms and collective action dynamics.

The authors integrate feminist theories of patriarchal norms, male peer support, and hegemonic masculinity into a formal game-theoretic model of community-driven social change. Under broad conditions, prevention of domestic violence becomes a social dilemma requiring coordinated action. The model predicts that fostering process-based, empathetic, and reciprocal altruism can mobilize collective action and reduce violence, whereas commonly used levers (e.g., lowering tolerance, small reductions in effort costs, or moderate increases in punishment) may have limited or even adverse effects. The paper provides a theoretical basis for future empirical research on altruism and collective action in violence prevention and calls for interdisciplinary collaborations and trials to test these mechanisms and refine models of system dynamics in community mobilisation.

- The model assumes rational utility maximization with social and psychological components; real-world decision-making may deviate from these assumptions. - It models women’s collective action but not men’s collective participation, which may follow different incentive structures and ethical considerations. - Beliefs (efficacy, pluralistic ignorance) and belief updating are not explicitly modeled; misperceptions could materially alter dynamics. - Parameter choices in scenarios are illustrative and not derived from empirical calibration; quantitative results are not forecasts. - The model focuses on social interactions and norms and does not capture the full complexity of domestic violence ecosystems or institutional responses. - Reciprocal altruism at high levels may threaten equilibrium existence, indicating potential instability not explored empirically.