The Three Blind Spots of Haiti’s Security Crisis

What the ASA survey reveals when we stop looking where everyone else is looking.

In January 2025, the Aksyon Sitwayen Angaje (ASA) consortium, which brings together four Haitian civil society organizations (Nègès Mawon, Initiative de la Société Civile, Koze Jèn Yo, and FIBESR), carried out a national opinion survey on the country’s security and political situation, in collaboration with the firm ProEco Haïti. Funded by the International Organisation of La Francophonie after an interruption caused by the suspension of funding from the National Democratic Institute (NDI), the study is one of the few empirical bases available for grasping, across the whole country, how Haitian citizens experience and perceive the crisis.

Its value rests on several features. First, its coverage: 1,291 adults interviewed face to face in the country’s ten geographic departments, including in rural areas (61.6% of the sample). Second, its methodology: sampling by the LQAS method (Lot Quality Assurance Sampling), validated in Haiti by CARE in 2020, with weighting by sex and by department based on data from the Haitian Institute of Statistics and Informatics (IHSI), a 95% confidence interval, and a 4% margin of error. Finally, its scope: the survey is not limited to security in the narrow sense. It covers access to essential services (food, health care, water, electricity, transport, telecommunications, education), forced displacement, institutional trust, voting intentions, and gender dynamics. It is this breadth that makes the document exceptional: it allows the security crisis to be read not as an isolated phenomenon, but as a system whose ramifications reach every dimension of national life.

It must be made clear from the outset, however, that the survey collects reported perceptions, not absolute counts of crimes or of displaced people. The figures that follow are percentages of respondents who consider that a given service is disrupted, or that a given household has been displaced or has lost a job. This is a real methodological limitation. But in public policy, the perception that triggers departure, non-return, mistrust, or electoral abstention carries strategic weight at least equal to that of the objective fact.

What does this survey reveal when it is read carefully?

There is, in the way we talk about insecurity in Haiti, a mechanism found in almost every urban conflict in the world: the eye fixes on the flames and ignores the smoke. We know where the hotbeds of violence are. Port-au-Prince. The Lower Artibonite. These names saturate the news bulletins and steer security budgets. The ASA data confirm it: the Ouest department records 32.1% forced displacement, 46.5% job losses, 58.9% disruption of food access. The Artibonite leads on kidnapping (15.3% of households having contributed to a ransom). None of this surprises anyone.

But the richness of the ASA survey lies elsewhere: in what it reveals when we look away from the epicenters. Three phenomena then emerge that public debate has not yet absorbed. Three blind spots in our reading of the security crisis. The first concerns what insecurity does where it does not shoot. The second, what it costs the women whom security discourse mentions without really listening to. The third, what is being quietly rebuilt from below while the institutions at the top collapse.

These three findings are linked by a common thread: in Haiti, everything that is distant fails; everything that is close holds up. Violence spreads through functional dependence on a distant central node. Women’s distrust is built by the absence of local services. Trust is rebuilt where institutions are tangible and local. This principle, put forward here as a hypothesis, will be tested against the data in each of the three sections that follow.

I. First blind spot: invisible contagion

Three figures from the ASA survey raise a problem. The Nord-Ouest department reports 72.6% disrupted transport. The Nord reports 76.1% affected phone signal. The Sud-Est reports 62.3% disrupted access to health care and 68.2% disrupted transport. Yet none of these departments is among the epicenters of armed violence. Their forced displacement rates remain low (7.7% in the Nord, 7.4% in the Sud-Est).

The explanation is in the ASA report itself: the telecommunications companies’ central systems are located in Port-au-Prince, where they are directly affected by the gangs. Violence does not physically travel to the Nord. It arrives there through cables, supply chains, blocked roads.

The contagion function

The impact suffered by a department j that is not itself violent can be expressed as a function of its dependence on the central node:

The idea is simple: the impact suffered by an area depends on the intensity of crime in the source area, on its degree of functional dependence on that source (denoted w: shared telecoms, common roads, supply chains), and on distance. The stronger the dependence, the more severe the contagion, even over great distances. It is connectivity, and not geographic proximity alone, that determines the extent of the spread.

To give a concrete measure of this contagion, we need an indicator that avoids a common methodological trap: selecting, for each department, only the most disrupted service. This practice, known as cherry-picking, artificially inflates the differences. To avoid this bias, we build a composite index of service disruption, defined as the arithmetic mean of three indicators explicitly available by department in the ASA report: the rate of disruption of transport, of access to health care, and of access to food.

This index measures the average perceived disruption of services in a given department. We then define the contagion ratio ρ as the ratio of this composite index to the forced displacement rate, which serves as a proxy for direct local violence:

When ρ is close to 1, service disruption is proportional to local violence: the department “produces” and “suffers” in comparable proportions. When ρ is well above 1, the department suffers disruption out of proportion to its own violence: it imports destabilization. Table 1 presents the results for the five departments for which all three indicators are explicitly available in the ASA report.

Table 1. Composite disruption index and contagion ratio. All columns are in percent, except the ratio in the last one. Author’s calculations based on the ASA/ProEco Haïti survey, January 2025 (n = 1,291). ICS: composite index of service disruption, the mean of the three self-reported indicators that precede it. Displ.: reported forced displacement rate, a proxy for direct violence. Ratio: the ρ of equation 3, the ratio of the ICS to forced displacement. The Nord-Ouest, with 72.6% disrupted transport, is not included because the report gives no explicit data on health care and food for it.

The hierarchy is clear. The Ouest, the epicenter of the violence, shows a ρ of 1.6: its service disruption is roughly proportional to its direct violence. The Sud (1.9) and the Centre (2.7) fall within comparable orders of magnitude. The Nord and the Sud-Est, by contrast, show ratios of 5.4 and 7.9: they report service disruption five to eight times greater than their level of direct violence would suggest.

Empirical check: Spearman’s correlation

If service disruption were purely local, that is, proportional to the violence in each department, we would expect a high rank correlation between the forced displacement rate and the composite disruption index. Spearman’s correlation coefficient, calculated on the five departments in Table 1, makes it possible to test this. The ranks are as follows:

Table 1b. Departmental ranks for the Spearman calculation. Rank D: rank by forced displacement. Rank C: rank by composite index. d²: square of the difference between the two ranks.

The result speaks for itself. Spearman’s coefficient between direct violence and service disruption is 0.30: a weak correlation. With n = 5, this coefficient is not statistically significant (the critical value at 5% for n = 5 is about 0.90). But the very weakness of this correlation is informative. If service disruption were determined locally by violence, we would observe an $r_s$ close to 1. The fact that it is 0.30 means that direct violence explains only a fraction of service disruption. Other factors are at work, and the main one is functional dependence on the central node. The Sud-Est, which drags the correlation down (d² = 9), is the most telling case: low violence, high disruption. It is the very hallmark of invisible contagion.

Two methodological caveats are in order. First, we are working with five observations, which rules out any claim to statistical robustness; Spearman’s correlation is used here as an exploratory tool, not as formal proof. Second, the three indicators used are reported perceptions, not objective measures of how services are functioning; the ratio ρ therefore measures a perceived gap, not physical contagion. With these limits stated, the convergence between the ratio ρ and Spearman’s correlation paints a coherent picture: service disruption in Haiti cannot be reduced to a local problem.

The strategic implication is direct. Securing Port-au-Prince does not serve Port-au-Prince alone. It means unblocking the phone signal in the Nord, transport in the Nord-Ouest, supplies in the Sud-Est. The central node concentrates the violence; but stabilizing it would release a national security dividend, measurable by the expected fall in the ρ ratios of the outlying departments.

II. Second blind spot: the gendered cost of insecurity

Among people displaced by insecurity, 49.4% of men are considering returning to their former neighborhood if the situation improves. Among women: 33.0%. A gap of sixteen points. This figure, buried in the gender section of the ASA report, is one of the most consequential in the document.

Two different calculations of return

Men and women do not measure “security” in the same way. The ASA report puts it clearly: for women, security is not limited to the absence of violence. It includes access to health care, children’s schooling, residential stability. This difference can be expressed by two functions:

In this first model, a displaced man reasons with two variables: the income he can recover in his area of origin and the risk of violence that persists there. If violence falls, the calculation tips and he returns. The function for women is different:

Here, four variables come into play. Even if violence decreases (β falls), the return remains negative if services are not restored (γ: schools, health care, water, transport) or if the risk to children persists (δ). The threshold for women’s return is structurally more demanding than the threshold for men’s. That is exactly what the 33% of women who are not considering returning are saying.

The following table quantifies the gender gaps measured by the survey.

Table 2. The gendered cost of insecurity: measured gaps. The Women and Men columns are in percent, the Δ gap in points. Source: ASA/ProEco Haïti survey, January 2025. The last column links each indicator to the variables in equations 5a and 5b; the last three indicators measure an orientation, not a parameter of the model.

Each row of the table feeds the model. Women have less income (α: 37.2% with no economic activity, versus 18.7%), more fragile access to services (γ: 64.6% reporting days without eating, versus 57%), and a weaker sense of security even after displacement (β: 81.4% versus 94.1%). Every variable pushes in the same direction: the threshold for women’s return is structurally higher than the threshold for men’s.

The direct consequence: a security strategy that restores order by force without restoring services (health, education, water, transport) will not bring women back. And without the return of women, there is no rebuilding of informal markets, no resumption of schooling, no restoration of community life. To define security as nothing more than the absence of armed violence is to use a male indicator in a country where 57.3% of the adult population is female.

III. Third blind spot: trust rising from below

In a landscape of widespread mistrust, one figure stands out. Local authorities, that is, the town halls, are distrusted by only 33.8% of respondents, compared with 60.3% for political parties, 52.4% for the presidential council in office at the time, and 51.0% for the Office of the Prime Minister. The hierarchy is clear: the further an institution is from people’s daily lives, the more it is rejected.

Table 3. The gradient of institutional trust. The two mistrust columns are in percent, the Δ gap in points. The last column indicates how close the institution is to citizens. Sources: ASA/ProEco 2025 and OCID 2021. The 2021 data come from a survey with a different methodology: the comparison is indicative. The CPT, the Transitional Presidential Council, has since ceased to exist. “n/a”: data not available for 2021.

The most striking figure is the Δ for local authorities: -22.6 points between 2021 and 2025. During the most violent period in Haiti’s recent history, trust in town halls rose by 22 points. It is counterintuitive. That is precisely why it is significant.

This gradient follows a rule that can be formalized:

Three factors determine trust in an institution: its physical and relational distance from citizens, its perceived presence in daily life, and its capacity to produce concrete, observable results. Town halls come out on top on all three. The Office of the Prime Minister and the political parties come out at the bottom. The last column of Table 3 brings out this correlation between proximity and trust. It is no coincidence. It is a gradient. And this gradient has a direct implication: the legitimacy of the Haitian state will not be rebuilt from the top. It will be rebuilt from the town halls, the communal section councils (CASEC), the places where citizens can see, touch, and judge public action.

One last figure cements this finding: 95.4% of respondents believe that security must be restored before any election. Electoral legitimacy is conditional on security legitimacy. And security legitimacy is demonstrated on the ground, not in palaces.

IV. One principle, three consequences

The three blind spots converge on a single principle: in Haiti, everything that is distant fails; everything that is close holds up. Violence spreads through functional dependence on a distant central node. Women’s distrust is built by the absence of local services. Trust is rebuilt through local institutional presence. The same logic. Three manifestations.

This convergence translates into three measurable strategic shifts:

Table 4. Strategic shifts, each grounded in a measure drawn from the data.

Formally, the partial derivative of national security with respect to the resources deployed at the central node is greater than that of a uniform deployment:

In other words, a gourde invested in stabilizing Port-au-Prince yields a higher national security return than the same gourde spread evenly across ten departments, because stabilizing the node frees the departments that depend on it. But this equation is not enough if “security” is not redefined to include the threshold that matters to women (services, stability, predictability), and if institutional investment does not target the places where trust still survives: local authorities.

What 1,291 voices are asking us to hear

We should take the measure of what this survey represents. For eleven days in January 2025, in a country where travel is a risk and access to the field a constant negotiation with violence, 66 interviewers crossed the ten departments to collect, face to face, the words of 1,291 citizens. These are perception data, not absolute counts. But in a country where the statistical apparatus has been dismantled, where no census has been carried out for years, and where administrative records are patchy when they exist at all, this survey is one of the few structured testimonies to what the Haitian population is living through. To ignore it would be an intellectual failing. Not to exploit it fully would be a waste.

Read carefully, this survey says three things that public debate still refuses to put into words. It says that insecurity destroys far beyond its visible perimeter: three departments report service disruption between six and ten times greater than their level of direct violence would suggest. It says that women pay a cost that our indicators do not capture, and that their failure to return jeopardizes the very reconstruction of the neighborhoods we claim to be securing. It says, finally, that public trust has not dissolved uniformly: it survives where institutions are close, tangible, observable, and it is from these places that the state could, if it wished, begin to rebuild itself.

The question, then, is not how many police officers to deploy. It is where to deploy them, so that stabilizing the central node releases a national dividend. It is for whom security is meant, knowing that women will not return unless the services that make life possible are restored. It is where to anchor the rebuilding, knowing that trust takes root in town halls and not in palaces.

Haiti does not lack diagnoses. The country is buried under reports, assessments, and matrices of recommendations. What is lacking is the capacity to read the available data with the rigor and honesty they demand, rather than through the prism of preconceived certainties. The ASA survey offers that opportunity. It lays the groundwork for a different conversation about security, based not on what we think we know, but on what citizens say they are living through.

Three blind spots. A single flaw in the lens. And a collective responsibility: not to look away once again.

Methodological limitations

Rigor requires that the limits of the exercise carried out in these pages be stated explicitly. First, the ASA survey data are reported perceptions, collected through face-to-face self-reporting. They may be affected by recall bias, social desirability bias, or availability bias. The ρ ratios measure a perceptual gap, not objectively documented physical contagion.

Second, territorial coverage is uneven. As the ASA report itself points out, the most dangerous areas of the Ouest and the Artibonite could not be covered because of access constraints. The results for these two departments therefore reflect the situation in accessible areas more than that of the most violent epicenters. This means that the disruption rates for the Ouest are probably underestimated, and that the Ouest’s ρ ratio (1.6) could in reality be closer to 1.

Third, Spearman’s correlation ($r_s$ = 0.30) is calculated on five observations. With such a small sample, it does not reach the threshold of statistical significance and can serve only as an exploratory indication, not as formal proof. Fourth, the longitudinal comparison with the 2021 OCID survey must be interpreted with caution: the methodologies, questionnaires, and contexts differ, which limits the significance of the changes observed.

Finally, the composite index of service disruption rests on three of the seven indicators documented in the report, chosen because they are explicitly available by department. An index incorporating all seven indicators (transport, telecoms, health care, water, electricity, schools, food) would be more robust, but the disaggregated data are not all explicitly reported for every department. These limits do not invalidate the trends identified, but they set the bounds of their validity: this is an analytical reading based on the best available data, not an econometric demonstration in the strict sense.

References

Aksyon Sitwayen Angaje (ASA) and ProEco Haïti (2025). Rapport d’enquête d’opinion sur la situation sécuritaire et politique en Haïti. With the support of the OIF and of France’s Ministry for Europe and Foreign Affairs (MEAE).

Braga, A., Papachristos, A., and Hureau, D. (2014). The Effects of Hot Spots Policing on Crime. Justice Quarterly, 31(4), 633–663.

CARE and UN Women (2020). Analyse rapide de genre: Haïti.

CNSA (2024). Analyse de la sécurité alimentaire en Haïti, mars-juin 2024.

OCID (2021). Enquête nationale sur la gouvernance démocratique en Haïti.

Sherman, L., Gartin, P., and Buerger, M. (1989). Hot Spots of Predatory Crime. Criminology, 27(1), 27–55.