The CEP’s dashboards show continuing registration, unequal participation and enormous rejection counters. Read alongside documented political ties to armed groups, they reveal why administrative activity has yet to produce a credible path to elections.
Haiti’s electoral machinery is recording activity. It is not yet demonstrating that it can deliver an inclusive, credible national vote.
The CEP dashboard screenshots reviewed for this analysis show 881,523 voter registrations, 6,657 candidate registrations and 1,351,535 recorded political memberships. Alongside those figures sit three membership-rejection counters whose arithmetic sum reaches 28,840,869.
That last number demands an explanation. It does not establish that 28.84 million different people attempted electoral fraud.
The distinction captures Haiti’s larger electoral problem: registration activity, institutional integrity and election readiness are different things. Software can process applications while insecurity excludes communities, political networks compromise competition and authorities revise the calendar.
The evidence supports describing democratic renewal as stalled. It does not support claiming that every electoral operation has stopped—or that these charts alone prove corruption inside the CEP.
Four processes, four different populations
The supplied material covers voter registration, political-organization membership, independent-candidate sponsorship and candidate registration. These figures cannot be combined into one electoral participation total.
A recorded party member is not necessarily a newly registered voter. A sponsorship is not a vote. An online candidacy is not an approved candidacy.
The live candidate dashboard corroborated the headline candidate counts in the screenshots. It also explicitly states that the figures are preliminary and that final candidacies depend on verification and validation. Appearing online does not establish legal eligibility or acceptance onto the final ballot.
The voter total also needs a clearly defined denominator. CEP communications describe outreach to first-time voters, while its electoral-register directorate maintains and updates the general electoral list. Until the CEP explains the dashboard’s relationship to the existing register, 881,523 should be reported as the displayed registration total—not automatically as Haiti’s entire eligible electorate. Dividing it by the total population would not produce a valid registration rate.
A registration network that needs a better readiness test
The screenshots show 1,306 active registration centers out of 1,442 listed centers, alongside 2,977 active operators. Approximately 90.6% of listed centers are classified active; 136 are outside that category.
That demonstrates continuing operations. It does not establish equal access.
An “active” center designation says little without operating hours, days actually open, equipment availability, connectivity and residents’ ability to reach it safely. The screenshots do not explain why each of the other 136 centers is not classified active. Attributing all of them to gang violence would exceed the evidence.
The geographic figures illustrate the problem:
| Area as labeled | Recorded registrations | Displayed configured-capacity utilization |
|---|---|---|
| OUEST 1 | 144,346 | 16.59% |
| SUD-EST | 95,862 | 22.92% |
| NORD | 76,782 | 8.79% |
| NORD-EST | 62,425 | 16.51% |
| ARTIBONITE | 58,450 | 7.44% |
| CENTRE | 67,642 | 16.41% |
| SUD | 135,521 | 17.32% |
| GRANDE-ANSE | 103,554 | 20.60% |
| NORD-OUEST | 67,823 | 13.30% |
| NIPPES | 69,118 | 20.30% |
Source: supplied CEP screenshots; calculations checked by Haiti Politic. “Configured-capacity utilization” is the dashboard’s measure, not a population coverage or turnout rate.
Artibonite has the lowest displayed utilization, compared with 22.92% in Sud-Est. This is consistent with a serious access problem, but it does not isolate the cause. Staffing, rollout dates, displacement, transport and the construction of the capacity denominator could affect the comparison.
The OUEST 1 label also needs clarification before it is treated as equivalent to the candidate dashboard’s OUEST category.
There are reassuring arithmetic checks: the departmental registrations sum exactly to 881,523. The age categories and female/male counts independently reach that same total. This establishes internal consistency, not the accuracy or completeness of individual records.
The 28.84 million question
The political-membership module displays:
| Category | Count |
|---|---|
| Recorded memberships | 1,351,535 |
| Rejected for invalid NINU identification number | 6,194,185 |
| Rejected for non-matching information | 16,904,051 |
| Rejected as duplicates | 5,742,633 |
The rejection counters sum to approximately 21.3 times the recorded membership total.
No defensible fraud percentage follows from that ratio. The screenshots do not establish whether the rejection categories overlap, whether repeated uploads are counted repeatedly, whether one entry generates multiple errors or whether all counters cover the same reporting period.
A duplicate could result from repeated submission, competing membership claims or deliberate double-counting. A mismatch could reflect a typing error or identity misuse. These are different problems requiring different remedies.
Rejections also indicate that validation controls are stopping entries. What remains unclear is whether those controls accurately distinguish legitimate applicants from improper submissions—and whether accepted memberships reflect informed consent.
There is nevertheless a documented reason to investigate manipulation. Reporting by Le Nouvelliste, reproduced by Karibinfo, attributed to ONI director Reynold Guerrier an allegation that political structures had attempted to fabricate identification numbers. The account also described complaints about people appearing under other political banners. ONI defended its data-protection measures. These are attributed allegations and institutional positions, not findings of guilt against every organization with rejected entries.
An independent audit should separate accidental errors, technical retries, non-consensual identity use and deliberate fabrication. Until then, both blanket accusations and blanket reassurances are premature.
The membership requirement is also a political filter
Of 105 political structures displayed—88 parties and 17 groupings—37 are shown as reaching 30,000 members.
| Structure type | Total | Reached displayed threshold | Share |
|---|---|---|---|
| Individual parties | 88 | 24 | 27.3% |
| Political groupings | 17 | 13 | 76.5% |
| Combined | 105 | 37 | 35.2% |
Groupings are substantially more likely to have crossed the displayed threshold. That is consistent with pooling organizational resources, but does not prove greater popularity, fraud or preferential treatment.
The remaining 68 structures are not shown as meeting the threshold in this snapshot. That is not a final disqualification decision.
The deeper question is whether the requirement measures genuine, voluntary organization. A membership threshold can simplify ballot administration while also rewarding money, transport networks and data-processing capacity. Where consent is difficult to verify, administrative resources can overshadow authentic political support.
The independent-presidential module presents another access problem: two prospective candidates, 235 recorded supporters and none reaching the displayed 150,000-supporter benchmark. Its rejection counters sum to 90,987, again without establishing a count of distinct people.
These figures establish a large administrative gap. They do not measure eventual electoral popularity.
Women register, but the path to national office narrows
Women account for 50.9% of recorded voters and approximately 56.0% of recorded organizational memberships. Yet they represent 29.1% of candidate registrations overall.
The breakdown by office is more revealing:
| Office category | Displayed female share |
|---|---|
| President | 0.0% |
| Senator | 13.4% |
| Deputy | 6.9% |
| Mayor | 35.6% |
| CASEC | 33.8% |
| ASEC | 39.4% |
Stronger representation in local categories lifts the overall average while concealing much lower representation in legislative and presidential registrations.
The charts do not explain individual decisions. They do justify examining party nominations, financing, harassment and access to organizational support. A national average also cannot, by itself, determine compliance with rules applying to particular lists or offices.
Youth representation raises another question. Ages 18–30 account for 7.2% of candidate entries; ages 31–60 account for 84.3%. Eligibility ages differ by office, so this is not a direct measure of discrimination. Nevertheless, promises to empower young people should be tested against actual recruitment and financing opportunities.
Some apparent discrepancies are correct
The candidate registrations total 6,657. The departmental table totals 6,639.
The difference is exactly 18 presidential registrations, which the departmental table expressly excludes because the presidency is national.
This reconciliation matters. A correct exclusion should not become a sensational allegation of missing candidates.
The registration-activity chart requires similar restraint. Its late-series values fall sharply below the August peak, but the final day may be incomplete. Without a timestamped daily export, the endpoint cannot establish a lasting collapse—or prove what caused it.
How insecurity turns into electoral exclusion
IOM reported nearly 1.5 million internally displaced people in September 2026, with displacement spreading beyond the capital.
Displacement creates electoral barriers even when registration software works. Citizens may have lost documents, moved away from their assigned locality or face danger and expense reaching a center. Replacement identification, address updates and access to the correct voter list become essential.
These are mechanisms of exclusion, not individual conditions measured by the supplied charts. Quantifying them requires anonymized information on registration transfers, document delays and displaced residents’ access.
The identification bottleneck has prompted an institutional response: the OAS and Japan announced a $3 million agreement supporting identification services, including procurement of 350,000 blank biometric cards and mobile units. An announced procurement is not proof that cards have reached voters.
An impressive center count means little to a citizen who cannot safely reach a center or obtain the documents needed to participate.
Corruption and insecurity reinforce each other
The evidence for political capture comes from sources beyond the registration dashboards.
The UN Panel of Experts’ report S/2026/714 describes political, economic and security actors sustaining gangs. It raises concerns about collusion between local authorities and armed groups to consolidate control and influence elections, documents weapons diversion involving corrupt officers and criticizes slow sanctions implementation. These are the Panel’s findings and allegations, not criminal convictions against every person discussed.
The electoral implications are substantial. Where access to a community depends on armed permission, campaigning can become dependent on relationships ordinary competitors do not possess. Where officials facilitate criminal networks, equal enforcement and voter protection become harder.
Corruption and insecurity can therefore form a reinforcing cycle: compromised institutions allow coercive power to endure; coercion restricts scrutiny and competition; weakened accountability makes institutional capture harder to reverse.
That is an explanation of the risk, not a statistical estimate of affected registrations.
Nothing in the supplied charts establishes theft of CEP funds, bribery by a named electoral employee or manipulation of a particular candidate’s file. Those claims require financial records, access logs, documented decisions and independent investigation.
A moving calendar is not an election
The CEP’s October 5 notice extended candidate registration and document submission to October 31, postponed the campaign period and said voter registration would continue. Its October 20 publication described consultations on a revised calendar.
Those notices establish a timetable being reworked while preparation continues. They do not establish insecurity as the sole cause of every delay.
Repeated changes nevertheless impose costs: applicants adapt to new requirements, administrators implement new instructions and voters lose certainty about how registration will lead to a ballot.
The strongest diagnosis is a stalled democratic transition with continuing administrative activity.
To make its dashboards meaningful, the CEP should publish definitions for every counter, distinguish people from submission attempts, explain overlapping rejection categories, provide correction and appeal procedures, report actual center availability, document displaced voters’ access and disclose spending against electoral deliverables.
Transparency should protect citizens’ identities and political affiliations. Aggregate reporting and independent audits can improve accountability without exposing personal identification numbers.
Haiti’s electoral crisis will not be resolved by announcing larger totals. Citizens need assurance that their identities are protected, candidacies are assessed fairly, communities can participate safely and the timetable leads to a vote that counts.


