The Haitian government’s CIEVE commission’s report does not find fraud. There is no example of:

  • Difference between polling return and posted result (as occurred on November 28, 2010)
    • Difference between manual count and election return or posted result
    • Disproportionate vote in a polling place for a candidate triggering further review by the Tabulation Center (2009 and 2010)
    • More votes than voters (2009)
    • Obvious tampering on polling returns (December 2006)
    • Conversion of blank votes into votes for candidates (February 2006)
    • Discarding of runner-ups’ votes (May 2000 senatorial)

Instead of finding fraud, the Verification Commission shows only that the possibility of it was opened up by the lack of two required reports, a set of purported discrepancies in election materials, mistakes in the copying of voter ID numbers, and the lack of individual registration of election monitors. Nowhere in its report does the commission name an instance of fraud or claim that it has found one. Its report is completely hypothetical. It rests its case on reporting and technical deficiencies. (Rapport de la Commission Indépendante d’Evaluation et de Vérification Electorale (CIEVE): Elections de 2015 [Port-au-Prince, le 29 Mai 2016])

Having failed to advance a single example of fraud, much less does the Verification Commission provide evidence that enough fraud occurred to affect the outcome. All it says is that the lack of required reports undermined the validity of enough votes to potentially, theoretically open the door to fraud.

Yet the Verification Commission also charges that the lapses it has discovered were “systematic (well organized) and had the intention of fraud.” Again it provides no evidence of such actual intent by any member of the electoral apparatus and in the body of its report more frequently attributes the lapses to ignorance and poor training.

A more fruitful field for the commission’s inquiry would have been the gate-keeping stage that preceded the actual election. There, qualified candidates were excluded for thinly-disguised political reasons and criminals were allowed to run. This is not a concern of the Verification Commission.

This concludes the Haiti Democracy Project’s main evaluation of the Verification Commission’s report, for even if all its accusations of reporting lapses were true, it has not proven fraud, still less that it rose to a level that would change the results and warrant overturning an election.

In the annex that follows, we will review its main accusations against the previous electoral commission’s reporting and assembling of election materials. In general, we find that other reports provided the information that the Verification Commission wants. As for discrepancies among election materials, a compilation mistake leads the commission to exaggerate them by five times.

Technical Annex

Voting by Election Monitors

The first issue raised by the Verification Commission is the voting by party poll watchers and electoral observers. Since their observation duties will take them away from their home polling places and they are obligated to stay in their assigned locations all day, the electoral law has long given them the right to vote there, even though their names don’t appear on the voters’ lists there, just at home.

The electoral law calls on the election workers to fill out a special form naming these voting poll watchers and observers, in addition to the regular polling return in which their votes are counted with all the others. But election workers bothered to do this in only 4 percent of the cases. Because of the absence of this form, the Verification Commission calls these votes “untraceable,” as opposed to the votes of the local voters whose names and IDs appear on the signature sheets. Given the high number of candidates, each of whom is entitled to send out poll watchers, the Verification Commission finds that this resulted in 29 percent more votes per polling place than are recorded on the local voters’ signature sheet. This the Verification Commission rather dramatically calls, “Le maillon responsible de la cassure de la chaine de surveillance du processus électoral.” “It means that the polling places allowed 448,000 citizens to vote without filling in the form for them stipulated in the electoral law . . . a massive absence” of the required form. In so doing the electoral workers showed “incompetence, negligence, unprofessionalism, and potentially fraud.” This absence of the special form is later cited as part of the proof that the presidential elections were laced with irregularities and fraud. (page 17).

Nevertheless, the Verification Commission claims no specific instances of fraud emanating from the poll watchers’ and observers’ voting without their names being recorded on the form. It does not, for example, claim that unauthorized people voted or that the poll watchers voted more than once.


Ideally, it would have been better if the electoral workers had filled out the form with the names and IDs of the visiting voters. This way one could better tell at the Tabulation Center whether the extra 29 percent of votes from the polling places were all those of election monitors, and that they had only voted once in that polling place. The absence of the form, however, does not mean that interlopers got to vote or that monitors voted more than once. The controls exercised by the election workers in the polling places are long-established and were sufficient to prevent either of these things from happening on a significant scale. Of 208 reports received from the accredited observers of the NOAH/Haiti Democracy mission, only six found that the poll watchers got to vote more than once in the October 2015 election. None reported that unauthorized people were allowed to vote. Thus even without the forms being filled out, we know that the extra 29 percent consisted of monitors who had a perfect right to vote and only voted once.

Thus, the Verification Commission’s use of the word “untraceable” to refer to that extra 29 percent not noted in the local voters’ list is excessive. The voting of that 29 percent was indeed “traced” by trained election workers on the scene and by the other election monitors present. Once those ballots went into the box, they became no more or less “traceable” than any other. This is a distinction without a difference.


There is no evidence that the handling of the election monitors’ vote changed the election outcome.

Cross-checking and Triangulation – A General Point

The question of the missing monitors’ form is consistent with the larger method of the Verification Commission in seeking to find all the means to verify and quantitatively cross-check the election post-facto in the files of the Tabulation Center. It wants the totals of the ballots sent out, the ballots sent back (used, unused, blank, or spoiled), the number of people listed as voting on the signature sheets or special forms, the votes on the election workers’ manual tallies, and the votes on the polling returns all to match. Otherwise it will declare the election fraudulent.


Even though the forms for the visiting voters’ names were missing, meaning that the number of voting names could never match up to the number of votes, the Tabulation Center still received the other three measurements in good order:

  1. The cast ballots
  2. The manual tallies (from 96 percent of polling places)
  3. The polling returns (from 100 percent) (Page 12-13).

Thus the Tabulation Center still had redundant means to cross-check the numbers of votes.

Nonregistration of Poll Watchers and Observers

In addition to the missing form, the Verification Commission says the electoral commission was derelict in handing out blank accreditation cards for the candidates and observer groups to fill in and distribute among their poll watchers and observers. “The election commission didn’t keep a register of these accreditation cards for verification or audit. These accreditation cards could be transferred among the political parties without any control or authorization of the election commission.”


It was logistically impossible for the electoral commission to individually register the large numbers of poll watchers and observers whom candidates and civic organizations wanted to send out. As customary, it left it up to the inscribed candidates and groups to decide who would represent them and how to keep track of them. The electoral law gives it this discretion. The election commission was already overwhelmed by the challenge of administering its own apparatus of 13,600 polling places without taking on new domains to administer. Indeed, it could barely issue even blank badges for the monitors. From the Haiti Democracy Project’s own experience, we recall that we and many other organizations didn’t receive our cards until the late afternoon of November 27, 2010 for the election that took place the next day. Similarly for the August 9, 2015 election we only received them at noon the day before. As for individually-registered badges, it took the observation unit five days to produce thirty-six of them because of difficulties with the equipment and transcription. The Verification Commission’s demand is impractical.

In place of registration, though, the election commission came with tighter regulation of monitors of the October 25, 2015 election and so effectively managed the issue raised by the Verification Commission. Strict orders went out to limit the number of poll watchers allowed in the room at one time and if necessary admit them in shifts. The longstanding rule to clip the corner of the monitors’ badges to signify that they had voted was enforced. The number of eligible observation organizations was severely slashed to the point of weakening observation.


There is no evidence that the non-registration of the individual election monitors by the electoral commission had any effect on the results of the October 25, 2015 election.

Failure to Include Voter List in Packet

Moving on from the issue of poll watchers and observers, the Verification Commission makes a considerable point of the failure of most polling places to include the voters’ list in the packet sent to the Tabulation Center. Under the electoral law, its absence could be grounds for disqualifying the polling return (Article 171.1(h). This list is distinct from the voters’ signature sheet described earlier – it is just a list of the eligible voters to be tacked up on the door of the polling place before the election so that people will know where to vote. Thus of 3,210 polling returns received, only 304 voting lists were included in the envelope. Says the Verification Commission, “The absence of practically all the voting lists (90%) is inexplicable and can be linked to many factors such as the lack of education or the negligence of the people at the polling-place level.”

But as the Verification Commission admits (page 11), the same information was included on the 2,822 signature sheets of the sample which did arrive safely in the envelope. Knowing this, the Verification Commission momentarily equivocates: “Although the absence of the voting lists can on the one hand be considered unimportant . . .” It quickly recovers: “The number of voting lists not found (2,454) is high and shows that the polling places do not pay attention to quality control.”

The Verification Commission is unwilling to let go of the voting lists because this high number enables it to declare that only 294 of 3,210 files examined were complete. Without this exclusion, it would have to admit that 2,748 were complete. “High number of incomplete files” will duly take its place in the list recited to prove fraud.


The absence of the voters’ list had no bearing on the election’s outcome.

Methodological Error Invalidates Finding of Discrepancies

“Analysis of the cast ballots versus the total of votes reported in the polling return by the election workers shows the inadequate level of quantitative analysis of the personnel trained by the electoral commission,” the Verification Commission charges. The polling places reported receiving 1,350,933 ballots. “However, after subtracting spoiled ballots (9,276), unused ballots (1,008,245), and void ballots (28,840), the number of ballots that are valid should be 304,572. But the number of ballots reported valid by the polling places is 369,668. The unjustified difference of 65,096 ballots isn’t accounted for anywhere.” The commission adds to this discrepancy another 19,198 votes said to represent the difference between valid ballots and valid votes on the returns, for a total discrepancy of 84,294.

This means, the commission continues, that the possibility of cross-checking important elements reported by the polling places is weak. “The numbers reported on the polling returns are not in agreement (difference of 84,000).” The commission presents the proof in Table 2 below:


Table 2: Reconciliation of the number of ballots and number of votes reported by the polling places
Nombre de bulletins reçus par les BV (2,818 PVs) 1,350,933
Nombre de bulletins gâtés (2,798 PVs) -9,276
Nombre de bulletins qui n’ont pas servi (3,016 PVs) -1,008,245
Nombre de bulletins nuls (2,973 PVs) -28,840
Sous Total 304,572
Variance non justifiée par les BV 65,096
Nombre de bulletins rapportés qui sont bons (3,027 PVs) 369,668
Variance entre bulletins rapportés et votes tabulés 19,198
Nombre de votes tabulés pour les 3,210 PV par la CIEVE 388,866
Nombre de votes tabulés pour les 3,084 feuilles de comptage par la CIEVE 369,902



A glance at the table shows, however, that the Verification Commission took data from differing numbers of polling places to make its comparison whereas this number must be held constant for the comparison to work. If you want to compare the number of ballots received minus those spoiled, unused, or null to the number of ballots reported valid by the polling places, then all these elements must come from the same polling places, otherwise you do not have a comparison.

If as an experiment we increase the number of polling places on the ballots-received line to equal the number of polling places on the valid-ballot line, then we find the discrepancy of 84,294 shrinking to 15,675.


Using a faulty methodology, the Verification Commission exaggerated the discrepancy between electoral elements by a factor of five. The remaining small discrepancy after correction of the methodology had no conceivable bearing on the outcome of the election.

Overall Conclusion

The Verification Commission has failed to make its case. It has not found one instance of fraud. It spends most of its time in the thick underbrush of issues far removed from the tabulation of the votes — lack of required reports for election monitors, lack of registration of monitors, lack of one type of  voters’ list, mistakes in copying ID numbers, an exaggeration of the extent of discrepancies in ballot inventories. The Verification Commission has found nothing that would change the results of the October 25, 2015 presidential election.



Ballot Bulletin
Election workers Membres des bureaux de vote (MBV)
Electoral commission Conseil Electoral Provisoire, CEP
Electoral law Décret électoral
ID number CIN, Carte d’Identification Nationale
Manual count Feuille de comptage
Null ballots Bulletins nuls
Party poll watcher Mandataire
Polling place Bureau de vote, BV
Polling return Procès-verbal de dépouillement, PV
Signature sheet Liste d’émargement
Special form naming voting election monitors Procès-verbal de carence
Spoiled ballots Bulletins gâtes
Tabulation Center Centre de Tabulation des Votes (CTV)
Unused ballots Bulletins qui n’ont pas servi
Verification Commission Commission Indépendante d’Evaluation et de Vérification Electorale
Valid ballots Bulletins qui sont bons
Voters’ list Liste électorale partielle (LEP)