Because counts are done in each polling station in the presence of parties, observers and media, and then a copy of the results sheet (edital) is posted at each polling station, it is relatively easy to monitor the count and check for fraud. Recent changes (noted below) improve the ability to monitor, and national media, parties and observers have become more skilled.
Nevertheless, there have been significant frauds and important secret changes to the results. These have never been enough to change the national outcome, but they have changed the result in two municipalities in 2013 and changed the number of parliaments seats in 2009. This special issue of the National Elections Bulletin sets out where fraud and unexplained changes have occurred in the past, and looks at ways to control this.
At the polling station
Ballot box stuffing is done in two ways
1) Most common is simply to report a very high turnout and simply write a false results sheet (edital). This sometimes also requires ticking off enough names in the register book. This can only be done when no one is watching - when opposition party polling stations staff (MMVs, membros das mesas do voto) or party delegates are not present or are not watching. This happens most often in places which are very pro-Frelimo where the opposition cannot find people to be in the polling station or where party agents and MMVs, who are usually local people, are too afraid to oppose and expose such practices, even if they are watching. Historically this has happened in Gaza, Tete and parts of Niassa.
2) Physically putting in extra ballot papers, particularly the introduction of unused ballot papers from that polling station, during the day or after voting closes. There have also been examples of ballot papers from other polling stations being introduced - this should be caught during the count because of numbering of ballot papers, but this does not always happen.
Invalidating ballot papers for the opposition. This was serious enough in Marromeu in 2013 to elect Frelimo as mayor and a majority in the assembly, when MDM actually won. This can be done in three ways:
1) Most common is to put a physical ink mark on opposition ballot papers, to make it look as if the person has voted for two people and thus the ballot paper is invalid (nulo). This can be done at various times but is usually done late at night, when opposition observers are tired, and it is dark (for example where there is no electricity), and someone will simply put an extra mark on a whole set of ballot papers.
2) Writing a wrong edital, as with ballot box stuffing.
All "nulos" are checked again at national level, which is why we have seen how ballots are invalidated. But this also means the third method used does not work:
3) Moving ballots stacked on the floor of the polling station from the pile for the opposition to the pile of nulos. (These are then accepted as valid at national level.)
At the district and province
The District Elections Commission (CDE) simply adds up the totals on all the polling station editais, and produces its own edital. But if no one is watching or if no one is doing their own calculation, it is easy to add "incorrectly". This happened in Gurué in 2013. It also probably happened in Marromeu in 2003 and in Beira and Nacala in 2008; in all three instances editais presented by the district election commissions were different from those of the respective PVTs (parallel vote tabulations).
By law the Provincial Elections Commission (CPE) should simply sum the district totals, but in fact they do a complete count, with a data input system. In past elections changes have been made to the totals. For example, in 1999 in Nampula, the results showed that 10% of voters only voted for president but not parliament, yet no one ever saw a voter putting a ballot paper in the presidential box but not in the parliament box.
Changed or damaged editais
The Constitutional Council found that in Gurué in 2013 two things happened with respect to original editais:
1) Official polling station editais were substituted by others, without official stamp or signature.
2) The provincial election commission ordered that changes be made to polling station editais.
And there are other examples of editais that have had ink poured over them, been lost, or been changed. This could happen in the polling station, at district or provincial level, or in transit.
At the National Elections Commission (CNE)
The election law says that the totals are to be based on district and city editais and the provincial results. But the CNE has always done its own count based on polling station editais. No formal comparison is made with 11 provincial results or with the district results, and the CNE's final result can be significantly different. In 2009, the CNE reported 104,000 fewer votes for president than the sum of the provincial totals, and for parliament transferred one seat in Sofala and one seat in Tete from Frelimo to Renamo, all without explanation.
In addition the CNE checks all invalid votes (nulos). Polling stations are normally too rigid in applying the rules about extra marks and marks outside boxes; the law says the vote is valid if the intent of the voter is clear. In 2009, 200,000 presidential votes (4.5% of all votes) were submitted to Maputo as invalid, but 24,000 of those were accepted as being valid and added to provincial totals. This can make a difference.
The law and changes in the law
Ballot papers are numbered and the law requires these to be checked during the count, to exclude any ballot paper which did not originally come from that polling station. This should prevent introduction of external ballot papers, but in 2013 was often not done because it was time consuming. Although the number of ballot papers is restricted to 10% more than the number of voters, in a polling station with average turnout this leave many unused ballot papers and thus does not stop the introduction of unused ballot papers from that polling station.
The law now requires that all ink and liquids be removed from the place where the ballots are going to be counted. This is an attempt is stop opposition ballot papers from being invalidated.
Posting editais outside the polling station allows a parallel count (PVT, parallel vote tabulation) and for Radio Moçambique to read out polling station results live the day after voting.
A copy of the edital goes to party monitors in polling stations and, now for the first time, to the three main party-nominated MMVs. This allows parties to do an independent tabulation. This was important for Gurué in 2013. But the gap here is that opposition parties often cannot find delegates or MMVs in exactly the places where Frelimo is very strong and finds it most easy to tamper with the results.
The CNE says district and provincial counts are open to parties, observers and media. But the law only refers to parties. Unclear at the moment if observers and press can attend.
Where editais are altered, damaged, or missing, provincial and national elections commissions can use the official copies given to parties. The Constitutional Council has emphasised the importance of this, so we expect it to be used more this year.
Recounts will be allowed. This has only ever been used once before, so there are no systems and no idea of the impact. But where turnout is given as 102%, it should be easy to demand a recount. Also, recounts are required for any polling station with an edital with mathematical errors – hundreds of editais have been thrown out in the past for that reason, but now recounts should be automatically done.
Comparisons and statistics
The main other weapon in the battle against fraud and error is to make better use of the data we have. The PVT does this, for example.
It is important to compare district results with provincial and national results, to see if changes were made.
The final step is statistical, using the PVT and also using data input at provincial and national level. There are two checks:
1) Are there a significant number of polling stations with a turnout close to 100%? A few are possible, but not many. This is an indication of ballot box stuffing.
2) Are there many polling stations with nulos over 10%? The average is around 4% and the distribution should be normal. If there are polling stations over 15%, or a significant number over 8%, then this is an indication of intentionally nullified ballot papers.
Finally, it will be important to do a sample comparison of polling station editais which we know to be correct, for example from observers or radio Mozambique, with editais from the same polling station being used at provincial and national levels. Differences could indicate fraud.
No election can be perfect, but we can reduce the potential for fraud.
Source: Mozambique political process bulletin Number NE-52 7 October 2014