Monday, 9 November 2009

2009 Elections Mozambique political process bulletinNumber 34

9 November 2009
Editor: Joseph Hanlon (
Deputy editor: Adriano Nuvunga Research assistant: Tânia Frechauth
Published by CIP, Centro de Integridade Pública and AWEPA, the European Parliamentarians for Africa
Material may be freely reprinted. Please cite the Bulletin.
To subscribe in English: Para assinar em Português:
Misconduct in hundreds of polling stations
Evidence is growing of misconduct in polling stations – both ballot box stuffing and improperly invalidating votes for the opposition. How widespread was the problem?
We can use the sample count conducted by the Electoral Observatory (Observatório Eleitoral) and EISA (Electoral Institute of Southern Africa) to make an estimate. This was a formal Parallel Vote Tabulation (PVT, Amostra Aleatória dos Apuramentos, AAA) based on a random sample of 8 percent of polling stations. An observer was stationed in each of the sample polling stations all day, and collected the final result at the end of the count in that polling station. We have analysed results from 967 polling stations across the country.
We believe there are indications of possible misconduct and fraud in perhaps 6% of polling stations – that is 750 polling stations across the country, which is a very large number. This is large enough to change the outcome in some places. We noted Friday that one parliament (Assembleia da República) seat may have been stolen from Renamo in Tete. In Angoche, where both MDM and Renamo are standing for provincial assembly, the misconduct might be large enough to affect the distribution of seats. 0510152025303540AfluênciaTurnout10%20%30%40%50%60%70%81%91%101%Number of polling stationsNúmero de assembleias de voto
This kind of analysis is not proof. It can only give an indication of where misconduct may have occurred. We have looked for very high turnouts, which could show ballot box stuffing, and very high levels of invalid votes, which could show where an extra mark has been added to ballot papers to make them invalid.
The first chart, on the previous page, shows the spread of turnout, grouped into 1% bands, with the lowest turnout on the left and highest on the right. Thus the three tallest lines in the chart show that there were 39 polling stations in the sample with a 39% turnout, 34 polling stations with a 40% turnout, and 33 with a 45% turnout. And this is what we would expect.
But two things look odd. First, at the left of the chart, there are a small group of polling stations with a turnout of less than 15%. That suggests something was wrong – the polling station opened late, or was hard to find, or there was a problem with the register book. This affected 12 polling stations in the sample, which is 1% of polling stations.
And at the right of the chart there are a group of polling stations with a turnout of over 95%. We find that highly suspicious, because normally some people on the register will have died or be too ill to vote or be away from the area. And turnouts of over 100% are even more suspicious. There were 21 polling stations in the sample with a turnout of 100% or more, and 15 with between 95% and 99%. We think there is a good chance that in these 36 polling stations (3% of the total), there was ballot box stuffing. A closer looks shows that of the 36, 21 were in Tete and 6 were in Gaza, areas where there has historically been ballot box stuffing. PVT Polling stations ranked by % of nulosAssembleias de voto da AAA ordenadas por % de votos nulos0%5%10%15%20%25%30%35%40%Polling stations ranked by nulosAssembleias de voto ordenadas por votos nulos% of nulos -
de votos nulos The second chart looks at invalid votes (nulos). Here we have ranked the 967 polling stations with those on the left having the lowest percentage of nulos and those on the right having the highest percentage. For the most part, the chart shows what one might expect – a range of nulos starting from very few and rising slowly to about 8%. But then it suddenly jumps. What the high values on the right of the chart show is that there are a few polling stations with very high numbers of invalid votes. It seems to us highly unlikely that such a large number of voters did not know how to vote or spoiled their ballot paper. There are 6 polling stations in the sample with over 30% nulos, 13 between 20% and 30%, and 12 between 15% and 20%. We consider the 29 polling stations (3% of the total) to be highly suspect. There are another 30 polling stations with between 10% and 15% nulos, which are also suspect.
It should be noted that 2 of the polling stations with more than 20% nulos are polling stations in Tete which are already suspect for have very high turnouts. Four others in Tete also have high numbers of nulos.
Of the 29 polling stations with more than 15% nulos, 6 are in Angoche, Nampula province, where there were complaints about putting extra ink marks on ballot papers for Renamo in past elections. The other big group is 7 in Sofala.
What is proof?
We must reiterate that this statistical test is not proof. The proof on falsely invalidated ballot papers could be seen during the requalification of the nulos, where we saw hundreds of ballot papers for Renamo and for Afonso Dhlakama with a highly suspicious extra ink mark. The proof on ballot box stuffing is that polling station staff have been seen ticking names off the register list after the polling station closed, and seen to be putting extra ballot papers in the ballot box.
Of course, this should have been caught by Renamo and MDM party delegates in the polling stations. In a few places, party delegates were improperly excluded. But all too often party delegates were poorly trained and not paying enough attention to the count.
Statistical techniques such as the ones we use here cannot identify individual polling stations with fraud. There will be a few polling stations that really do have a very high turnout, for example of young people who registered to vote for the first time this year. Similarly, we also saw many hundreds of ballot papers which were genuinely invalid, where the same hand had marked multiple boxes.
But we can be sure that an unbelievably high number of polling stations had 100% turnout, and that it is not credible that in so many polling stations there were so many people marking their ballot papers with both a cross from one candidate and a fingerprint for another.
Despite stern warnings in both the polling station manual and in the polling station staff code of conduct, literally thousands of polling station staff participated in, or acquiesced to, improper activity in polling stations. And Armando Guebuza and Frelimo won such an overwhelming victory that they did not need the illegal help that has again tainted a Mozambican election. Joseph Hanlon
LSE study showed large scale fraud in 2004
A study by the London School of Economics Crisis States Research Centre of the 2004 Mozambican election reported that: "The evidence presented favours the conclusion that ballot box stuffing, improper ballot nullification and organisational failure did indeed take place on a large scale. Furthermore, we conclude that fraud was most prevalent in those provinces where Renamo complained about misconduct." The study looked at ballot box stuffing, nulification of Renamo votes, and places where people could not vote. That study is posted on our website:
MDM says delegates expelled
or refused credentials
The MDM at a meeting in Beira yesterday collected together reports of improper actions it faced on election day, 28 October. In particular, MDM says district election commissions refused to issue credential to MDM delegates in Chinde, Zambézia, and Changara, Tete
MDM says its delegates were barred from polling stations in Tsangano, Tete and expelled from polling stations in Angonia, Tete, and at Emopesca, Angoche, Nampula.
In many place in Zambezia, MDM says its delegates were not permitted to be present for the opening of the polling stations, and in Gurué, Maganja da Costa, Gilé and Pebane districts and Quelimane were not allowed to be present for the count.
Other incidents reported by MDM include:

In Angoche, Nampula, people claiming to be national observers were involved in ballot box stuffing.

Polling station staff in Angonia, Tete, telling electors to vote for Armando Guebuza.

Early closing of some polling stations in Manica; campaigning close to some polling stations in Manica.

Priority in queues given to known Frelimo voters in Maringué, Sofala.

In Savane, Dondo, Sofala, polling station 0634 was moved to more than 7 km away fro the registration location.

In various polling stations in Gaza, Maputo province, and Maputo city, results sheets (editais) were not posted at polling stations and not given to party delegates.
MDM cites specific crimes
The MDM also cites a series of specific criminal actions involving identified people, which could be prosecuted:

At polling station 2108, Escola Primária Completa (EPC) de Mussolo Novo, Nicoadala, the results sheet (edital) was changed after it was written, changing 810 votes for Daviz Simango to only 15. The president of the polling station refused to accept the protest, which has now been submitted to the district election commission.

At polling station 623 Escola Primaria Samora Machel in Mafambisse, Sofala, the MDM says it caught the polling station president, Sra Eufrásia, and a teacher named Conde putting extra ballot papers in the box. The police were called to respond to the resulting scuffle.

In EP1-Monzue, Savane, Dondo, Sofala, the Frelimo party secretary was permitted to vote for two people he said were ill.

At polling station 0056 at EPC Esturro, Beira Escola Primaria do Esturro, Beira, a polling station staff member only identified as Teresinha was caught putting an extra ink mark on 124 ballot papers for Daviz Simango, to invalidate them:
0056 at EPC Esturro, Beira Escola Primaria
do Esturro, Beira

MDM also says there was the improper adding of an extra mark to make votes for MDM invalid, in Tsangano, Tete.

Refusal by polling station presidents to accept protests in Lichinga, Niassa and Pebane, Zambézia.
MDM cites register problems
The MDM cites a series of problems with register books. These include:

Incorrect register books in polling stations in Ilha de Moçambique, and the refusal of polling station staff to use the handwritten register books until 1600.

Register books with more than 1000 people in Cabo Delgado.

In Lichinga, people whose names were not in the register were allowed to vote and their names were listed. MDM also claims there were 19 false register books, and there are discrepancies between numbers of registered voters on various lists.

Incorrect register books in Tsangano, Tete.

Books missing in Machanga, Ceramica, and Mafambisse, Sofala.

Repeated names in some register books in Manica.
Mozambique Political Process Bulletin
Editor: Joseph Hanlon (
Deputy editor: Adriano Nuvunga -- Research assistant: Tânia Frechauth
Material may be freely reprinted and circulated. Please cite the Bulletin.
Published by CIP, Centro de Integridade Pública and AWEPA, the European Parliamentarians for Africa
To subscribe: Para assinar:
In English: Em Português:
To unsubscribe:
Also on the web: Também na internet:
In English: News on the elections:
Previous issues of the Bulletin:
Em Português: Noticias sobre as Eleições:
Boletims anteriores:

No comments: