After the country's first elections in 1994 had been applauded by the international community, the local election four years later - marked by bitterness between the two major parties and boycotted by most of the opposition with a resulting voter turnout of only 15 percent - seriously questioned the consolidation of the young democracy. This gave the second general elections in December 1999 - the first under the sole responsibility of Mozambican authorities - even more significance as the concluding test for the sustainability of the new democratic system.
Realising the growing importance of the upcoming elections, the major parties struggled through to a substantial programme designed to enhance overall confidence in the electoral process. The electoral law was changed to once again incorporate all parties represented in the national assembly in the electoral organs at national, provincial and district level. The parties also agreed to conduct a completely new voter registration exercise in order to eliminate doubts about the trustworthiness of the existing registry. The registration process, monitored by national and international observers, was considered a huge success with 86% of the estimated eligible voters registered.
Election Days
Presidential and parliamentary elections were scheduled for December 3 and 4 1999, but extended for a third day after heavy rainfall delayed the delivery of materials to several dozen of the more than 8,000 polling stations in the country. A few stations in the rural areas of the most afflicted regions in Zambezia province never opened and thus a total of approximately 8,000 registered citizens were not given a chance to cast their vote.
In general, there was a polling station for every voters list with a maximum of a thousand registered voters. In some cases, voters lists were split in two and separate polling stations established to shorten the way for voters in rural areas. The rule that citizens should vote at the same location where they had registered could therefore not be observed in all cases.
The stations were staffed with five polling officers. Additional staff members (called auxiliaries) organised the queues and answered basic questions outside of the polling stations. The personnel was due to arrive at the station at 5:00 to set up the voting material before the station opened to the public at 7:00. Accredited party representatives, national and international observers as well as the media were allowed to be present during the entire day (7:00 - 18:00) as well as during the opening and closing procedures.
Upon arrival, voters presented their voting card to a polling official who verified that their names were on the voters list. The presiding officer handed out ballot papers for both presidential and parliamentary elections and demonstrated how to fold the papers to avoid unintended ink marks. The ballot papers were different in size, but had an otherwise similar design. The parliamentary ballot paper listed the candidate´s party name, abbreviation and symbol with a separate line for each party or candidate. The presidential ballot paper listed the candidate's name, party symbol, abbreviation and a photograph. The last column on both papers was left blank allowing for voters to mark their choice either with a cross or a thumbprint. Two mobile voting booths that were a part of the material kit for polling stations, provided for the secrecy of the vote. After marking and folding the ballots, voters returned to put them in the respective ballot boxes in front of the presiding officer. Finally, voters were made to dip their index finger in indelible ink to establish that they had already voted. Their names were checked off and initialled in the appropriate column of the voters list before the presiding officer returned their voter cards.
The Count at the Polling Station
The counting process at the polling station level began immediately after the stations were closed on the last day of voting in the presence of party representatives and observers. As a first step, the total number of voters checked off in the voters list was counted after all empty fields representing non-voters had been crossed out. Unused and spoiled ballots were counted, the result noted in the polling station report and the ballots packed in tamper evident plastic bags. The ballot box for the presidential elections was then opened to count the total number of ballots. The ballots were not unfolded, but repacked in the box afterwards. The process was repeated with the ballots for the parliamentary elections, although in this case, the ballots were not repacked, because the officials subsequently carried out the actual counting. The total number of voters as well as the number of ballots in each of the two boxes was noted in the polling station report and on an additional document that was immediately posted outside the polling station for public information.
The counts for presidential and parliamentary elections were conducted separately, but the procedures were similar. The presiding officer took one ballot at a time, unfolded it and indicated aloud the classification given to the vote (party or candidate name, invalid or blank). The deputy officer repeated the classification and showed the ballot to the party representatives and observers present. Ballot papers were arranged in separate stacks representing each party or candidate, invalid and blank votes. A third polling official kept a running tally of the classified votes on a notepad or blackboard. When all ballot papers had been counted, each stack was counted separately and the results compared with the tally. The party representatives were then allowed to examine the stacks and contest the decision of the presiding officer. If a decision was challenged, the complaint was noted on the back of the respective ballot and signed by both the presiding officer and the contestant. The challenge had no immediate impact on the counting results, but the ballot paper was later sent to the National Electoral Commission for a final decision.
A separate document (or edict) was filled out for each election, showing the number of registered voters, the number of ballots in the box, the number of ballots per candidate or party and the number of blank and invalid votes. The edicts were signed by the polling officials and posted outside of the polling station for public information. The same information along with the number of contested votes, and all other irregularities or problems was noted in the polling station record.
Valid and blank votes were sealed in the same tamper evident plastic bag, while invalid and contested ballots were sealed in another. The polling station record, the voters list and carbon copies of the edicts for both elections were packed and sealed in separate bags. All plastic bags were then stowed with the other material in the metal electoral kit boxes and sent to the district office of the Technical Secretariat for Electoral Administration (STAE). While valid and blank ballots were kept at the district level, all other materials and documents were forwarded to the provincial STAE headquarters. STAE personnel or police escorts accompanied the transports. In most cases, party representatives also travelled with the materials.
Tabulation at the Provincial Level
Upon arrival at the provincial STAE headquarters, the materials were separated. General supplies were stored in warehouses. The voters list, a copy of the polling station documents (report and edicts) and the plastic bag with invalid and contested votes were sent to the National Electoral Commission (CNE) in Maputo, the capital, while a second copy of the polling station documents was kept as a basis for the tabulation of the provincial results. These results, which were scheduled to be published within seven days after the elections, were only preliminary, as the final adjustments made necessary by the re-evaluation of invalid and contested ballots were made at the national level.
A special software program based on Microsoft Access had been designed for the elections. All information from the ballot account forms was entered into this database in the respective provincial capital. The data was entered independently by two teams to detect errors in arithmetic and data entry or attempts of fraud. The computer program then compared the two data sets and reported any discrepancy to the administrator whose responsibility it was to examine and resolve the problem. In the case of a simple typing mistake, the edict was again entered by both teams. If the edict itself was faulty (e.g. the sum of all votes did not match the number of ballots in the box), it was given to the Provincial Electoral Commission (CPE) for follow-up. There were no explicit rules or regulations as to how and when the CPE should reconcile the discrepancies, so that in some provinces the commissions waited until all other work was done before they met, thus delaying the re-entry of the faulty edicts and with it the whole tabulation process.
Party representatives and national and international observers were given access to the results database, allowing them to see the data that had already been entered and compare it with the figures they had gathered during the election. In case of doubt, observers could ask for the original edict to verify that the numbers on the document matched those in the database. Unfortunately, there was no way to see how many polling stations had yet to be entered.
Due to technical problems with the software, access to the observer computer was restricted to short periods of 15 to 30 minutes between the shifts of the data entry personnel. In most provinces, CPE and STAE personnel seemed to be rather reluctant to co-operate with the observers, often interpreting the regulations rigidly or, in a few cases, even denying access to either the observation terminal or the original edicts. In most cases, these problems originated from insufficient training regarding the role and rights of observers and were resolved after a visit of a deputy director of National STAE, who was able to convince the personnel of the validity of the observer regulations.
As a consequence of its late completion (only days before the elections were held), the personnel had very little experience with the new software, and observers had not been trained at all on how to use it. This was especially problematic for many of the national or party observers who lacked any computer experience whatsoever. Thus, most observers had difficulties accessing the available data and no clear idea of what they should do with it, consequently relying on the advice of other (often international) observers on how to compare their own information with the computer data. A lack of understanding of the procedures led to confusion on part of the observers, who interpreted technical problems as attempts of fraud. In some cases, observers did not ask for clarification of their doubts, but simply reported their sometimes faulty allegations to their party headquarters, adding to an already tense political situation.
Despite all the problems, the observation system successfully detected two attempts of fraud, where data entry personnel had exchanged the number of votes for the two presidential candidates whenever the incumbent had won the majority at a polling station, thus favouring the opposition candidate. With the help of computer audit logs the culprits were found and the results were corrected before serious harm was done. In almost all provinces, technical problems, lack of training with the computer software and a high number of edicts with wrong or insufficient data delayed the publication of the preliminary results. A considerable number of edicts showed mistakes so grave that they could not be processed at all or the CPEs felt unable to reconcile them, in the latter case passing on the responsibility to the CNE. As a result, hundreds of edicts representing several hundred thousand votes were not reflected in the preliminary results, adding doubts about accuracy and transparency of the counting process to the delay.
(Re-)Evaluation of Invalid Ballots at the National Level
While the tabulation process was ongoing in the provinces, edicts and ballots started to arrive in Maputo. STAE and CNE members began with the reclassification of invalid and contested votes. After the arrival of the materials had been registered, the sealed transport bags were organised by province. The bags were then opened and every ballot was reviewed by two CNE members appointed by different parties. The regulations allowed a liberal interpretation of valid ballots only asking for the voter's will to be clearly marked. In other words, a cross or thumbprint (or both) could be anywhere in the line with the party or candidate information. Even though polling and counting officials had been provided with a comprehensive set of examples for invalid and blank votes, the CNE was overwhelmed by the large number of ballots they had to examine. A document was filled out for each polling station, specifying the number of invalid votes that had been reclassified as valid and to which party or candidate they had been assigned. The results were later summarised for the whole province on a separate form.
Another team of CNE members worked on the problematic edicts that had been forwarded by the CPEs. Whenever the problems were sorted out, the data was entered into the database with the provincial results that by then had arrived in the capital. Finally, the summarised results of the reclassification of invalid and contested votes were added to the databases with the provincial results, forming the base for the distribution of parliamentary seats. To determine the outcome of the presidential race, the figures were summed up for nation-wide results. The official results were publicly announced by the CNE with a delay of two days that stemmed partly from the delays caused during the tabulation process in the provinces, partly from the unexpectedly high number of invalid or contested votes and partly due to problematic edicts which had to be reviewed by the CNE.
Lessons Learned
Although some of the problems that occurred during the 1999 Mozambican elections were country specific, their analysis allows drawing some conclusions worth generalising:
- transparency is of paramount importance during the entire electoral process, especially during all stages of vote counting and tabulation,
- rules and procedures must be clear and precise and known in advance to all people concerned with the elections,
- all electoral officials must be sufficiently trained in these rules and procedures and in correctly filling out the official forms and records,
- when computers are used, system and software must be tested well before the elections to allow for modifications and improvements; additionally both electoral officials and observers must be trained how to use the software,
- polling and counting officials need to comprehend the task of the observers and understand that they are not working against them,
- to avoid mistakes during the count, working conditions must be appropriate (e.g. enough light) and officials should not be exhausted, as often was the case after three full days of voting.
