`GECOM managed to print a final voters list of reasonably high accuracy which can stand up to international scrutiny and is comparable internationally. In essence, in a voters list of approximately 440,185, the errors discovered were fewer than 1,000 and were mostly due to polling station misallocation. Even if the problems had been 1% or 4,400, this would have had virtually no effect on the distribution of seats, due to the fact that the electoral system is highly proportional'

`The Audit team has not been able to find that any real electoral advantage or prejudice was suffered by a particular party, as a result of the errors and administrative problems that occurred in the electoral process. The problems that occurred, while they may have had more effect in Region Four, because of the density of the population in that area, impacted upon the country as a whole. The entire electorate was generally disadvantaged by the problems rather than a single political party. In this respect, no evidence was presented by any political party which demonstrated that a particular error or problem affected the result in any constituency, region, or even at (the) national level'

Investigators who scrutinized the systems for the March 19 polls have found that concerns raised by stakeholders did not prejudice or profit any particular party and following an array of tests it said the voters list was highly accurate.

"The Audit Team was unable to find that the issues and concerns raised by the political parties and other stakeholders in relation to the 2001 electoral process, prejudiced or advantaged any particular political party or parties. Consequently, the Audit Team is able to declare that the issues and concerns detailed in this report would not have affected the result of the election", the highly anticipated report from the Sweden-based International Institute for Democracy and Electoral Assistance (IDEA) said.

It declared that the problems identified affected all regions of the country though more densely populated areas such as Region Four suffered more purely because of the concentration of voters.

IDEA was commissioned to conduct the systems review and audit of the elections in the aftermath of the political protests that followed the March 19 elections. The PPP/Civic was declared the winner of the general elections but there were strenuous objections to the electoral process from the main opposition People's National Congress REFORM (PNC/R) and as a result the Elections Commission - comprising representatives of all the political parties - agreed to invite IDEA to conduct the review.

The PNC/R concerns included claims of disenfranchisement of its supporters and flaws in the final voters list that prevented its backers from casting ballots. In the days following the election, the swearing in of the new president was challenged in court and there were violent demonstrations by PNC/R supporters.

For the purposes of the review, the Commission's database was frozen following the March 19 polls to prevent the possibility of any tampering prior to IDEA's investigation.

IDEA's mandate was to review the compilation of the final list of electors including the registration of voters and the processes culminating in the Official List of Electors; assess the production and distribution of the national identification card; arrangements for the casting of votes including the siting of polling stations and operational planning; review the recruitment and training of elections staff. A host of recommendations were made by the auditors on each of these areas. No where in the report was any political party named in the discussion of complaints raised by stakeholders.

The auditors found that the one issue which could have impacted prejudicially on a political party, had it not been addressed, was the issuing of the national identity card where Region Four was affected the most. "This was not the case in view of the decision taken by GECOM to allow voters to use alternative identification" on polling day, the report said.

In its summary of findings, the 130-page report said the eight-man team of auditors was unable to find any evidence of purposeful manipulation or electoral fraud in relation to the issues that were the subject of the audit. "Extensive discussions were held with the political parties and stakeholders and numerous allegations were made. However, very little evidence was presented to the Audit Team to substantiate the allegations that were made. Notwithstanding this, the Audit Team spent considerable time conducting interviews and examining documents at GECOM (Guyana Elections Commission), as well as running numerous IT (Information Technology) and database "tests", to investigate these allegations".

From the multitude of tests done on the database, the IDEA report said that no evidence of corruption or manipulation of the database was found. While errors occurred, the auditors were unable to find evidence of conspiracy to manipulate systems or the election result.

"While it is possible that there may have been individual instances of abuse of the system, as is the case in most elections around the world, there is no evidence to suggest that these abuses would have had any effect whatsoever on the election result", the report asserted.

The report further found that GECOM and its staff were too accommodating of the demands and sensitivities of the political parties and cited the recruitment and appointment of staff as one case where this was reflected.

Voters list

It was found by the IDEA team that the plan for holding the elections was compressed into an unrealistic time frame. GECOM's system development plan was allocated seven months and the IDEA report said that based on international experience 18 months would have been a more feasible period.

The schedule for the elections had been compressed because of various delays in implementing the Herdmanston Accord terms which were agreed in January 1998 for the holding of elections by January 17, 2001. Following agreement, the elections were subsequently held on March 19.

The IDEA investigators asserted that many of the problems that arose in the electoral process were attributable to changed plans and the resultant increase in the workload required to assuage concerns by political parties. It listed numerous cases of these including the GECOM Field Test Exercise and block objections to the Preliminary Voters List.

There were errors in the name field, date of birth and other information fields, the team found. These were mostly due to data capture error and the absence of certain checks.

Most of the errors in the Revised Voters List (RVL) happened because of an error in the claims and objections application. "The error is directly attributable to the time constraints that applied during the development and testing of this application which was compressed into just one week", the report said. The sequence for the voters list was as follows: Preliminary Voters List (PVL) to RVL to Official List of Electors (OLE) to OLE plus addendum.

Despite the additional workload and time consuming decision making, the report said that GECOM achieved a highly accurate voters list that was internationally comparable and therefore GECOM and its Information Systems Department (ISD) and data entry staff must be credited for this.

"In essence, in a list of approximately 440,185 (voters), the errors discovered were fewer than 1,000 and were mostly due to polling station misallocation. Even if the problems had been 1% or 4,400, this would have had virtually no effect on the distribution of seats, due to the fact that the electoral system is highly proportional", the report posited.

It made a number of technical recommendations for the improvement of the list process and forwarded that a six-month system development plan should be crafted at GECOM to enable local government elections to take place any time after March 2002.

National ID card

The report found that 440,185 persons were photographed by the close of the exercise on January 4, 2001. "To that extent at least, the process and procedures (as they finally evolved) that were used to accomplish this task, were effective", the auditors declared.

They said that in relation to allegations about the photographic exercise statistics, a party representative was unable to say how any discrepancy compromised the process but said "if there was a difference of two, there was a problem".

As a result of this position, the IDEA team discussed a discrepancy in the figures tabulated from hand written returns from the field and that contained in a subsequent report. While the discrepancy was not significant, the auditors said the provision of an incorrect estimate before it was properly checked was unwise and damaged the credibility of GECOM.

Turning to the claim that many duplicate ID cards were issued and this compromised the election result, the IDEA report said that GECOM undertook a reconciliation exercise and this indicated that no more than 324 duplicate national ID cards were issued.

"There is no evidence to support claims that the erroneous issue of such a small quantity of duplicate ID cards could have a material effect on the result of the election. The possession of a duplicate ID card did not mean that a person could vote more than once, unless that person also appeared twice on the voter's list. This is unlikely as there were checks and balances in place, such as the use of indelible ink to prevent multiple voting", the report contended.

It acknowledged that delays in the distribution of ID cards gave rise to concerns and led to GECOM's decision of March 16, 2001 to allow other forms of ID. "Due to this demonstration of pragmatic policy flexibility, no electors were disenfranchised as a result of not having received their National ID card and therefore there was no impact on the result of the election".

A raft of suggestions were also made about how to make the entire ID card production and distribution process more secure.

Data driven forensic probe of complaints

The auditors say numerous discussions were held with GECOM and political parties during which complaints were ventilated in relation to the software and database employed to compose different versions of the voters list.

"It should be stated that, despite numerous requests, very little evidence was produced to corroborate a number of allegations made", the report disclosed.

Nevertheless, the allegations were listed in the report and then subjected to a battery of tests, the results of which were also provided. These tests homed in on the validity of the division/subdivision structure of the database and the correct storing of these in the address table among other areas. The results were acceptable.

In one case, there was an invalid combination of divisions and subdivisions in 6,380 records on the RVL. This was linked to a programming error in the correction and objections application but was later corrected in time for the OLE.

On allegations that data entry operators might have deliberately entered wrong information, the auditors assessed the error rate of each operator who had entered invalid combinations of divisions and subdivisions.

It was found that 64 operators had entered invalid combinations. The average error rate was 1.8%. Seventeen operators had an above average error rate while 11 registered a rate of more than 3% and three had an error rate of 7% or more.

"Although the programming error in the application can be explained as the case of the invalid combination of the divisions and subdivisions, the high error rate of the three operators cannot be dismissed without further analysis of the operators' skills levels and their understanding of the application. However, we could not obtain the information on the skills levels of the operators. Therefore, there is no conclusive evidence of the abuse of the system through this test", the report said.

In what it described as the most conclusive test of duplicate registration - people with the same surname and first name and birth date living at the same address, the auditors found that there were nine such duplications.

"We have established that there are 62 identified duplicate electors on the database. Our test to establish whether capturing errors have concealed other duplicated electors has established that where this may have occurred, the impact would not be material to the electoral process... it appears that the duplications are mostly as a result of elector negligence, and possibly to a very small extent there may be elector fraud", the report concluded.

In its tests of the PVL to RVL and RVL to OLE transactions, the auditors averred that these assays provide "conclusive evidence that the OLE was not manipulated in any material way".

Recommendations were made to improve systems security and to undertake checks of audit trails. It also found that GECOM's systems documentation is haphazard and incomplete and this puts it at the mercy of the systems programmer.

Operations and logistics

The auditors ruled that the existence of polling stations with extremely small voter complements is a serious flaw. One hundred and forty-five stations had 50 or less voters and the report said that in addition to creating unnecessary administrative and logistical problems it trespassed on international norms as the secrecy of the ballot could be compromised.

Apart from this, GECOM's procedures appeared to comply with international standards and the report said the problems raised in relation to polling stations occurred countrywide.

The auditors said that GECOM's operational planning was devoid of systematic contingency planning.

One area that came in for significant criticism was the communication system for results transmission which the auditors said was severely hamstrung by the late arrival of mobile phones and insufficient numbers of these.

"The lack of a contingency plan for communication caused a situation where the officials in question had nothing to fall back on when the phone communication system did not work", the report found.

It said that the partial breakdown in the communication system caused a significant delay in transmitting results and that polling stations lacked proper communication during polling day. Polling day observers had pinpointed this as a problem when late in the evening of March 19 GECOM apparently had great difficulty communicating with its stations on what the official closing time for polling was.

Lack of time and experience was cited by the auditors for the communication problem but they also took a swipe at international advisors and consultants "who suggested high tech solutions to problems with insufficient time for effective implementation".

The report stated that the delays and errors in the transmission of results occasioned a breakdown in the confidence of the public.

The counting process employed on polling day was said by the auditors to have followed international practice with the necessary security measures built in.

Several proposals were made for bettering the results transmission process.