The type and sequence of registration activities vary considerably according to the specific system and methods chosen for gathering voters’ data. These methods can be divided into three broad categories based on the levels of technology currently applied to the process:
Biometric systems are used to measure physical characteristics and behaviours (fingerprints, speech, face, iris and the like) with the objective of recognizing patterns to distinguish those that match closely enough to be considered identical from those that are different enough to be considered non-identical. Biometric systems have proven particularly useful in two specific segments of the electoral cycle, namely voter registration and voting operations. If Automatic Fingerprint Identification System (AFIS) technology is used during voter registration, duplicate registrations can be detected. AFIS technology refers to software applications capable of establishing the identity of an individual through fingerprints by the use of biometric functions. AFIS technology has been included in many voter registration processes in the last years, and in some cases it has been included in the legal provisions governing voter registration.
Biometric systems are in fact increasingly considered to be the definitive solution to voter impersonation and multiple voting practices. They are especially popular in post-conflict countries and emerging democracies with either very limited or non-existent forms of civil registry identification, such as DRC, Togo, Guinea, Angola, Nigeria, Mozambique and Pakistan, and are under consideration in a large number of developing countries.
The importance of the centralised matching function – It is imperative, for a biometric registration system to be successful, that the data for all voters is matched against all other voters to allow the system to identify all potential double registrants. Many biometric registration kits, for an additional cost, can be fitted with an inbuilt “Micro AFIS” that allows a suspected double registrant to be immediately detected by the software on the laptop computer where he or she registers more than once, i.e. in the same registration centre. As most voters that attempt to double register will do so in different registration centres, however, there is a requirement for the data on all kits to be matched against the data of all other kits in a centralised location. If there is a live (e.g. by satellite) link-up between all kits at all times during the registration process, it may be possible to “catch” a double registrant that is attempting to double register in two different registration centres when he registers at the second centre. This would be an extremely expensive method of detecting double registrants, however.
If the data from all the kits is only recorded on compact disc or external memory cards and periodically delivered to the centralised data centre at various times during the voter registration process (e.g. once every week), then all data from all the biometric registrations kits will need to be merged in order to allow the technology to throw up all the suspected double registrations when the data for each voter is matched against every other voter, and therefore potential double registrants may only definitely be caught at the end of the process.23 Electoral management bodies should not underestimate the amount of time and resources that need to be allocated to the matching task, depending, particularly, on how they address the following issue.
Prior to commencing the data matching, the electoral management body that decides to employ AFIS or other biometric technology will be faced with the issue of whether or not to allow the technology to definitely determine what is a double registrant, or whether such a decision, which may, in some cases, result in sanctions such as the forfeiture of the right to vote (as in Kenya), should be taken by an EMB official on presentation of a suspected double registration by the technology. Conducting a “naked eye” inspection of a suspected double registration from two digital fingerprints, however, is not an accurate way to decide on a suspected double registrant. In order to improve the accuracy of the decision, therefore, the biometric software can sometimes employ fused algorithms to match both the digital fingerprints with the facial features from the digital photos to arrive at more definitive possible duplicates.
EMB staff will then be able to take a decision on what is a double registrant from the photos. Implementation of such a system, however, needs to be well planned. If, for example, 2 million voters are registered using biometric technology, then the central matching exercise will need to conduct up to 2 trillion matches. An enormous ampount of computing power is necessary for that number of matches to identify a possible, lets say, 0.001% double registrants. EMBs need to plan the manpower, the computing power and the time necessary to conduct the central matching exercise , particularly in cases where the EMB officials make the final decision on each suspected double registrant case thrown up by the system.
Dealing with double registrants – Different countries and EMBs deal with confirmed double registrants in different ways. In some countries, the EMB will simply make a decision as to which of the voter’s registrations (usually the first) is considered the legitimate one, and the second and subsequent registrations will be removed from the voter lists database. In other countries such as Kenya, however, double registration is a criminal offence. The double registrant not only will lose his or her right to vote, but the details of the double registrants may be handed over to the police authorities for prosecution.
The issuance of voters cards at the time of registration may complicate matters further. If the registrant has been issued with his or her voters card at the time of registration (a common practice) then the electoral management body will be faced with the challenge of either trying to recover the voters cards from the double registrants, or, which is the more usual practice, of notifying each polling centre on polling day, of the voters that registered in both that centre and other centres. This list may be required if the double registrant attempts to vote with the voter card issued by that centre. If the voter is not on the real voter list for that centre,24 and has a legitimate voter card for that centre issued during the voter registration period, then the polling staff should be instructed to check the list of double registrants to see whether the voter is a confirmed double registrant.
An exhibition period, where a provisional voter register is displayed some time before the elections in order to allow voters to confirm that their details were entered correctly and lodge objections against voters who they claim are not entitled to vote at that centre (or at all), can be an excellent opportunity for the EMB to address the double registrants issue prior to polling day.
Four potential intervention points to support registration are:
Nigeria is one country in which AFIS has been used for election purposes. The EMB identified and removed millions of duplicates from the voters register based on decentralized datasets (using a number of local government areas rather than the whole country as one dataset) using AFIS. AFIS has also been used in the voter registration process in the Democratic Republic of Congo and Venezuela and electoral rolls have in previous elections been produced by Pakistan’s National Database and Registration Authority (which has database systems including fingerprint biometrics) on behalf of the Elections Commission of Pakistan. It was also considered for the 2003 elections in Yemen to address problems with the voters register. In Yemen, the issue was not to develop a new AFIS system, but rather to explore the possibility of merging the voter register with the existing AFIS-based civil registry system.
ACE Consolidated Reply: "EMBs using Automatic Finger Identification Systems (AFIS)"
