Elections and Registration in Afghanistan (ERA) Project
Voter Registration Sequencing and Related Activities
ERA Topical Report #8
January 25, 2003
1. At the moment, there are several related population tallying activities in execution or planning stages for 2003, including
- Household prelisting with a test application for voter registration and census preparation currently in train in the seven districts of Kunduz province
- Voter Registration lists, cards and identification of polling stations
- DDR campaign at a national and regional level possibly linked with PRT deployment
- Tazkira (National Identity Card) renewal of a system operated by the Ministry of Interior which was never complete and which has deteriorated since 1973-78
- Full census with any accompanying demographic and economic sample surveys
Activities a. and b. must take place before the election scheduled for June 2004.
It seems activity e. is not contemplated until after the Bonn Accord date.
Activity c. necessarily accompanies a. and b. as it deals with election security.
Activity d. could conceivably lead a. and b. with a current Tazkira being evidence of voting eligibility. But to avoid system overload, d. could also follow and benefit from successful campaigns in a., b. and c.
Each of the above activities will involve mobilization and data collection at the Woleswali (District) and sub-district levels. For example, in most of the rural areas there are regionalized village (Quarya) clusters which could support one or more polling stations within a district and in the major urban areas there are recognized Nahyas (Wards) within the district perimeters.
2. Inevitably, because of the difficult terrain, poor transport, large population and extensive area of Afghanistan, these activities will be carried out in time phases (district by district or province by province). Because of the shortage of trained and experienced personnel, the same individuals might be engaged in more than one of the activities sequentially.
What follows is very preliminary and subject to discussion, correction or further elaboration in the next few weeks.
3. History offers some lessons to be learned from previous large-scale efforts to tally Afghanistan's population by district. These include:
- The 1974 Afghan Demographic Studies (ADS) Demographic Survey and "Sample Census"
- The 1979 PDPA partial census and household pre-enumeration tallying
- The 1990 USAID estimates combining a. and b. with best available estimates of population growth including war-related deaths and refugee outflow. (This study includes bound computer outputs showing area and population estimates for 1974, 1979 and 1990 by 325 districts together with a 1:1,000,000 map showing district IDs, names and boundaries resulting from the ADS project.)
The author was heavily engaged in a. and, with Afghan help, solely responsible for c. with any of its strengths and weaknesses. These sources have been described elsewhere and useful lessons learned can be applied in the future. The important thing to recollect for now is that anything beyond these three sources is largely derivative or incomplete in area coverage until very recently. The further we are from the critical year of 1979, the shakier the inferences.
The population emphasis recognizes potential voters as a definable and internationally accepted sub set, namely males and females age 18 and over. The district emphasis recognizes these as administrative building blocks for the larger and more heterogeneous provinces. The distribution of services—schools, health centers, humanitarian assistance—is usually allocated on some basis of observed or claimed needs of the local population, and herein lies the propensity in Afghanistan and elsewhere to inflate numbers in order to call forth resources or accrete political influence.
4. Technology now available offers some short cuts and quality control measures and also some possible pitfalls, compared to the three previous attempts to estimate or tally the population by district. The following technologies have been deployed, or could be deployed, successfully in Afghanistan for voter registration and related activities.
- The computer mapping capabilities of UN AIMS now in place will greatly assist all future population estimating and voter registration activities. To date, the capabilities have been used largely for humanitarian assistance. They can be turned easily to support the above five population activities and indeed are being employed in the CSO Kunduz application.
- Hand-held GPS are already in use by several organizations and have been supplied by UN FPA to the CSO Kunduz team. Intelligently used, they can resolve ambiguities about the location, name and extent of settlements and the site of potential polling centers.
- Computer databases at UN AIMS or elsewhere, organized at recognized district and sub-district levels for storage, retrieval and analysis of population- related data (including refugee estimates for Iran and Pakistan), so long as these databases are not susceptible to hacking and modification for the purpose of electoral fraud.
- Automated Identification Cards specifically for voter registration and Tazkiras. The technology and application of these is now quite common. At least three vendors have been active in soliciting contracts in Afghanistan: Thomas de la Rue (UK), Thales and Sagem (France) and possibly Dyn Corps (US). Several factors require further examination by impartial experts:
- Cost relative to benefits and alternative uses of funds. One vendor made a rough quote of $20 million to $50 million.
- Complexity where a simpler manual system might be cheaper, less disruptive and more fraud resistant (Nigeria 1963, Liberia 1987, and c. above).
- Advantages of such a system include international acceptability, multiple uses, and likely reduction of double registration errors, but not necessarily errors of coverage and under-registration of potential eligible voters.
5. There are three new population data sources by district in addition to the pre-1990 sources listed above. They are all available at UN AIMS.
- CDC (Centers for Disease Control) estimations are derived by plausible, but likely unverifiable, assumptions about coverage, drawn from recent child immunization data from Afghanistan.
- ORL (Oak Ridge Laboratories) estimations are based on commercial satellite imagery correlated to intensity of land cultivation and urban morphology and light emissions. To our knowledge, there has been no ground truth household or structure count data to validate the model. However the correlation between population density and vegetative cover and land-use intensity has been established in the past for rural areas (roughly 80-85% of the population). The technique, compared to the other two sources, might underestimate urban centers.
- CSO (Central Statistics Office, Kabul) publishes annual estimates. In the past they may have been tied to the 1979 sources and the chosen growth rate may be plausible, but it is applied rather uniformly across the country. The results are bedeviled by new districts which may not have been gazetted, and it is not certain whether the village and populations in these new districts have been subtracted from their "parent" districts, thus opening the way for population inflation.
All sources come out about 21-22 million for the total population, which accords with a rough extrapolation from the 1990 figures. As there is no "correct" basis for comparison, results based on calculator correlations and regressions among the three new sources are suggestive at best.
6. Household prelisting (Para. 1a. above) is undergoing a test in the seven districts of Kunduz province at the moment. Fifty CSO employees are directly involved. The operation is planned as a baseline or feeder to voter registration and a full census (Paras. 1b. and d. above) and a field test of the CSO estimates. To achieve speed and accuracy of coverage, only the number of males and females over age 18 is tallied, with a 1/200 sample of ages and household relationships.
A concurrent geo-coded Settlement Facilities Survey should serve to establish polling places within each district. Good practice would have potential registrants assigned to a polling place in advance of voting.
7. It appears, as intended, that the best path to an effective voter registration system is through the UN FPA-assisted CSO Household prelisting exercise. It is hoped this Kunduz pre-listing and facilities survey, modified as necessary by field experience, can be completed on schedule and country-wide in advance of a separate registration campaign closer to the election date.