In 2003 the government of Afghanistan launched a unique national assessment to support the development work of different government and non-government organizations working in different sectors. Prior to this assessment, most of these organizations were doing their own surveys based on their demands. The World Food Programme (WFP) of the United Nations was managing a nationwide assessment and collecting information on food security, and there was a wish among the development community to have an improved methodology for nationwide assessments. A workshop was held in Mazar-e-Sharif in April 2003 to design a Nationwide Assessment based on government decision and stakeholder demand, and the outline of a questionnaire for the 2003 survey was prepared. This questionnaire was further developed jointly by the WFP and the Vulnerability Analysis Unit (VAU) within the Ministry of Rural Rehabilitation and Development (MRRD).
The National Risk and Vulnerability Assessment (NRVA) 2003 was implemented between July and September of 2003 lead by WFP and supported by the Vulnerability Analysis Unit (VAU) of the Ministry of Rehabilitation and Rural Development (MRRD). A workshop was held in June 2004, in which some preliminary results of the 2003 were discussed and recommendations were made for the NRVA 2005.
The First Phase of the National Surveillance System (NSS) project was launched in September 2003 by MRRD, with financial support of the European Commission and the United Nations; the objective of NSS was to lead and coordinate, among NGOs and government agencies, the efforts to monitor the situation in the country. A final report of NRVA 2003 was released on December 2004. A consumer satisfaction survey was carried out at the end of 2004 to ascertain the extent of use of NRVA-
2003 data and to collect demands for data for the NRVA 2005 from different governmental and non governmental organizations.
Preparatory work for the NRVA 2005 was underway in the spring 2005, coinciding with the beginning of the Second Phase of the National Surveillance System (NSS) Project, of which both the Central Statistic Office (CSO) and MRRD are the implementing agencies. While CSO leads on data gathering and its custody MRRD helps on survey implementation, analysis and interpretation, as well as dissemination of findings. The goal of the NSS project is to reduce poverty in Afghanistan and it has four objectives: to conduct nationwide assessments such as the NRVA, to conduct emergency assessments and response to protect deterioration of people’s livelihoods (early warning systems), to carry out special studies upon demand, and to contribute to the capacity building of government staff.
The NRVA 2005 survey, implemented between June and August 2005, was a massive and concerted effort in which very isolated areas in the country were sampled. Since the completion of the field survey both CSO and MRRD have been actively sharing data, information and analytical services to the development community within Afghanistan and internationally. The NRVA 2005 is an instrument that allows in-depth analysis of different aspects of the Afghan economy. As it will be explained below, the NRVA 2005 is not statistically comparable with NRVA 2003, as such it is not possible to assess rates of change with respect to 2003 (i.e., NRVA 2005 is statistically representative at provincial and national level while NRVA 2003 is not); however, NRVA 2005 but sets a baseline that can be used for future assessments, such as the NRVA 2007 and others to follow. As it will be shown here, NRVA 2005 is a forward-looking instrument that has been used to develop the capacities of the national staff at CSO and MRRD in reference to Millennium Development Goals. Household quantitative data form the building blocks that support this report. These blocks are complemented with both the perceptions of shura assemblies and perceptions of households.
Preparatory arrangements for the NRVA 2007 are underway and it is expected that all lessons learnt during the implementation and analysis of NRVA 2003 and 2005 will built a cornerstone in the planning and implementation of NRVA 2007. In particular it is currently discussed to randomly distribute the upcoming NRVA samples into 12 months rather than to carry out a whole assessment in one season of the year. This shall ensure a further significant reduction of non-sampling errors in the field.
The success of the NSS project should be measured in terms of empowerment of the Afghan staff to design, implement, analyze, interpret and disseminate findings of the nationwide surveillance system.
The primary objective of NRVA 2005 is to collect information at community and household level to better understand livelihoods of Kuchi (nomadic pastoralists), rural and urban households throughout the country, and to determine the types of risks and vulnerabilities they face. National and international stakeholders can benefit from the summarized findings of the report or the data set made available for in-depth analysis to develop strategies to address the short, medium, and long-term needs of the nomadic, rural and urban populations through better informed and timely policy development and intervention strategies.
The 2005 Assessment takes into account a series of recommendations made by several stakeholders during a workshop held in June 2004 when the preliminary NRVA 2003 results were discussed. The assessment includes urban households allowing a more comprehensive appreciation of the status of the country in the summer of 2005.
Kind of Data
Sample survey data [ssd]
Unit of Analysis
Community (Shura), Households, and Individuals
Version available at the World Bank, converted into Stata.
Section 17 missing.
The scope of the National Risk and Vulnerability Assessment survey includes:
- Household level, male: Household register & education, housing, household Facilities, drinking water, assets & credit, livestock, agriculture & land tenure, migration remittances & social, sources of income, household expenditures, cash-for-work, food aid & iodised salt, household shocks & coping and HIV / AIDS.
- Household level, female: Food consumption – women’s, maternal child health, children 0-59 months and HIV / AIDS & Literacy test.
- Community level, male: Community information, access / infrastructure, markets access, health access, education, community roles and governance, program activities, community priorities and water table.
- Community level, female: Health access, community bodies & governance and community priorities
The survey covered 34 provinces excluding 6 districts.
Producers and sponsors
Central Statistics Office (CSO)
Ministry of Rural Rehabilitation and Development
A sample of 30,822 households from 34 provinces (1,735 Kuchi, 23,220 rural and 5,867 urban) was drawn excluding 6 districts that were not enumerated (as CSO household listing data was not available at the time of sampling the Livestock Census [FAO, 2003] data was used). Twelve districts were enumerated only by male surveyors in all Zabul (11 districts) and Maruf district in Kandahar due to security restrictions; however, in the se districts the food consumption part of the female questionnaire was filled out by male enumerators interviewing male respondents.
Rural and Urban Settled Households
The analytical domain, the unit at which the data are statistically representative, is at the level of 34 rural provinces; in contrast to NRVA 2003, the province of Uruzgan was split into smaller Uruzgan and Daykundi; the same happened to Parwan, which was split into Parwan and Panjsher. In addition to these 34 provincial analytical domains, there are 10 urban areas with populations larger than 10,000 households.
The survey has also collected data representative of these 10 urban domains. Thus, there are 44 settled analytical domains. Because Kuchi have been considered as one national analytical domain, there are a total of 45 analytical domains for NRVA 2005. Collecting representative data with a proportional sample at the provincial level creates a challenge because of the large variation in provincial population from the smallest population in the province of Nimroz, with only 13,941 rural households, to Hirat, with 226,650 rural households. To adjust the sampling to the available budget, the province Jawzjan with 50,900 rural households, has been used as the base analytical domain for which the sampling fraction has been determined. For those domains with populations less than Jawzjan, and where the sample fraction delivered less than 350 households, further clusters were added to ensure a minimum sample size of 350 households. The sample is therefore not self-weighting.
For those provinces or districts within provinces where the sample frame was not yet available at the time of sampling (42 districts), the Livestock Census database was used to draw a sample. On arrival at a village, the number of households was determined during the male community interview. As it was difficult for the enumerators to predict the number of households within dwellings, an additional question was asked for the total number of dwellings in the village. This number was divided by 12, to create a sampling interval for households within the community. The enumerators then selected a household each time they counted the sampling interval houses. By using this method, the sampled households were randomly and spread equally throughout the village.
The household listing conducted by CSO did not effectively include the migratory Kuchi population to the date of the survey; hence there was no effective sampling frame for this population. Apparently, this lack of enumeration of the Kuchi population includes those that have recently settled. This is exactly the same population that was surveyed during winter/spring 2004 by the National Multi-Sectoral Assessment for Kuchi (NMAK), i.e. the Kuchi that is still nomadic and those that have recently settled since the onset of the last drought period. This is the best estimate of the current Kuchi population. The unit of observation for the survey was the Kuchi communities in their winter location, where one or more Kuchi communities may have been located. The sample frame for the survey was created by constructing the predicted Kuchi populations in their summer location, for which information was collected from the NMAK 2004 survey.
Dates of Data Collection
Data Collection Mode
Data Collection Notes
Government and non-government organizations at the provincial and national level were contacted for provision of male and female enumerators, as the team composition was designed for four enumerators (two males and two females, based on the questionnaire design). Full advantage was taken from WFP’s regional level enumerators who had previous experience in surveys. Training was conducted in eight regions using two trainers. A test was given to the enumerators, and only those who passed were included in a four-person team (two females and two males, excluding Zabul).
Support from local mullahs and teachers in the Maruf district of Kandahar, Logar and Zabul provinces was sought by the NRVA management team to overcome security issues. The field work started in June and was completed in August 2005. Readings with Global Positioning System were taken to verify the locations of the sampled villages. Eight regional coordinators were hired by the NSS project to supervise the enumerators in eight regions. In addition, the WFP/VAM regional team leaders and monitors were involved in the supervision of field work of NRVA 2005 enumerators.
Central Statistics Office
Ministry of Rural Rehabilitation and Development
The core of NRVA 2005 is being formed by the household questionnaire. The household questionnaire consisted of the following 18 sections; the first 14 were answered by the male head of household or male respondent, and the last four by the female members of the household:
- Household register and education;
- Household facilities;
- Drinking water;
- Assets and credit;
- Agriculture and land tenure;
- Migration, remittance and social networks;
- Sources of income;
- Households expenditures;
- Cash for work;
- Food Aid and iodized salt;
- Household shocks and coping strategies;
- Food consumption;
- Maternal child health;
- Children 0 – 59 months;
- HIV/AIDS and literacy test.
The total number of questions that were asked to the sampled households exceeded 260 but not all questions were answered because some of them were eliminated based on the responses provided (with skipping rules). The household is regarded as the unit of analysis. In Afghanistan there is a need to address the questions to males and females depending on their nature. In every sampled community 12 households have been interviewed. On average the time required to answer the household questionnaire was less than two hours. Besides the household questionnaire, information was gathered at community level. Therefore, two community questionnaires were designed – one male and one female. These two questionnaires addressed the following topics:
Male shura questionnaire:
- Community information;
- Access to infrastructure;
- Markets access;
- Health access;
- Community roles and governance;
- Programme activities;
- Community priorities;
- Water table.
Female shura questionnaire:
- Health access;
- Community bodies and governance;
- Community priorities.
Automated data entry
Teleform Enterprise version 8 (Cardiff software, donated by WFP) was used throughout the process to scan the NRVA 2005 Teleform questionnaires filled in the field. Teleform is an electronic pre-programmed method of gathering data (optical readable software), often used for its speed and accuracy in large surveys and censuses. A scanner capable of processing 60 sheets per minute was used. Unlike NRVA 2003, where Teleform was only used for the shura and wealth group data after being transcribed by VAM and key enumerator staff into scan able formats; finally the information was scanned into a Microsoft Access database using Teleform.
The NRVA 2005 was completely designed in Teleform; then the enumerators filled in the pre-designed questionnaire sheets and the data were directly scanned into the Access database. Scanning 1.3 million data sheets took two to three months more than anticipated; the process was finally finished in February 2006. These delays were partially due to the quality of enumeration of questionnaires, computer hardware that was not powerful enough to sustain the processing required (alleviated by the loan of a high-speed server from UNOPS) and the absence of a stable electricity supply (alleviated by the loan of the power generator from WFP).
Once the data were scanned, the programme logically checked if the number of responses per question was not exceeded. Unfortunately, within NRVA 2005 a decision was taken to insert the number of the response within the answer circles. This resulted in some false positive answers as a high percentage of the answer circles were already coloured. Only when a true answer was also indicated (giving two responses) the programme stopped asking for verification, if there was no response then the false positive was accepted and these responses were taken out during normal cleaning practices. Once a questionnaire was validated, the image file was deleted and the data was written to the Access database. Descriptive statistics were estimated with SPSS and Genstat. Cluster analysis using ADATTI software was used for food security profiling. Provincial statistics produced are included in the Annex; those for national, Kuchi, rural and urban categories are included in the main body of the document.
Data constraints and limitations
In spite of the time spent on the design of the questionnaire and its implementation in NRVA 2005, the data gathered have the following limitations:
- Seasonality. Food security assessment and household perceptions are only valid for the summer season, rather than for the whole year.
- Limited data on non-food consumption. Due to the multilateral nature of the assessment most of the non-food consumptions (except communication costs) have been included as groups to avoid an exhaustive questionnaire with a strong risk of lowering the quality of data.
- Income. The module on income was designed to look at the number of income generating activities in a household; these can be used for profiling household livelihoods, but should not be used for quantification of income.
- Infant and child mortality. Although the age of each child bearing woman was included in the male questionnaire, it was impossible to construct an unambiguous link of these ages to the multiple women in the female questionnaire. Therefore, it is not possible to calculate child mortality for a standard reference period with this design.
- Rounding food consumption quantities. The food consumption quantities were obtained by asking female respondents of households to estimate the weights of the different types of 69 foods consumed by the household in the past seven days. If weights were provided in local units, then the enumerator, with support from the respondent, estimated the weights in kilograms. Inevitably, it must be assumed that there was wide scale rounding up or down.
- Inseparable Kuchi information. Kuchi, as the nomadic pastoralists are known in Afghanistan, were sampled as one national analytical domain regardless of considering their migration pattern. While there are two main types of Kuchi population in Afghanistan: long range and short range migratory Kuchi, NRVA 2005 did not collect this information from the Kuchi communities that it enumerated. While there might be socio-economic differences between these two types of Kuchi, these groups cannot be identified within NRVA 2005 sample.
- Security limitations to female enumerators. Due to poor security for female enumerators in 11 districts of Zabul province and one other district (Maruf district of Kandahar) only the food-consumption module of the female questionnaire was enumerated by men to male heads of households or male respondents, to provide food security and household perceptions from these highly insecure areas.
NRVA 2005 in relation to NRVA 2003
It is very tempting to compare the figures of NRVA 2003 with those of NRVA 2005. However, the methodology used in NRVA 2003 is different from the one used for NRVA 2005. The main differences between NRVA 2003 and NRVA 2005 can be summarised as follows:
NRVA 2003: Sample frame not proportional to population. Sample selection was not random (based on different agro-ecological zones). Household selection was not random (based on wealth groups).
NRVA 2005: Proportional to population (used updated CSO figures). Sample selection was based on random start method to have a better geographical distribution of the sample. The household selection was based on the random start method within villages.
NRVA 2003: Four levels of data collection: District level, Community shura level, Wealth group level, Household level.
NRVA 2005: Two levels of data collection: Community shura level, Household level. Only market data is collected at district level.
NRVA 2003: Covered only rural areas and Kuchi.
NRVA 2005: Covered rural and urban areas and Kuchi.
NRVA 2003: Used common questionnaire and after completion of the field work, the data were transcribed into Teleform format.
NRVA 2005: Teleform questionnaires were used in the field and used as the data entry instrument.
NRVA 2003: Coverage: 32 provinces / 368 districts / 1,853 villages / 5,559 wealth groups / 11,757 rural households
NRVA 2005: Coverage: 34 provinces / 392 districts / 2,597 clusters / No wealth groups / 30,822 households
NRVA 2003: Female enumerators were not involved in the south and most port of eastern areas; so, female information is lacking. Poor female coverage.
NRVA 2005: Female enumerators participated in the whole assessment in all provinces except Zabul. Good female coverage.
NRVA 2003: Managed by WFP/VAM.
NRVA 2005: Managed by Government (CSO-MRRD).
NRVA 2003: Partial stakeholder participation for questionnaire design.
NRVA 2005: Full stakeholder participation questionnaire design.
NRVA 2003: Several different trainers were involved in the training of the enumerators. This allowed the possibility for variation among enumerators.
NRVA 2005: Uniform training: just 2 trainers who were involved from design to implementation for the whole country.
NRVA 2003: Household food consumption, but no household non-food consumption. Food poverty calculation only possible at the household level.
NRVA 2005: Household food consumption and about 25 items of non-food consumption at the household level.
Thus, it may be safely assumed that the quality of the NRVA 2005 data is superior to that of NRVA 2003. The development of these two rounds of NRVA must be regarded as a learning curve for all Afghan stakeholders. Apart from NRVA 2003 and 2005, other household surveys have been conducted in Afghanistan. Some of them covered similar topics as NRVA, for example the Demographic and Health Survey and the Multiple Indicator Cluster Survey (UNICEF, 2003) are some of the surveys that deal with public health.
Central Statistics Organization
Use of the dataset must be acknowledged using a citation which would include:
- the Identification of the Primary Investigator
- the title of the survey (including country, acronym and year of implementation)
- the survey reference number
- the source and date of download
Central Statistics Office, Kabul
Ministry of Rural Rehabilitation and Development
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.