National Risk and Vulnerability Assessment 2003
Afghanistan, 2003
Identification
AFG_2003_NRVA_v01_M
National Risk and Vulnerability Assessment 2003
| Name | Country code |
|---|---|
| Afghanistan | AFG |
Integrated Survey (non-LSMS) [hh/is]
The primary objective of the study is to collect information at community and household level to better understand livelihoods of rural settled populations and nomadic pastoralists (Kuchi) throughout the country, and to determine the types of risks and vulnerabilities they face throughout the year. The many stakeholders can then use the findings of the study to develop strategies to address the short, medium, and long-term needs of these populations through appropriate and timely policy development and intervention strategies.
Sample survey data [ssd]
Community (Shura), Households, and Individuals
Scope
The 2003 Afghanistan National Risk and Vulnerability Assessment Survey covered the following topics which were answered by male and female respondents:
HOUSEHOLD LEVEL
- Household Identification
- Household Register
- Migration
- Education
- Health
- Employment
- Household Assets
- Household Amenities
- Perceptions
- Shocks
- Programme Participation
- Land
- Crop Production
- Farming Constrains
- Livestock
- Dietary Diversity
COMMUNITY / SHURA
MALE
- Community Identification and Geo-reference
- Livelihood Identification
- Seasonality and Labor
- General Community Situation
- Water
- Markets
- Heath Facilities
- General Information
- Literacy and Education
- Wheat Production
- Livestock
- Tractors
- Grazing Land
MALE WEALTH GROUP
- Community Wealth Group Identification
- Education
- Markets and Health
- Livestock
- Land Tenure
- Agricultural Cropping Practices
- Orchards
- 1st Season Cereal Crop Production
- Cash Crops
- Labor Opportunities for Men
- Labor for Children Under 14
- Labor Opportunities for Women
- Expenditure
- Debts
- Remittances
- Coping Strategies
- Priorities
- Intervention Preferences
COMMUNITY / SHURA
FEMALE
- Community Identification and General Community Information
- Markets and Health
- Livestock
- Land Tenure
- Coping Strategies
- Priorities
FEMALE WEALTH GROUP
- Community Identification
- Labor Opportunities for Women
- Household Budget
- Coping Strategies
- Livelihood Constraints
- Priorities
- Intervention Preferences
DISTRICT LEVELE
- District Information
- Kuchi
• Livestock ownership ranges from best-off and poorest kuchi communities in the district
• 21BSummary of population and household number by kuchi type - District Price of Items
| Topic | Vocabulary | URI |
|---|---|---|
| basic skills education [6.1] | CESSDA | http://www.nesstar.org/rdf/common |
| housing [10.1] | CESSDA | http://www.nesstar.org/rdf/common |
| consumption/consumer behaviour [1.1] | CESSDA | http://www.nesstar.org/rdf/common |
| migration [14.3] | CESSDA | http://www.nesstar.org/rdf/common |
| general health [8.4] | CESSDA | http://www.nesstar.org/rdf/common |
Coverage
The survey covered 32 provinces.
Producers and sponsors
| Name |
|---|
| Ministry of Rural Rehabilitation and Development |
| Name | Role |
|---|---|
| European Union | Financial assistance |
Sampling
Afghanistan's last census was implemented in 1979. While the Central statistical Office (CSO) is currently conducting a pre-census exercise, the lack of current census data hindered the ability to design a sample that was based on a framework that would allow the estimation of representative statistics at national or sub-national levels. Instead, the sample design was implemented in two stages: (i) the community selection was done using a number of agro-ecological zones based on estimated land areas for each zone; and (ii) selection of households within a community was based on three wealth group classifications which were defined during the community interviews.
The lack of a population-based sampling frame implies that results from the NRVA do not statistically represent all of rural Afghanistan and are relative, rather than absolute. Still, a number of questions in the NRVA related to the community and wealth group population allow the correction of the household selection probabilities. Therefore, all sample estimates calculated are corrected to adjust the intra-community selection probabilities of each household. Beyond the NRVA sample, the data do not represent more aggregate rural regions or provinces, so all results and conclusions derived from the survey can be used only to make relative comparisons across provinces and agro-ecological zones.
Weighting Systems
The system of weights used when analyzing the NRVA data was devised to adjust the sample to be representative of the communities from which the data was sampled. This weighting does not allow one to make statistical inferences at the national or provincial level, but does allow one to make inferences only to the sample population.
Weighting
The lack of reliable population and village location data ensures that NRVA 2003 could not be designed using a proportional sampling frame. Because of the lack of information on the relative populations of the zones within districts not to mention districts and provinces, the only truly representative statements that the data can make is about the population sampled, i.e. 1853 villages, which represent 175,026 households, and 1,273,314 household individuals (assuming average NRVA household size was 7.27) in all but 11 districts. If we assume the rural population is as the latest Central Statistics Office (CSO) population estimates indicates, 16.06m, then this population of individuals represents 12.6% of the Afghan rural population.
Offsetting this lack of statistical representative sampling frame is the large sample size for this type of survey. While we are unclear how to weight between villages, the large sample size gives us a reasonable sense of confidence that the data is probably quite robust. It is hoped that the current CSO pre-census listing accompanied by geo-references will eventually enable a retro weighting to be applied to this sample.
Therefore the weighting system can only accommodate for the different sizes of wealth groups and number of households and individuals interviewed within each wealth group.
Weighting system for Shura level statements (male and female)
The weighting system, for producing means, medians, percentages, etc., where maintaining the correct degrees of freedom is not important. Since one interview was done in each shura (village/community), the weight for each shura interview is simply the number of households in that shura. This system of weights adjusts the sample to be representative of the number of households in the sample.
This weighting system will increase the n dramatically, creating the pseudo-replication explained above. This is irrelevant when calculating means, medians, percentages, and other descriptive statistics where degrees of freedom are not considered, but becomes an issue when using statistical tests where the number of degrees of freedom is important in determining is significance. The way different statistical packages handles such a weighting system may vary.
Note: When drawing conclusions at the shura or community level (e.g.- % of shuras producing handicrafts in the winter) no weighting system should be used.
Weighting system for Wealth groups within village/community statements
Wealth groups within village/community were the next unit of observation, and these should be weighted by their population size. In a wealth group we typified one household to represent the wealth group, and therefore the weight of this wealth group information is equal to the population of that wealth group.
Weighting system for Household level statements
- Weighting system, for producing means, medians, percentages, etc., where degrees of freedom are not important.
The weight for each household is calculated using the following formula:
(#hh_in_wealth_group)/(#_interviews_in_wealth_group)
This system of weights adjusts the sample to be representative of the households the data was sampled from (that is, representative of the villages included in the survey). The sum of these weights is equal to the number of households in all of the villages sampled.
Weighting system for individual level statements
The following formula should be used with weighting system for households:
((#hh_in_wealth_group)/(#_interviews_in_wealth_group))*#_members_in_hh
This is equal to:
= HHweight*#_members_in_hh
Survey instrument
The district questionnaire was used to collect information from Key Informants, such as District Authorities, Kuchi leaders, and Veterinary Field Units, in order to determine the different agro-ecological or livelihood zones within a district. This information was used to rank districts according to their vulnerability to food insecurity. The ranking exercise used information on access and availability to markets, health facilities, water, and education as well as the general physical environment, security, and presence and location of land mines. In addition, population estimates were collected to facilitate planning and targeting of potential interventions. It is understood that these are rough estimates that will need to be updated by the pre-census survey currently being undertaken by the Central Statistics Office (CSO).
Through focus group discussions and key informant interviews, the shura questionnaire
provides an overview of the community access to markets and health facilities, along with estimates of education levels and literacy, past and anticipated exposure to shocks, and priorities for community members. The shura's were also asked to stratify the households in the community into wealth groups: very poor, poor, medium, and better off families. This information was then used to estimate the population in each category. Where possible, both male and female focus groups were interviewed. Women’s discussions focused more on their roles in the community and households, education, constraints to livelihoods, female-headed households and women’s decision-making roles. Where it was not possible to conduct female shura focus groups, the male focus group was asked to provide information on female labour activities and opportunities.
Focus group discussions for the wealth group questionnaire were conducted for community members in the very poor, poor, and medium wealth groups only. Separate male and female wealth group interviews were conducted where possible. The better-off groups were excluded because they were not expected to be vulnerable. The focus group interviews collected information on: typical agricultural activities, livestock, labour and income (activities and amounts), and access to markets, health and education. In addition, focus groups also provided their inputs on priority interventions to improve the quality of life for members of their communities.
For the household questionnaire, approximately 6-7 household interviews were conducted in each community. The questionnaires included modules on household demography, education, health, migration, housing, income activities, household asset ownership, risk exposure and response, agricultural activities, livestock ownership, and food consumption (7-day food frequency).
Data collection
| Start | End |
|---|---|
| 2003-07 | 2003-10 |
| Name | Affiliation |
|---|---|
| Central Statistics Office | Ministry of Rural Rehabilitation and Development |
The NRVA was launched in the third quarter of 2003 with WFP VAM taking on the role of coordinating the assessment on behalf of the Ministry of Rural Rehabilitation and Development (MRRD). The data collection for the NRVA was implemented between July and September 2003.
Eleven trainings were run by the VAM team with support from TUFTS University throughout all WFP Area Offices in July, to prepare the enumerators for the data collection component of the NRVA. The training sessions, which were both theoretical and practical, lasted 5 days and were followed by a test to ensure that only those participants that clearly understood how to collect the data were selected as surveyors. Of the total 351 participants that were trained, 158 men and 111 women were selected and actively participated in the NRVA data collection. Of these, 114 were from various line Ministries and the remainder were from other agencies, NGOs, or were locally hired and recruited. In the middle of July, the data collection began, with most teams comprised of two men and two women visiting District Authorities and villages, conducting focus group discussions, and household interviews.
The original plan was to include villages throughout the country, though insecurity in some areas (mostly in Uruzgan, Zabul, and Paktika provinces) resulted in these areas not being fully assessed. Throughout the south, insecurity also prevented women from participating in the assessment in many districts. For those populations, extrapolations and comparisons from surrounding areas will form the basis of understanding their situation. In all, more than 90 villages originally selected for the assessment were not visited for security reasons.
Each enumerating team had a Team Leader who was responsible for quality control. Each province and/or region had a VAM Team Coordinator, whose responsibility was to visit the enumerating teams, providing technical support. On completion of the survey in a region, teams were brought together by the Team Coordinators for a final screening of the questionnaires and a debriefing, prior to sending the questionnaires to Kabul for data entry.
Although the NRVA was logistically implemented and coordinated by WFP VAM, a rotating coordinator from the stakeholder group of the NRVA took the lead in providing information updates and assisting field teams with any key problems that may have arisen. This initiative also proved to be successful in increasing the understanding and ownership of the NRVA as a Ministry led multi-stakeholder assessment.
Data processing
Data entry was conducted in three ways, depending on the data level. Data from the district questionnaires were entered manually by VAM teams in WFP Area Offices, and sent electronically to Kabul. Shura and wealth group data was transcribed by VAM and key enumerator staff onto scannable formats, and forwarded to Kabul for electronic scanning into an Access database using TELEform Enterprise, a data scanning software package licensed by Cardiff Software. Household questionnaires were forwarded to Kabul and entered by staff from the Ministry of Rural Rehabilitation and Development, and the Ministry of Agriculture and Animal Husbandry. Data cleaning and analysis utilized a broad variety of software packages, including Access, Excel, SPSS, and GenSTAT.
Data appraisal
Data Constraints and Limitations
The NRVA data is able to answer many questions about the status of populations in rural Afghanistan. However, there are several constraints and limitations to the data that must be taken into account when interpreting the results. These data limitations may also serve as a guide for improving future assessments.
-
When mapping data at the district level, the 1984 AIMS districts are used. However, these districts do not consistently line up with the current district definitions. Province definition has not changed, and is accurately represented in the maps.
-
The use of ‘agro-ecological zones’ in data collection and analysis is subjective. In a country as topographically and climatically diverse as Afghanistan, the variation of agro-ecological zones is immense and cannot be fully represented by only five zones. For example, irrigated land in the north is different from that in the south. Seasonality of crops, number of crops per season, and other factors can vary significantly within one agro-ecological zone.
-
It is important to highlight the non-random method of village selection. An attempt was made to represent all agro-ecological zones present in each district, using a purposeful sample. Due to the lack of sampling frame, this method of selection was chosen. Although confidence intervals, standard deviations, and p-values can be calculated, they are only reliable when making estimates to the sampled population, and not to the provincial or national level.
-
These constraints in the strata and sampling mean that results cannot be reliably inferred to a level higher than the population selected from. Certain data, such as the shock of an earthquake flood, or insecurity, may be more reliably applied to the district or provincial level. Other data such as frosts, access to health services or public transportation may also be relatively robust despite the non-random sampling. However, data such as literacy, dietary diversity, school attendance, and labour must be interpreted carefully. National or provincial estimates for these indicators then serve as benchmarks for relative comparisons or for monitoring trends rather than actual estimates. In these cases, two-way and multi-way analyses will provide information that may be more reliable.
-
The term ‘Wealth group’ is used throughout the report. The data collection was structured around the concept of four wealth groups (better off, medium, poor, and very poor). Households were categorized into these 4 groups by village shuras, based on perceptions of social economic status. However, the perception of wealth group is subjective, and the definition of wealth group can differ between communities.
-
Further, the wealth group data was not gathered, for most indicators, from the better off households. However, in order to be able to more accurately make general statements about the populations, the medium wealth group was weighted to include the better off population. This means that the better off households are represented by the medium households which may result in the population, when not stratified by wealth group, appearing poorer than it actually is.
-
It is also important to highlight again that female shura and female wealth group data have not been gathered in several districts, particularly in the south, due to cultural and security restrictions. This is important to note when interpreting any of the female shura or wealth group data results for these areas.
-
The consumption data are simply an estimate of the weight of the different types of food consumed by the household in the past 7 days without the use of scales or other measuring devices. The number of people in a household was determined using the household register section of the questionnaire. Due to an oversight in questionnaire design the number of people specifically present at meal times during the 7 day recall period was not recorded. Also, the method did not provide an opportunity to account for food wasted or fed to animals.
Data Access
| Name |
|---|
| 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.
Contacts
| Name | URL | |
|---|---|---|
| Central Statistics Organization | mail@cso.gov.af | www.cso.gov.af |
Metadata production
DDI_WB_AFG_2003_NRVA_v01_M
| Name | Role |
|---|---|
| World Bank, Development Economics Data Group | Production of metadata |
Metadata version
Version 01