Household Poverty Survey 1998
Gambia, The, 1998
Created on
Apr 25, 2019
Last modified
Apr 25, 2019
Page views
10
Identification
Survey ID Number
GMB_1998_HPS_v01_M
Title
Household Poverty Survey 1998
Country
| Name | Country code |
|---|---|
| Gambia, The | GMB |
Study type
Priority Survey (hh/ps]
Series Information
The 1998 National Household Poverty Survey Report is based on a nationwide poverty survey which was conducted in March and April of 1998 by the Central Statistics Department. The study was commissioned by the Strategy for Poverty Alleviation Coordinating Office (SPACO), Department of State for Finance and Economic Affairs. The report is an integral element of the National Poverty Monitoring System which is designed to track selected socioeconomic indicators at the household level in order to establish the incidence, nature and characteristics of poverty in The Gambia. The current study will constitute the baseline against which future surveys will be assessed.
Abstract
Rather than studying the entire population, the 1998 Household Poverty Study opted for a sample survey. The advantages of sampling against a complete coverage are well documented and will not be dwelt on here. This notwithstanding, it is worth mentioning that this option allowed for a wide range of issues to be studied. In all, the survey collected information on issues such as education, health, employment and earnings, anthropometry, demography, among others.
Kind of Data
Sample survey data [ssd]
Unit of Analysis
- Households
- Individuals
- Individuals
Version
Version Description
Version 1.0 (final dataset).
- Missing section 0, section 2 on fertility, section 8, section 11 and the sections on Non-farm enterprises.
- Missing section 0, section 2 on fertility, section 8, section 11 and the sections on Non-farm enterprises.
Version Date
1998-04-01
Version Notes
This version was taken from the World Bank archives. Except for expenditure and a few others, almost all the variables were declared as strings in the original dataset. They have all been converted to numeric in this study to facilitate the analysis. Additional data cleaning has been done on the datasets. These were mostly errors linked to the creation of key variables and the removal of redundant records.
Scope
Notes
The 1998 Household Poverty Survey covered the following topics:
- Household Particulars
- Household Demographic
- Health
- Education and Literacy
- Employment
- Non-Farm Enterprise
- Housing
- Environment
- Household Income
- Consumption of Own Produce
- Household Expenditure (non-food items)
- Household Expenditure (food items)
- Miscellaneous Income and Expenditure
- Transfer Payments Made by Household
- Transfer payments Received by Household
- Anthropometry
- Household Assets
- Household Particulars
- Household Demographic
- Health
- Education and Literacy
- Employment
- Non-Farm Enterprise
- Housing
- Environment
- Household Income
- Consumption of Own Produce
- Household Expenditure (non-food items)
- Household Expenditure (food items)
- Miscellaneous Income and Expenditure
- Transfer Payments Made by Household
- Transfer payments Received by Household
- Anthropometry
- Household Assets
Coverage
Geographic Coverage
National
Producers and sponsors
Primary investigators
| Name | Affiliation |
|---|---|
| Gambia Bureau of Statistics (GBOS) | Department of State for Finance and Economic Affairs |
Producers
| Name | Affiliation | Role |
|---|---|---|
| United Nations Development Programme | UN | Technical assistance |
| International Labour Organization | UN | Technical assistance |
Funding Agency/Sponsor
| Name | Abbreviation | Role |
|---|---|---|
| United Nations Development Programme | UNDP | Funding |
| International Labour Organization | ILO | Funding |
Sampling
Sampling Procedure
SURVEY DESIGN
The sample size of any study depends a to large extent on three key factors:
1. The degree of accuracy required
2. The extent of variation in the population with regards to key characteristics of the study.
3. The population size.
The sample size also needs to be sufficiently large to allow for meaningful analysis bearing in mind the objective of the study, which was mainly to provide a wide range of indicators which will form benchmark information from which poverty will be monitored over time and space.
Against this background, the sample size for the 1998 Household Poverty Study was set at 2000 households. This was deemed sufficient because it would provide enough cases for subgroup analysis. Two thousand households would also provide sufficient cases given the resource constraints in terms of financing, personnel and time.
SAMPLE SELECTION
In order to have a sample that is representative of the country and to avoid conducting interviews in rural areas with scattered population, cluster sampling procedure was adopted using the existing geographical clusters in the form of Enumeration Areas (EA). Technically, Enumeration Areas are mapped to contain about 500 persons but in reality, they range from 300 to 1000 persons. The EA demarcation covers the whole country and conforms to the administrative boundaries.
Another consideration in the sampling process was the number of households to be selected since this has implications on costs and sampling error. According to Scott (cited in CSD, 1994), a constant take of households per enumeration area has no effect on the sampling error over a Probability Proportional to Size (PPS) technique at the first stage of sampling.
Unlike rural areas where the rich and poor normally live in the same area, the urban population is more residentially homogeneous. In other words, rich people tend to live in certain areas whilst the poor also tend to cluster together. Given the above considerations, a multistage sampling procedure using the PPS technique was adopted. Therefore, 18 households were randomly selected in rural areas against nine in the urban areas.
In summary, a multi-stage sample with probability proportional to size (PPS) was taken. Enumeration areas were stratified into 15 groups based on division and density within divisions. A fraction of these EAs (same as in 1993 Population Census) was selected with PPS and 18 households for rural EAs (or 9 for urban EAs) selected using simple random sampling procedure.
Note: See detailed sampling information in the survey final report which is presented in this documentation.
The sample size of any study depends a to large extent on three key factors:
1. The degree of accuracy required
2. The extent of variation in the population with regards to key characteristics of the study.
3. The population size.
The sample size also needs to be sufficiently large to allow for meaningful analysis bearing in mind the objective of the study, which was mainly to provide a wide range of indicators which will form benchmark information from which poverty will be monitored over time and space.
Against this background, the sample size for the 1998 Household Poverty Study was set at 2000 households. This was deemed sufficient because it would provide enough cases for subgroup analysis. Two thousand households would also provide sufficient cases given the resource constraints in terms of financing, personnel and time.
SAMPLE SELECTION
In order to have a sample that is representative of the country and to avoid conducting interviews in rural areas with scattered population, cluster sampling procedure was adopted using the existing geographical clusters in the form of Enumeration Areas (EA). Technically, Enumeration Areas are mapped to contain about 500 persons but in reality, they range from 300 to 1000 persons. The EA demarcation covers the whole country and conforms to the administrative boundaries.
Another consideration in the sampling process was the number of households to be selected since this has implications on costs and sampling error. According to Scott (cited in CSD, 1994), a constant take of households per enumeration area has no effect on the sampling error over a Probability Proportional to Size (PPS) technique at the first stage of sampling.
Unlike rural areas where the rich and poor normally live in the same area, the urban population is more residentially homogeneous. In other words, rich people tend to live in certain areas whilst the poor also tend to cluster together. Given the above considerations, a multistage sampling procedure using the PPS technique was adopted. Therefore, 18 households were randomly selected in rural areas against nine in the urban areas.
In summary, a multi-stage sample with probability proportional to size (PPS) was taken. Enumeration areas were stratified into 15 groups based on division and density within divisions. A fraction of these EAs (same as in 1993 Population Census) was selected with PPS and 18 households for rural EAs (or 9 for urban EAs) selected using simple random sampling procedure.
Note: See detailed sampling information in the survey final report which is presented in this documentation.
Weighting
The 1993 Enumeration Areas were used as the sampling frame to select the enumeration areas for the survey. Some households were undersampled whilst others were oversampled, therefore there was a need to apply weighting factors both for the household and population in each division. These weighting factors were applied during analysis throughout the report. Table 2.1.3.2 shows the proportion of households and population in the 1993 Population and Housing Census and the 1998 National Household Poverty Survey, Household and Population Weights by Division.
Data Collection
Dates of Data Collection
| Start | End |
|---|---|
| 1998-04-01 | 1998-05-31 |
Data Collection Mode
Face-to-face [f2f]
Supervision
An essential ingredient of any survey plan is the development of quality control systems and procedures, that is, means of assuring that the survey specifications are being carried out satisfactorily. The most important aspect of quality control of this survey is the control over data collection. This included a number of elements such as review of questionnaires by field supervisors, observation of interviews by the Coordinator, re-interviews or second interviews of a selected household. This meant that supervisors checked on enumerators work for missing data, duplicated information, and inconsistent data. The Coordinator visited each team at several points in the field during data collection on a number of occasions for consultation and progress reports. The Coordinator also made a 30 per cent check on each team's work and identified the errors. Enumerators were alerted to rectify and take note of such problems before leaving the EA. Each questionnaire was examined and checked again by a member of the professional staff once it reached head office. Missing or suspected data detected at this point resulted in the return of the questionnaire to the team with a request to call back on the household and obtain or verify the data.
Data Collection Notes
The data was collected from the beginning of April to the end of May covering a period of 48 days consecutively in the field. Data was collected by teams of 5 (1 supervisor and 4 enumerators) in various parts of the country. In order to ensure complete coverage and accessibility of all the survey respondents with minimum fuel consumption, the entire country was divided into five supervision areas as per the following:
TEAM 1: The whole of Banjul, Old Jeshwang, Bakau, Latrikunda, Dippakunda comprises of 39 E.As
TEAM 2: New Jeshwang, Serekunda, Bundungkakunda, Manjaikunda, Bakoteh, Tallinding, Latrikunda Sabiji, Fajikunda, Abuko, Hamdalai(Kerr Sering) Sukuta and Sukuta Sanchaba all of Kanifing administrative area and Kombo North, an area covering 39 EAs.
TEAM 3: Nemakunku, Lamin, Mandinari, and the rest of Kombo North, Kombo South, Kombo Central, Kombo East, Foni Bintang Karanai, Foni Kansala, Foni Bondali, and Foni Jarol. The team covered 26 EAs.
TEAM 4: Lower Niumi, Upper Niumi, Jokadu, Lower Baddibu, Central Baddibu, Lower Saloum, Upper Saloum, Nianija, Niani, and Sami covering a total of 29 EAs.
TEAM 5: Kiang West , Kiang Central, Kiang East, Jarra Weast, Jarra Central, Jarra East, Niamina Dankunku, Niamina West, Niamina East, Fuladu West, Janjanbureh, Fuladu East, Wuli, Sandu consisting of 31 EAs.
In rural areas a field team conducted roughly 17 interviews in 5 EAs (84 interviews) per week. As the team had to conduct two rounds of interviews apart, this means that a team spent roughly one week altogether in a rural EA. The team had no base in the location and were moving around the selected EAs in the provinces. Interviews took place in Mandinka (50 per cent) or some other language, e.g. Sarehuleh (8 per cent). Interpreters were used in 2 per cent of cases. Households were defined as a group of persons/person acknowledging one head and with some sharing of food and budgets. In The Gambian context this meant that most polygamous households were counted as one large household.
Data collection was done in three stages: household listing, part one of the questionnaire and part two.
TEAM 1: The whole of Banjul, Old Jeshwang, Bakau, Latrikunda, Dippakunda comprises of 39 E.As
TEAM 2: New Jeshwang, Serekunda, Bundungkakunda, Manjaikunda, Bakoteh, Tallinding, Latrikunda Sabiji, Fajikunda, Abuko, Hamdalai(Kerr Sering) Sukuta and Sukuta Sanchaba all of Kanifing administrative area and Kombo North, an area covering 39 EAs.
TEAM 3: Nemakunku, Lamin, Mandinari, and the rest of Kombo North, Kombo South, Kombo Central, Kombo East, Foni Bintang Karanai, Foni Kansala, Foni Bondali, and Foni Jarol. The team covered 26 EAs.
TEAM 4: Lower Niumi, Upper Niumi, Jokadu, Lower Baddibu, Central Baddibu, Lower Saloum, Upper Saloum, Nianija, Niani, and Sami covering a total of 29 EAs.
TEAM 5: Kiang West , Kiang Central, Kiang East, Jarra Weast, Jarra Central, Jarra East, Niamina Dankunku, Niamina West, Niamina East, Fuladu West, Janjanbureh, Fuladu East, Wuli, Sandu consisting of 31 EAs.
In rural areas a field team conducted roughly 17 interviews in 5 EAs (84 interviews) per week. As the team had to conduct two rounds of interviews apart, this means that a team spent roughly one week altogether in a rural EA. The team had no base in the location and were moving around the selected EAs in the provinces. Interviews took place in Mandinka (50 per cent) or some other language, e.g. Sarehuleh (8 per cent). Interpreters were used in 2 per cent of cases. Households were defined as a group of persons/person acknowledging one head and with some sharing of food and budgets. In The Gambian context this meant that most polygamous households were counted as one large household.
Data collection was done in three stages: household listing, part one of the questionnaire and part two.
Questionnaires
Questionnaires
The survey was administered using structured questionnaire that consists of two parts. Part one of the questionnaire collected demographic, health, education and crop production, among other information. Part two collected information mostly on household expenditure and anthropometric measures.
Data Processing
Other Processing
DATA PROCESSING
Ten data entry operators were give five days intensive training so that they could understand the questions and responses provided as well as to familiarize themselves with the data entry screen. It was envisaged that data entry would start about a week after the end of field data collection. However, this was not possible owing to numerous callbacks that had to be done. The callbacks were numerous for two reasons:
- The questionnaire was lengthy
- Data collection was at the beginning of the rainy season when most of the rural communities were busy on their farms and had little time to sit through long hours of interview.
Data entry could only start after fieldwork had started and lasted for about three months.
DATA CLEANING AND ANALYSIS
By the third week in September, the computer specialist was able to submit some sections of the data in SPSS format to the data analysis team who engaged in further data cleaning in preparation for analysis. Data cleaning was unduly delayed for many reasons which included the complicated nature of the questionnaire, its length as mentioned earlier, coupled with the data entry operators' inexperience with that kind of questionnaire. It was therefore necessary to re-enter most of the information on the questionnaire for most households.
Data Analysis commenced as soon as cleaning was completed. The analysis plan, which was approved by the User Group, formed the basis for preliminary analysis.
Cognisance of the fact that the ILO provided technical assistance in the setting up of the Poverty Monitoring System (PMS), it was deemed prudent to continue soliciting their input during the conduct and analysis of the 1998 National Household Poverty Study. This was crucial, as the results of the study are to form the basis for poverty monitoring.
Their input was in the following areas:
- Commenting on the draft questionnaires and the analysis plan.
- Reviewing the preliminary analysis.
- Providing a three-week training for in-depth poverty analysis.
- Reviewing and commenting on various drafts of this report.
Constraints
The whole exercise has had some delays due to a number of setbacks:
- The irregular power supply for most of June, July and August 1998 when data entry was supposed to be going on in earnest.
- The Computer specialist had to take about one and a half weeks off in order to attend a seminar in Malta on the Millennium Bug.
Ten data entry operators were give five days intensive training so that they could understand the questions and responses provided as well as to familiarize themselves with the data entry screen. It was envisaged that data entry would start about a week after the end of field data collection. However, this was not possible owing to numerous callbacks that had to be done. The callbacks were numerous for two reasons:
- The questionnaire was lengthy
- Data collection was at the beginning of the rainy season when most of the rural communities were busy on their farms and had little time to sit through long hours of interview.
Data entry could only start after fieldwork had started and lasted for about three months.
DATA CLEANING AND ANALYSIS
By the third week in September, the computer specialist was able to submit some sections of the data in SPSS format to the data analysis team who engaged in further data cleaning in preparation for analysis. Data cleaning was unduly delayed for many reasons which included the complicated nature of the questionnaire, its length as mentioned earlier, coupled with the data entry operators' inexperience with that kind of questionnaire. It was therefore necessary to re-enter most of the information on the questionnaire for most households.
Data Analysis commenced as soon as cleaning was completed. The analysis plan, which was approved by the User Group, formed the basis for preliminary analysis.
Cognisance of the fact that the ILO provided technical assistance in the setting up of the Poverty Monitoring System (PMS), it was deemed prudent to continue soliciting their input during the conduct and analysis of the 1998 National Household Poverty Study. This was crucial, as the results of the study are to form the basis for poverty monitoring.
Their input was in the following areas:
- Commenting on the draft questionnaires and the analysis plan.
- Reviewing the preliminary analysis.
- Providing a three-week training for in-depth poverty analysis.
- Reviewing and commenting on various drafts of this report.
Constraints
The whole exercise has had some delays due to a number of setbacks:
- The irregular power supply for most of June, July and August 1998 when data entry was supposed to be going on in earnest.
- The Computer specialist had to take about one and a half weeks off in order to attend a seminar in Malta on the Millennium Bug.
Access policy
Access authority
| Name | Affiliation | URL | |
|---|---|---|---|
| The Gambia Bureau of Statistics | Department of State for finance and Economic Affairs | gamcens@qanet.gm | http://www.gbos.gm |
Contacts
| Name | Affiliation | URL | |
|---|---|---|---|
| The Gambia Bureau of Statistics | Department of State for Finance and Economic Affairs | info@gbos.gm | http://www.gbos.gm |
Confidentiality
All users must sort approval from GBoS before any dissemination of the data. The data shall not be used for any other purpose than the above specified request.
Access conditions
Data and metadata are provided to you for exclusive use. The data and/or metadata may not be transferred to any other user without prior written authorization from GBoS.
Citation requirements
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 acronym and year of implementation)
- the survey reference number
- the source and date of download
- the Identification of the Primary Investigator
- the title of the survey (including acronym and year of implementation)
- the survey reference number
- the source and date of download
Disclaimer and copyrights
Disclaimer
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.
Metadata production
DDI Document ID
DDI_GMB_1998_HPS_v02_M
Producers
| Name | Abbreviation | Role |
|---|---|---|
| World Bank, Development Economics Data Group | DECDG | Generation of DDI documentation |
Date of Metadata Production
2007-04-19
DDI Document version
Version 1.1 (June 2011) Adopted from "PS1998" Nesstar file that was done by Ivo Njosa.