Integrated Household Survey 2003-2004 - Sierra Leone

Abstract
---------------------------

Sierra Leone is a country recovering from years of conflict. There has been a great deal of effort placed on rebuilding the statistical institutions and collecting benchmark data for monitoring and evaluating poverty reduction policies.

The current household survey provides the data establishing this benchmark using standard household survey questionnaires and provides important information for defining money metric poverty measures. Among the uses of the data, have been: computation of poverty statistics for the PRSP (Poverty Reduction Strategy Paper), fills in gaps in the National Accounts system, and re-base the consumption basket for the CPI (Consumer Price Index). The data will also help in identifying primary target groups for government assistance and forms the basis for rebuilding the capacity of Statistics Sierra Leone (SSL).

The specific objectives of the survey are:

1.To provide information on the patterns of household consumption and expenditure disaggregated to the district level.

2. To provide a database for national and regional planning and poverty monitoring.

3. To produce weights required for the revision or construction of the Consumer Price Indices;

4. To provide statistics for the PRSP.

5. To Improve on the estimation of national income.

6. To give an in-depth view into the structure and distribution of wages and conditions of work of the country's labour force, regions, occupations and gender.

Geographic coverage
---------------------------

This was a national survey representative to the district level. The country is divided into the Northern, Southern, and Eastern Provinces and the Western Area. Each Province/Area is divided into Districts/Wards. There are 14 strata. These include 13 districts and Freetown.

Analysis unit
---------------------------

- Households
- Individuals
- Farm Plots
- Consumption Expenditure Items

Universe
---------------------------

This survey was a standard type household survey. In the survey, a household has been defined as a group of related or unrelated people, who live in a dwelling unit or its equivalent, take their meals together and eating from the same pot, and share common housekeeping arrangements for at least 9 months preceding the date of the interview.

Kind of data
---------------------------

Sample survey data [ssd]

Sampling procedure
---------------------------

The target sample size selected for the 2003/2004 Sierra Leone Integrated Household Survey was 3,700 households, of which 2400 were households drawn from the Rural Areas. In the survey, a household has been defined as a group of related or unrelated people, who live in a dwelling unit or its equivalent, take their meals together and eating from the same pot, and share common housekeeping arrangements for at least 9 months preceding the date of the interview.

The primary source of the Sampling Frame was the 1985 Population Census of Sierra Leone. The country is divided into the Northern, Southern, and Eastern Provinces and the Western Area. Each Province/Area is divided into Districts/Wards. The Districts are further divided into Chiefdoms, Towns and Villages hierarchically. The Chiefdoms and Towns are divided into Census Enumeration Areas (EAs), which formed the Primary Sampling Unit, from which information on the number of households and population in each EA was derived. Each District or Ward is stratified into Rural and Urban Areas. The Ultimate Sampling Unit was the Household.

The overall sampling frame was stratified into two strata, urban and rural, with sampling carried out separately in each stratum. The Enumeration Area was the Primary Sampling Unit with an average population of 1,645. Based on revised population estimates, about 66% of Sierra Leoneans households live in the rural areas so the sample stratification ensured that 66% of selected households came from the rural areas. In the selection of the 226 Enumeration Areas from the 2,553 Enumeration Areas for the Listing of Households, 66 Urban EAs and 160 Rural EAs were selected systematically with a random start and fixed interval of 11. The selection of the EAs within the Urban and Rural list was done to ensure that they were spread out evenly among all the Provinces or Regions in Sierra Leone.

The sample for the Sierra Leone Integrated Household Survey was drawn in two stages with the Enumeration Areas chosen as the Primary Sampling Unit (PSU). A total of 226 EAs were selected systematically out of the 2,538 EAs without probability proportional to their population size. The second stage involved the listing of all households within each selected EA and randomly selecting 1,300 households in the Urban Area (20 Households per Urban EA) and 2,400 households in the Rural Area (15 Households per Rural EA) using a table of Random Numbers. For each Enumeration Area, eight substitute Households were selected.

The existing Sampling Frame is based on the 1985 Population Census. Sierra Leone is divided into the Southern, Eastern, Northern Province and the Western Area with each divided into Districts / Ward. The Districts / Wards are divided into Census Enumeration Areas from which information on the number of Households are obtained. Each District is stratified into Rural and Urban.

The 2,553 Enumeration Areas (EAs), constituting 1,320 Urban and 2,400 Rural EAs was created out of the 1985 Population Census and used as the Sampling Frame for the Sierra Leone Integrated Household Survey. The Sampling Design adopted for the Survey was a 2 - Stage Stratified Sampling. Stage 1 constituted 226 Enumeration Areas selected from the 2553 Enumeration Areas. In Stage 1, the Sampling of 2553 / 266 = approximately 11 was determined and a random number between 1 and 11 inclusive selected. Starting with that randomly selected number we systematically select every 11th EA to give the required 226 EAs. The Sample was not selected with Probability Proportional to Population Size since the population figures are based on the 1985 Census and out of date.

The List of Households and Maps existing at the start of the Survey were derived from the 1985 Population Census. These Frames are more than 17 years old and significant changes have taken place, especially in the Urban Areas, because of population growth, migration and other factors. Since to update the Frame and Maps for all the EAs will be too expensive, Listings of all Households in the 226 EAs was done in order to control both Sampling and Non-Sampling Errors. A Household is defined as a group of related and unrelated persons living in a dwelling unit, eating from the same pot and sharing the common housekeeping arrangement. The Household member, with some exceptions, is a normal resident of the house who spent more than nine months in the household during the past 12 months.

Stage 2 constituted the selection of 20 Households from each of the 66 Urban EA to give 1,320 Urban Households, and 15 Households from each of the 160 Rural EAs to give 2,400 Rural Households. Thus, the total number of Households drawn for the Survey, using 2-Digit and 3-Digit Tables of Random Numbers, was 3720.

Mode of data collection
---------------------------

Face-to-face [f2f]

Cleaning operations
---------------------------

The data were cleaned in various stages. An initial stage of consistency edits were applied using CONCOR edits (IMPS module). A second stage of consistency edits were designed in CSPro. This included the treatment of outliers and range checks using various methods (as the data indicated) for most of the expenditure data. See Process Manual (External Resources) for more information.

摘要
---------------------------
塞拉利昂是一个正在从多年冲突中恢复的国家。大量努力已经投入到重建统计机构和收集基准数据，以监控和评估减贫政策。当前的家庭调查通过标准的家庭调查问卷建立这一基准数据，并为定义货币度量贫困提供了重要信息。数据的使用包括：计算减贫战略文件（Poverty Reduction Strategy Paper）的贫困统计数据，填补国民账户系统中的空白，以及重新编制消费者价格指数（CPI）的消费篮子。数据还将有助于确定政府援助的主要目标群体，并构成重建塞拉利昂统计局（Statistics Sierra Leone，简称SSL）能力的基石。

调查的具体目标包括：

1. 提供关于家庭消费和支出模式的信息，这些信息按区级进行细分。
2. 为国家和区域规划以及贫困监控提供数据库。
3. 生成修订或构建消费者价格指数所需的权重。
4. 为减贫战略文件提供统计数据。
5. 改进国民收入的估计。
6. 对国家劳动力、地区、职业和性别的工资结构和分布提供深入的了解。

地理覆盖范围
---------------------------
这是一项以区级为代表性的全国性调查。国家分为北部、南部和东部省份以及西部地区。每个省份/地区分为区/街道。共有14个层。这包括13个区和弗里敦。

分析单位
---------------------------
- 家庭
- 个人
- 农场地块
- 消费支出项目

总体
---------------------------
本调查是一种标准类型的家庭调查。在调查中，家庭被定义为至少在采访日期前9个月内，在住宅单位或其等效住宅中居住、共同用餐、使用同一锅进食，并共享共同家务安排的相关或无关的人群。

数据类型
---------------------------
样本调查数据 [ssd]

抽样程序
---------------------------
2003/2004年塞拉利昂综合家庭调查的目标样本量为3,700户，其中2,400户来自农村地区。在调查中，家庭被定义为相关或无关的人群，他们居住在住宅单位或其等效住宅中，共同用餐并使用同一锅进食，并且在采访日期前至少9个月共享共同的家务安排。

抽样框架的主要来源是1985年塞拉利昂人口普查。国家分为北部、南部和东部省份以及西部地区。每个省份/地区分为区/街道。区进一步分为宗族、城镇和村庄，形成层次结构。宗族和城镇分为人口普查登记区（EAs），形成初级抽样单位，从中获取每个EAs中家庭和人口数量信息。每个区或街道被划分为农村和城市地区。最终抽样单位是家庭。

总体抽样框架分为两个层，城市和农村，在每个层中进行单独抽样。人口普查登记区是初级抽样单位，平均人口为1,645。根据修订后的人口估计，约66%的塞拉利昂家庭居住在农村地区，因此样本分层确保了66%的选样家庭来自农村地区。在从2,553个人口普查登记区中选择226个登记区进行家庭列表时，系统性地选择了66个城市EAs和160个农村EAs，随机开始，间隔为11。在城乡名单中选择EAs的目的是确保它们在塞拉利昂的所有省份或地区中均匀分布。

塞拉利昂综合家庭调查的样本通过两个阶段进行抽取，登记区被选为初级抽样单位（PSU）。共选择了226个登记区中的226个，没有任何与人口规模成比例的概率。第二阶段涉及在每个选定的登记区中列出所有家庭，并使用随机数字表随机选择城市地区的1,300户家庭（每个城市EAs 20户）和农村地区的2,400户家庭（每个农村EAs 15户）。对于每个登记区，选择了8个替代家庭。

现有的抽样框架基于1985年人口普查。塞拉利昂分为南部、东部、北部省份和西部地区，每个地区划分为区/街道。区/街道划分为人口普查登记区，从中获取家庭数量信息。每个区被划分为农村和城市。

由1985年人口普查创建的2,553个人口普查登记区（EAs），包括1,320个城市和2,400个农村EAs，被用作塞拉利昂综合家庭调查的抽样框架。调查采用的抽样设计是两阶段分层抽样。第一阶段包括从2,553个登记区中选择226个登记区。在第一阶段中，确定了2553/266 = 大约11的抽样量，并随机选择1到11之间的一个数字。从随机选择的那个数字开始，系统性地选择每个第11个登记区，以给出所需的226个登记区。样本不是按与人口规模成比例的概率进行选择的，因为人口数字是基于1985年人口普查的，并且已经过时。

调查开始时存在的家庭列表和地图来自1985年人口普查。这些框架已有超过17年的历史，由于人口增长、迁移等因素，特别是城市地区发生了显著变化。由于更新所有EAs的框架和地图将过于昂贵，因此在226个EAs中对所有家庭进行了列表，以控制抽样和非抽样误差。家庭被定义为居住在住宅单位中、使用同一锅进食并共享共同家务安排的相关和无关人员。家庭成员（除某些例外外）是在过去12个月中在家庭中居住超过9个月的正常居民。

第二阶段包括从每个66个城市EAs中选择20户家庭，以给出1,320户城市家庭，以及从每个160个农村EAs中选择15户家庭，以给出2,400户农村家庭。因此，使用两位数和三位数随机数字表抽取的样本总数为3,720户。

数据收集方式
---------------------------
面对面 [f2f]

数据清洗操作
---------------------------
数据在多个阶段进行了清洗。最初阶段应用了一致性编辑，使用CONCOR编辑（IMPS模块）。第二阶段设计了一致性编辑，在CSPro中进行。这包括使用各种方法（如数据所示）处理异常值和范围检查，用于大部分支出数据。有关更多信息，请参阅流程手册（外部资源）。