Living Standards Measurement Survey 2001 (Wave 1 Panel) - Bosnia and Herzegovina

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

In 1992, Bosnia-Herzegovina, one of the six republics in former Yugoslavia, became an independent nation. A civil war started soon thereafter, lasting until 1995 and causing widespread destruction and losses of lives. Following the Dayton accord, BosniaHerzegovina (BiH) emerged as an independent state comprised of two entities, namely, the Federation of Bosnia-Herzegovina (FBiH) and the Republika Srpska (RS), and the district of Brcko. In addition to the destruction caused to the physical infrastructure, there was considerable social disruption and decline in living standards for a large section of the population. Alongside these events, a period of economic transition to a market economy was occurring. The distributive impacts of this transition, both positive and negative, are unknown. In short, while it is clear that welfare levels have changed, there is very little information on poverty and social indicators on which to base policies and programs. In the post-war process of rebuilding the economic and social base of the country, the government has faced the problems created by having little relevant data at the household level. The three statistical organizations in the country (State Agency for Statistics for BiH -BHAS, the RS Institute of Statistics-RSIS, and the FBiH Institute of Statistics-FIS) have been active in working to improve the data available to policy makers: both at the macro and the household level. One facet of their activities is to design and implement a series of household series. The first of these surveys is the Living Standards Measurement Study survey (LSMS). Later surveys will include the Household Budget Survey (an Income and Expenditure Survey) and a Labour Force Survey. A subset of the LSMS households will be re-interviewed in the two years following the LSMS to create a panel data set.

The three statistical organizations began work on the design of the Living Standards Measurement Study Survey (LSMS) in 1999. The purpose of the survey was to collect data needed for assessing the living standards of the population and for providing the key indicators needed for social and economic policy formulation. The survey was to provide data at the country and the entity level and to allow valid comparisons between entities to be made. The LSMS survey was carried out in the Fall of 2001 by the three statistical organizations with financial and technical support from the Department for International Development of the British Government (DfID), United Nations Development Program (UNDP), the Japanese Government, and the World Bank (WB). The creation of a Master Sample for the survey was supported by the Swedish Government through SIDA, the European Commission, the Department for International Development of the British Government and the World Bank. The overall management of the project was carried out by the Steering Board, comprised of the Directors of the RS and FBiH Statistical Institutes, the Management Board of the State Agency for Statistics and representatives from DfID, UNDP and the WB. The day-to-day project activities were carried out by the Survey Management Team, made up of two professionals from each of the three statistical organizations. The Living Standard Measurement Survey LSMS, in addition to collecting the information necessary to obtain a comprehensive as possible measure of the basic dimensions of household living standards, has three basic objectives, as follows: 1. To provide the public sector, government, the business community, scientific institutions, international donor organizations and social organizations with information on different indicators of the population's living conditions, as well as on available resources for satisfying basic needs. 2. To provide information for the evaluation of the results of different forms of government policy and programs developed with the aim to improve the population's living standard. The survey will enable the analysis of the relations between and among different aspects of living standards (housing, consumption, education, health, labour) at a given time, as well as within a household. 3. To provide key contributions for development of government's Poverty Reduction Strategy Paper, based on analysed data.

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

National coverage

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

Households

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

Sample survey data [ssd]

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

(a) SAMPLE SIZE
A total sample of 5,400 households was determined to be adequate for the needs of the survey: with 2,400 in the Republika Srpska and 3,000 in the Federation of BiH. The difficulty was in selecting a probability sample that would be representative of the country's population. The sample design for any survey depends upon the availability of information on the universe of households and individuals in the country. Usually this comes from a census or administrative records. In the case of BiH the most recent census was done in 1991. The data from this census were rendered obsolete due to both the simple passage of time but, more importantly, due to the massive population displacements that occurred during the war. At the initial stages of this project it was decided that a master sample should be constructed. Experts from Statistics Sweden developed the plan for the master sample and provided the procedures for its construction. From this master sample, the households for the LSMS were selected. Master Sample [This section is based on Peter Lynn's note "LSMS Sample Design and Weighting - Summary". April, 2002. Essex University, commissioned by DfID.] The master sample is based on a selection of municipalities and a full enumeration of the selected municipalities. Optimally, one would prefer smaller units (geographic or administrative) than municipalities. However, while it was considered that the population estimates of municipalities were reasonably accurate, this was not the case for smaller geographic or administrative areas. To avoid the error involved in sampling smaller areas with very uncertain population estimates, municipalities were used as the base unit for the master sample. The Statistics Sweden team proposed two options based on this same method, with the only difference being in the number of municipalities included and enumerated.

(b) SAMPLE DESIGN
For reasons of funding, the smaller option proposed by the team was used, or Option B. Stratification of Municipalities The first step in creating the Master Sample was to group the 146 municipalities in the country into three strata- Urban, Rural and Mixed - within each of the two entities. Urban municipalities are those where 65 percent or more of the households are considered to be urban, and rural municipalities are those where the proportion of urban households is below 35 percent. The remaining municipalities were classified as Mixed (Urban and Rural) Municipalities. Brcko was excluded from the sampling frame. Urban, Rural and Mixed Municipalities: It is worth noting that the urban-rural definitions used in BiH are unusual with such large administrative units as municipalities classified as if they were completely homogeneous. Their classification into urban, rural, mixed comes from the 1991 Census which used the predominant type of income of households in the municipality to define the municipality. This definition is imperfect in two ways. First, the distribution of income sources may have changed dramatically from the pre-war times: populations have shifted, large industries have closed, and much agricultural land remains unusable due to the presence of land mines. Second, the definition is not comparable to other countries' where villages, towns and cities are classified by population size into rural or urban or by types of services and infrastructure available. Clearly, the types of communities within a municipality vary substantially in terms of both population and infrastructure. However, these imperfections are not detrimental to the sample design (the urban/rural definition may not be very useful for analysis purposes, but that is a separate issue).

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

Face-to-face [f2f]

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

(a) DATA ENTRY

An integrated approach to data entry and fieldwork was adopted in Bosnia and Herzegovina. Data entry proceeded side by side with data gathering to ensure verification and correction in the field. Data entry stations were located in the regional offices of the entity institutes and were equipped with computers, modem and a dedicated telephone line. The completed questionnaires were delivered to these stations each day for data entry. Twenty data entry operators (10 from Federation and 10 from RS) were trained in two training sessions held for a week each in Sarajevo and Banja Luka. The trainers were the staff of the two entity institutes who had undergone training in the CSPro software earlier and had participated in the workshops of the Pilot survey. Prior to the training, laptop computers were provided to the entity institutes, and the CSPro software was installed in them. The training for the data entry operators covered the following elements:

- Introduction to the LSMS Survey questionnaire; Introduction to the personal computers/ lap top computers; Copying data on diskette and printing of output
- The Data entry programme (CSPro). Understanding of the Round 1 data entry screens (Modules 1-10)
- Practice of Round 1 (data entry trainees enter questionnaires completed by interviewer trainees during practice interviews)
- Understanding of Round 2 Data entry screen (Modules 11-13)
- Practice of Round 2 Data entry screens (data entry trainees entered the questionnaires completed by interviewer trainees)
- Control Procedures; Copying data on diskette and printing lists of errors; Transfer of the data through email to the institutes

The data entry programme was fine-tuned during the training. Some unexpected responses during the interviews had to be accommodated and a few skip patterns fixed. The training emphasized the role of the data entry operator as a member of the survey team, and how the outputs of the programme (error lists) were to be provided to the supervisors and interviewers for necessary correction. The goal was to produce high quality data. Several of the key features of this were:

1. Pre-coded verbatim questionnaires
2. Error detection at the time of data entry
3. Data entry that was concurrent with fieldwork
4. Correction of suspected errors in the field.

(b) QUALITY CHECKS

The following checks were incorporated in the data entry software:
1. Value Range: The program checked to ensure that the values entered were within the valid range for each variable;
2. Reference tables: Where appropriate, the entered data were checked against reference values ( e.g. the price of a kilo of tomato could not exceed 10 KM
3. Skip checks: The program checked that all appropriate skips were followed, both within and between different units of observation;
4. Checks for consistency between different responses: The program checked for internal consistency. For example, whether the age of a person was sufficient for the education level attained, if a filter question for agriculture had a positive response that the module had all relevant information entered and the like.

After the data entry was completed in the field, the data were transferred through email to the central offices in Sarajevo and Banja Luka with the help of PCAnywhere Software. The data entry programme was designed to detect many of the errors even at the stage of data entry, thereby minimizing the need for ex-post facto data editing. Once all data was compiled in the entity offices, a check was made to ensure the structural consistency of data files, i.e. that no records were duplicated or omitted. When the RS and FBiH data files were merged it became apparent that a last minute decision on the treatment of decimal places in several modules had been different in the two entities. Thus, the two data bases were not compatible. A correction was made and data from these modules were re-entered. Once this was done, the data sets were compatible, and a countrywide data set was created. During this process some additional double entry was carried out to correct any data entry operator errors that had occurred. Data Cleaning It is important to note what is meant by 'data cleaning' in terms of the BiH-LSMS data set. In the sense that the data set is a faithful reflection of the responses of all interviewees the data set can be considered 'cleaned'. Every effort was made to ensure that the information provided during the interviews was correctly entered in electronic format.

Response rate
---------------------------

82 percent

Data appraisal
---------------------------

As in any survey, this does not mean that the data set is perfect. As participation in the survey is voluntary, informants had the option to refuse to answer specific questions and may have provided information that is not always consistent. The interviewers resolved as many inconsistencies as possible with the informants but there are, of course, limits. However, given the widely differing needs of the range of analysts who will use the BiHLSMS data, nothing further has been done to the original data. While some data sets are processed so that all missing values are imputed, all outliers revalued and all inconsistencies fixed based on some set of assumptions, this has not been done here. The reason being that there is no correct way to resolve the problems of missing data, outliers and inconsistencies. Each person will need to make his or her own decision on how to treat such data problems based on the type of analysis being carried out. For some analyses, the information in outlier values is key while for others, such outliers would distort findings and would need to be dropped or provided an imputed value. The same for missing values. Some analysts will choose to drop cases with missing values for the variables of interest to them while others will impute such values, using medians, mean or complex multi-variate techniques. In order to ensure the usefulness of the data set for all users, no attempt has been made to impute missing values, reconcile inconsistencies, re-value outliers, or in any way alter the responses provided by the respondents.

摘要
---------------------------
1992年，波斯尼亚和黑塞哥维那，作为前南斯拉夫六个共和国之一，成为了一个独立的国家。在此之后不久，一场内战爆发，持续至1995年，造成了广泛的经济破坏和人员伤亡。在《代顿协定》之后，波斯尼亚和黑塞哥维那（BiH）作为一个由两个实体组成的国家出现，即波斯尼亚和黑塞哥维那联邦（FBiH）和塞尔维亚共和国（RS），以及布里奇克地区。除了对物理基础设施造成的破坏外，还发生了相当大的社会动荡和大量人口生活水平下降。在此事件的同时，国家正经历着向市场经济过渡的经济转型期。这一过渡的分配影响，无论是积极的还是消极的，都是未知的。简而言之，尽管福利水平已经改变，但关于贫困和社会指标的信息很少，这些指标是制定政策和计划的基础。在战后重建国家经济和社会基础的过程中，政府面临着在家庭层面缺乏相关数据的难题。国家的三个统计机构（波斯尼亚和黑塞哥维那国家统计局（BHAS）、塞尔维亚共和国统计局（RSIS）和波斯尼亚和黑塞哥维那联邦统计局（FIS））一直积极工作，以提高政策制定者可用的数据：无论是宏观层面还是家庭层面。他们活动的其中一个方面是设计和实施一系列家庭系列调查。这些调查中的第一个是生活水平测量研究调查（LSMS）。后来的调查将包括家庭预算调查（收入和支出调查）和劳动力调查。LSMS家庭子集将在LSMS之后的两年内进行再次访谈，以创建一个面板数据集。

三个统计机构于1999年开始设计生活水平测量研究调查（LSMS）。调查的目的是收集评估人口生活水平以及制定社会和经济政策所需的关键指标所需的数据。调查旨在提供国家和实体层面上的数据，并允许在实体之间进行有效的比较。LSMS调查于2001年秋季由三个统计机构进行，得到了英国政府国际发展部（DfID）、联合国开发计划署（UNDP）、日本政府和世界银行（WB）的财政和技术支持。调查的总体管理由由RS和FBiH统计局总监、国家统计局管理委员会代表以及来自DfID、UNDP和WB的代表组成的指导委员会负责。日常的项目活动由由每个统计机构的两名专业人员组成的调查管理团队负责。生活水平测量调查LSMS，除了收集获取尽可能全面的家庭生活水平基本维度信息所需的信息外，还具有以下三个基本目标：

1. 向公共部门、政府、商业界、科研机构、国际捐助组织和社会团体提供关于人口生活条件不同指标以及满足基本需求所需资源的情报。
2. 提供信息，以评估旨在提高人口生活水平的不同形式政府政策和计划的结果。该调查将能够分析在特定时间以及家庭内部生活水平（住房、消费、教育、健康、劳动）的不同方面之间的关系。
3. 提供基于分析数据的政府扶贫战略文件的关键贡献。

地理覆盖范围
---------------------------
全国覆盖范围

分析单位
---------------------------
家庭

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

抽样程序
---------------------------
(a) 样本量
确定了一个总样本量5,400户家庭，足以满足调查需求：其中2,400户在塞尔维亚共和国，3,000户在波斯尼亚和黑塞哥维那联邦。困难在于选择一个能够代表国家人口的概率样本。任何调查的样本设计都取决于国家家庭和个人的总体信息。通常这来自于人口普查或行政记录。在波斯尼亚和黑塞哥维那的情况下，最近的普查是在1991年进行的。由于时间的流逝以及更重要的战争期间发生的巨大人口迁移，该普查的数据已经过时。在此项目的初期阶段，决定构建一个主样本。来自瑞典统计学的专家制定了主样本的计划，并提供了其构建的程序。从这个主样本中，选择了LSMS的家庭。主样本 [本部分基于Peter Lynn的笔记“LSMS样本设计和加权 - 摘要”。2002年4月，埃塞克斯大学，由DfID委托。] 主样本基于对市政当局的选择和所选市政当局的全面普查。理想情况下，人们希望使用比市政当局更小的单位（地理或行政单位）。然而，虽然考虑市政当局的人口估计是相当准确的，但较小地理或行政区域的人口估计并不是这样。为了避免使用具有非常不确定的人口估计的较小区域进行抽样的误差，市政当局被用作主样本的基础单位。瑞典统计团队根据这种方法提出了两种选择，唯一的区别在于所包含和列举的市政当局数量。

(b) 样本设计
出于资金原因，团队提出的较小选项被采用，即选项B。市政当局的分层
创建主样本的第一步是将国家的146个市政当局分为三个层 - 城市、乡村和混合 - 在每个两个实体中。城市市政当局是那些65%或更多的家庭被认为是城市的，而乡村市政当局是那些城市家庭比例低于35%的市政当局。其余市政当局被归类为混合（城市和乡村）市政当局。布里奇克被排除在抽样框架之外。城市、乡村和混合市政当局：值得注意的是，在波斯尼亚和黑塞哥维那使用的城市-乡村定义是不寻常的，因为像市政当局这样大的行政单位被归类为完全同质的。它们的分类为城市、乡村、混合，来自1991年的人口普查，该普查使用市政当局中家庭主要收入来源来定义市政当局。这种定义在两个方面是不完美的。首先，收入来源的分布可能已经发生了巨大的变化：人口已经转移，大型工业已经关闭，由于地雷的存在，大量农业用地仍然无法使用。其次，这种定义与其他国家不同，其他国家根据人口规模将村庄、城镇和城市分类为乡村或城市，或者根据可用的服务和基础设施类型进行分类。显然，市政当局内的社区类型在人口和基础设施方面都存在很大的差异。然而，这些不完美并不损害样本设计（城市/乡村的定义可能对分析目的不是很有用，但这是一个单独的问题）。

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

清理操作
---------------------------
(a) 数据录入
在波斯尼亚和黑塞哥维那采用了数据录入和实地工作的一体化方法。数据录入与数据收集同时进行，以确保在现场进行验证和纠正。数据录入站位于实体研究所的区域办事处，并配备了计算机、调制解调器和专用电话线。每天完成的问卷被送到这些站点进行数据录入。二十名数据录入操作员（来自联邦的10名和来自RS的10名）在萨拉热窝和巴尼亚卢卡举行的为期一周的两次培训中接受了培训。培训师是两个实体研究所的员工，他们之前接受了CSPro软件的培训，并参加了试点调查的工作坊。在培训之前，实体研究所提供了笔记本电脑，并在其中安装了CSPro软件。数据录入操作员的培训涵盖了以下要素：

- LSMS调查问卷简介；个人计算机/笔记本电脑简介；磁盘上的数据复制和输出打印
- 数据录入程序（CSPro）。理解第一轮数据录入屏幕（模块1-10）
- 第一轮（数据录入实习生在实践访谈中录入由访谈实习生完成的问卷）的实践
- 理解第二轮数据录入屏幕（模块11-13）
- 第二轮数据录入屏幕的实践（数据录入实习生录入了由访谈实习生完成的问卷）
- 控制程序；在磁盘上复制数据并打印错误列表；通过电子邮件将数据传输到研究所

在培训期间，对数据录入程序进行了微调。在访谈期间出现了一些意外的响应，需要适应，并修复了几个跳过模式。培训强调数据录入操作员作为调查团队一员的作用，以及如何将程序的输出（错误列表）提供给监督员和访谈员进行必要的纠正。目标是产生高质量的数据。以下是一些关键特征：

1. 预编码的原始问卷
2. 数据录入时的错误检测
3. 与实地工作同时进行的数据录入
4. 在实地中纠正可疑错误

(b) 质量检查
在数据录入软件中包含了以下检查：
1. 值范围：程序检查输入的值是否在每个变量的有效范围内；
2. 参考表：在适当的情况下，输入的数据被检查与参考值（例如，一公斤西红柿的价格不能超过10 KM）；
3. 跳过检查：程序检查是否遵循了所有适当的跳过，无论是在不同的观察单位之间还是在不同的观察单位内部；
4. 不同响应之间的一致性检查：程序检查内部一致性。例如，一个人的年龄是否足够达到其教育水平，如果农业筛选问题有肯定的响应，则该模块已输入所有相关信息，等等。

在实地完成数据录入后，数据通过PCAnywhere软件的帮助传输到萨拉热窝和巴尼亚卢卡的中心办事处。数据录入程序被设计成在数据录入阶段就检测到许多错误，从而最大限度地减少事后数据编辑的需求。一旦所有数据在实体办事处汇总，就会进行检查以确保数据文件的结构一致性，即没有记录被重复或遗漏。当RS和FBiH数据文件合并时，很明显，在两个实体中对几个模块的小数点处理的决定在最后时刻是不同的。因此，这两个数据库不兼容。进行了纠正，并重新输入了这些模块的数据。完成此操作后，数据集兼容，并创建了一个全国数据集。在此过程中，进行了一些额外的双重录入，以纠正任何数据录入员可能发生的错误。数据清理
值得注意的是，在波斯尼亚和黑塞哥维那-LSMS数据集中，“数据清理”的含义是什么。从数据集忠实反映所有受访者响应的角度来看，该数据集可以被认为是“清理过的”。已尽一切努力确保在访谈中提供的信息被正确地以电子格式录入。

响应率
---------------------------
82百分比

数据评估
---------------------------
与任何调查一样，这并不意味着数据集是完美的。由于调查是自愿的，受访者有选择拒绝回答具体问题的权利，并且可能提供了不总是一致的情报。访谈员尽可能地与受访者解决了不一致之处，但当然有限。然而，鉴于广泛不同的分析人士对BiH-LSMS数据的需要，对原始数据没有进行任何进一步的修改。虽然一些数据集被处理，以便所有缺失值都进行插补，所有异常值重新评估，所有不一致之处都根据某些假设进行调整，但这在这里并没有发生。原因是没有正确解决缺失数据、异常值和不一致之处的问题的方法。每个人都需要根据自己的分析类型自行决定如何处理此类数据问题。对于某些分析，异常值中的信息是关键，而对于其他分析，此类异常值会扭曲结果，需要被删除或提供插补值。同样，对于缺失值。一些分析人士会选择删除与他们感兴趣的变量相关的缺失值的案例，而其他人则将使用中位数、平均值或复杂的多元技术来插补这些值。为了确保数据集对所有用户的有用性，没有试图插补缺失值、解决不一致之处、重新评估异常值或以任何方式改变受访者的响应。