Longitudinal Socioeconomic Study of Refugees and Host Communities 2022-2023 - Kenya

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

The World Bank in collaboration with the United Nations High Commission for Refugees (UNHCR) and the University of California-Berkeley conducted a cross-sectional survey on refugee and host populations living in camp (Dadaab, Kakuma and Kalobeyei) and urban (Nairobi, Nakuru and Mombasa) locations. This dataset contains information from wave 1 which is part of a panel survey that targets refugees and their hosts communities. Enumerators were trained in a participatory approach where they learned how to collect data specific for refugees as well as nationals. Data quality monitoring was conducted daily during data collection to provide feedback to the field team and correct any arising errors. Data was collected with computer Aided Personal Interviews (CAPI) technique through the Ipsos iField software.

The sample is representative of camp based (Dadaab, Kakuma and Kalobeyei) refugees and their host communities in Turkana and Garissa; and urban based (Nairobi, Mombasa and Nakuru) refugees and their host communities. Refugees were sampled from two frames, the UNHCR registration database proGress and the Socioeconomic Surveys (SES Kakuma, Kalobeyei and Nairobi) in two stages. An overlapping frames correction was used to account for observations with a positive probability of being selected from either frame. Hosts were sampled using a two-step cluster design, whereby a set of Enumeration Areas (EAs) and replacement EAs was drawn with probability proportional to size (PPS) from the universe of eligible EAs. Subsequently, a listing exercise was carried out to eliminate ineligible households including households where one member was registered with UNHCR as a refugee or asylum seeker. In the second stage, 10 households and a list of 5 replacements for the case of non-response was drawn for each selected EA using simple random sampling.

Targeted modules on women empowerment and child education were administered to women aged 15 and above and 5th grade children, respectively. A part of the education module was also answered to the child’s caregiver (adult 18 years and above). Unlike other modules, the education module was only implemented in the Kakuma, Kalobeyei, and Dadaab refugee strata and the Turkana and Dadaab host community strata. Targeted modules were also administered to a ‘representative respondent’, a randomly drawn adult from the household, covering employment, migration, savings and mental health.

The dataset contains four files. hhm.dta contains data at the level of the individual for all household members. Each household member is uniquely identified by ‘hhid’ and ‘hhmemid’. hr.dta contains household level information. rr.dta contains data at the representative respondent level. wer.dta contains data for the women empowerment module. The ‘hhid’ variable uniquely identifies all households.

This survey was conducted between May 2022 to June 2023. It comprises a sample of 9,304 households in total, of which 5,840 are refugees and 3,464 are hosts.

Further information on the Kenya Analytical Program on Forced Displacement can be found on the following dashboard: https://www.worldbank.org/en/data/interactive/2024/06/18/dashboard-kenya

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

Garissa, Turkana, Nairobi, Nakuru and Mombasa counties, Kenya

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

Household, individual.

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

Sample survey data [ssd]

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

The survey has two primary samples contained in the ‘status’ variable: the refugee sample and the host community sample. Refugees were sampled from two frames, the UNHCR registration database proGres and the Socioeconomic Surveys (SES) conducted in Kakuma, Kalobeyei and Nairobi. ProGres holds information on all refugee and asylum seekers families in Kenya registered with UNHCR including their phone numbers and the approximate location, data on nationality. The second frame consists of the samples of the Kalobeyei SES (2018), Kakuma Refugee Camp SES (2019) and Urban SES in Nairobi (2020-21). It holds rich socioeconomic baseline data and is a subset of the first. In the first stage, a subsample of proGres families is drawn from the entire proGres frame using implicit stratification by sub-county and nationality. Notably, proGres families are not always the same as households defined in standard surveys. For K-LSRH, a household is defined as a “person or group of people living in the same compound (fenced or unfenced); answerable to the same head and sharing a common source of food/share cooking arrangement.” Meanwhile, proGres families refer to the group of people in which refugees are registered to UNHCR. A household can comprise of multiple proGres families and the selection probability has been adjusted accordingly.

In the second stage, an independent subsample was drawn from the SES frame for each of the strata with the existing SES. Households who had arrived in Kenya before the SES and who had not left the country, represented an overlap group which has a positive probability of being selected into the sample through both the SES frame and through proGres directly. For this subset of households, weights were adjusted using an overlapping frames correction.

Hosts were sampled using a two-step cluster design. The sampling frame of host households consists of communities living in close proximity to refugees who are potentially impacted by their presence. The host community of camp refugees in Turkana and Dadaab are defined as those living within a 15 km radius of the camp borders, while host communities of urban refugees are defined as neighborhoods hosting at least 10 percent of proGres families living in each city of the stratum including; Kayole, Eastleigh North, and Kasarani in Nairobi, Old Town and Majengo in Mombasa, and Lanet, Shabaab, and Langalanga in Nakuru. Sampling follows a two-step clustered design, whereby a set of Enumeration Areas (EAs) and replacement EAs is drawn with probability proportional to size (PPS) from the universe of eligible EAs. Subsequently, a listing exercise was carried out to eliminate ineligible households including households not living within the boundary of selected EAs and households not providing consent to the listing interview. Households where one member was registered with UNHCR as a refugee or asylum seeker were also not interviewed, to rule out overlap with the refugee sampling frame. In the second stage, for each selected EA, 10 households and a list of 5 replacements for the case of non-response was drawn using simple random sampling. This resulted in a sample of 2,000 households for the host community of camp refugees and of 1,500 households for the host community of urban refugees.

A representative respondent (RR) was randomly selected from among the household members aged 18 and above and administered a set of in-depth modules on employment and individual-level characteristics. In addition, the survey included a module on women’s empowerment, administered by trained female enumerators to randomly selected female respondents aged 15 and above. If the RR was a woman, the same respondent also completed the woman empowerment module; otherwise, a woman aged 15 years and above was selected among the women in the household based on simple random sampling. Finally, a module on child education and aspirations was administered to a cohort of upper primary school children and their main caregivers. One Child Respondent (CR) was randomly selected among 5th-grade students currently attending school of any age up to and including 17 years (to also capture over-age students) in selected households. The child’s main caregiver (CG) was also interviewed to capture parental aspirations and main barriers to education. The Caregiver refers to the household member responsible for making the educational decisions for the child. Unlike other modules, the Education Module was only implemented in the Kakuma, Kalobeyei, and Dadaab refugee strata and the Turkana and Dadaab host community strata.

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

Computer Assisted Personal Interview [capi]

Research instrument
---------------------------

Targeted modules on women empowerment and child education were administered to women aged 15 and above and 5th grade children, respectively. A part of the education module was also answered to the child’s caregiver (adult 18 years and above). Unlike other modules, the education module was only implemented in the Kakuma, Kalobeyei, and Dadaab refugee strata and the Turkana and Dadaab host community strata. Targeted modules were also administered to a ‘representative respondent’, a randomly drawn adult from the household, covering employment, migration, savings and mental health.

The Questionnaire is provided as external resources in pdf format. Questionnaires were produced through the Ipsos developed iField software.

摘要
---------------------------
世界银行与联合国难民署（UNHCR）及加州大学伯克利分校合作，对生活在营地（达达布、卡库马和卡洛贝伊）及城市（内罗毕、纳库鲁和蒙巴萨）地区的难民及东道国人口进行了横断面调查。本数据集包含第一波次的数据，该波次为针对难民及其东道国社区的面板调查的一部分。调查员在参与式方法下接受了培训，学习了如何收集针对难民的特定数据以及国民的数据。在数据收集过程中，每日进行数据质量监控，以向现场团队提供反馈并纠正任何出现的错误。数据通过计算机辅助个人访谈（CAPI）技术，利用Ipsos iField软件进行收集。

样本代表了生活在营地（达达布、卡库马和卡洛贝伊）的难民及其东道国社区，以及生活在城市（内罗毕、蒙巴萨和纳库鲁）的难民及其东道国社区。难民从两个框架中抽取样本，即联合国难民署注册数据库proGress和社会经济调查（SES Kakuma、Kalobeyei和内罗毕）。ProGress包含所有在肯尼亚注册的难民和寻求庇护者的家庭信息，包括他们的电话号码和近似位置，以及国籍数据。第二个框架包括Kalobeyei SES（2018年）、Kakuma难民营SES（2019年）和内罗毕城市SES（2020-21年）的样本。它包含了丰富的社会经济基线数据，是第一个框架的子集。在第一阶段，从整个proGress框架中抽取了一个子样本，通过次县和国籍进行隐式分层。值得注意的是，proGress家庭并不总是等同于标准调查中定义的家庭。对于K-LSRH，家庭被定义为“居住在同一院落（有围栏或无围栏）中、对同一户主负责并共享共同食物来源/烹饪安排的个人或人群。”而proGress家庭指的是那些在联合国难民署注册的难民群体。一个家庭可以包含多个proGress家庭，且选择概率已相应调整。

在第二阶段，从每个具有现有SES的层中独立抽取了SES框架的子样本。在SES之前抵达肯尼亚且未离开国家的家庭构成了一个重叠组，该组有正概率通过SES框架和通过proGress直接被选入样本。对于这个子集的家庭，使用重叠框架校正调整了权重。

东道国人口的抽样采用两阶段聚类设计。东道国家庭的抽样框架由靠近难民的社区组成，这些社区可能受到难民存在的影响。在 Turkana 和 Dadaab 的难民营东道国社区被定义为生活在营地边界15公里范围内的居民，而城市难民东道国社区被定义为在每个层级的城市中至少有10%的proGress家庭居住的社区，包括内罗毕的Kayole、Eastleigh North和Kasarani，蒙巴萨的Old Town和Majengo，以及纳库鲁的Lanet、Shabaab和Langalanga。抽样遵循两阶段聚类设计，其中从合格的EA总体中按规模成比例（PPS）抽取一组枚举区域（EA）和替代EA。随后，进行了一次清单作业，以排除不合格的家庭，包括不在选定EA边界内居住的家庭以及未提供清单访谈同意的家庭。那些在联合国难民署注册为难民或寻求庇护者的家庭成员也没有接受访谈，以排除与难民抽样框架的重叠。在第二阶段，对于每个选定的EA，通过简单随机抽样抽取了10个家庭和5个备选名单以应对非响应情况。这导致了难民营东道国社区的样本量为2000个家庭，城市难民东道国社区的样本量为1500个家庭。

从18岁以上的家庭成员中随机选择了一位代表受访者（RR），并对其进行了就业和个人层面特征的深度模块调查。此外，调查还包括一个关于女性赋权的模块，由经过培训的女性调查员对随机选择的15岁及以上的女性受访者进行。如果RR是女性，则相同的受访者还完成了女性赋权模块；否则，在家庭中的女性中根据简单随机抽样选择了一位15岁及以上的女性。最后，对上小学高年级的学生及其主要监护人进行了关于儿童教育和抱负的模块调查。在选定家庭中，随机选择了一位目前就读于任何年龄的5年级学生（包括17岁及以下的学生，以捕捉超龄学生）作为儿童受访者（CR）。该儿童的主要监护人（CG）也被访谈，以了解父母的抱负和主要教育障碍。监护人指的是负责为儿童做出教育决策的家庭成员。与其他模块不同，教育模块仅在卡库马、卡洛贝伊、达达布难民层以及Turkana和达达布东道国社区层实施。

数据集包含四个文件。hhm.dta包含所有家庭成员的个体层级数据。每个家庭成员通过‘hhid’和‘hhmemid’唯一标识。hr.dta包含家庭层级信息。rr.dta包含代表受访者层级的数据。wer.dta包含女性赋权模块的数据。‘hhid’变量唯一标识所有家庭。

本调查于2022年5月至2023年6月进行。共包含9304个家庭的样本，其中5840个是难民，3464个是东道国人口。

有关肯尼亚强制位移分析计划的更多信息，可在以下仪表板上找到：https://www.worldbank.org/en/data/interactive/2024/06/18/dashboard-kenya

地理覆盖范围
---------------------------
加里萨、图尔卡纳、内罗毕、纳库鲁和蒙巴萨县，肯尼亚

分析单元
---------------------------
家庭、个人。

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

抽样程序
---------------------------
调查包含两个主要样本，包含在‘status’变量中：难民样本和东道国社区样本。难民从两个框架中抽取样本，即联合国难民署注册数据库proGress和在社会经济调查（SES Kakuma、Kalobeyei和内罗毕）中进行的调查。ProGress包含所有在肯尼亚注册的难民和寻求庇护者的家庭信息，包括他们的电话号码和近似位置，以及国籍数据。第二个框架包括Kalobeyei SES（2018年）、Kakuma难民营SES（2019年）和内罗毕城市SES（2020-21年）的样本。它包含了丰富的社会经济基线数据，是第一个框架的子集。在第一阶段，从整个proGress框架中抽取了一个子样本，通过次县和国籍进行隐式分层。值得注意的是，proGress家庭并不总是等同于标准调查中定义的家庭。对于K-LSRH，家庭被定义为“居住在同一院落（有围栏或无围栏）中、对同一户主负责并共享共同食物来源/烹饪安排的个人或人群。”而proGress家庭指的是那些在联合国难民署注册的难民群体。一个家庭可以包含多个proGress家庭，且选择概率已相应调整。

在第二阶段，从每个具有现有SES的层中独立抽取了SES框架的子样本。在SES之前抵达肯尼亚且未离开国家的家庭构成了一个重叠组，该组有正概率通过SES框架和通过proGress直接被选入样本。对于这个子集的家庭，使用重叠框架校正调整了权重。

东道国人口的抽样采用两阶段聚类设计。东道国家庭的抽样框架由靠近难民的社区组成，这些社区可能受到难民存在的影响。在 Turkana 和 Dadaab 的难民营东道国社区被定义为生活在营地边界15公里范围内的居民，而城市难民东道国社区被定义为在每个层级的城市中至少有10%的proGress家庭居住的社区，包括内罗毕的Kayole、Eastleigh North和Kasarani，蒙巴萨的Old Town和Majengo，以及纳库鲁的Lanet、Shabaab和Langalanga。抽样遵循两阶段聚类设计，其中从合格的EA总体中按规模成比例（PPS）抽取一组枚举区域（EA）和替代EA。随后，进行了一次清单作业，以排除不合格的家庭，包括不在选定EA边界内居住的家庭以及未提供清单访谈同意的家庭。那些在联合国难民署注册为难民或寻求庇护者的家庭成员也没有接受访谈，以排除与难民抽样框架的重叠。在第二阶段，对于每个选定的EA，通过简单随机抽样抽取了10个家庭和5个备选名单以应对非响应情况。这导致了难民营东道国社区的样本量为2000个家庭，城市难民东道国社区的样本量为1500个家庭。

从18岁以上的家庭成员中随机选择了一位代表受访者（RR），并对其进行了就业和个人层面特征的深度模块调查。此外，调查还包括一个关于女性赋权的模块，由经过培训的女性调查员对随机选择的15岁及以上的女性受访者进行。如果RR是女性，则相同的受访者还完成了女性赋权模块；否则，在家庭中的女性中根据简单随机抽样选择了一位15岁及以上的女性。最后，对上小学高年级的学生及其主要监护人进行了关于儿童教育和抱负的模块调查。在选定家庭中，随机选择了一位目前就读于任何年龄的5年级学生（包括17岁及以下的学生，以捕捉超龄学生）作为儿童受访者（CR）。该儿童的主要监护人（CG）也被访谈，以了解父母的抱负和主要教育障碍。监护人指的是负责为儿童做出教育决策的家庭成员。与其他模块不同，教育模块仅在卡库马、卡洛贝伊、达达布难民层以及Turkana和达达布东道国社区层实施。

数据收集方式
---------------------------
计算机辅助个人访谈 [capi]

研究工具
---------------------------
针对女性赋权和儿童教育分别实施了针对15岁及以上女性和5年级儿童的针对性模块。教育模块的一部分也由儿童的主要监护人（成人18岁及以上）回答。与其他模块不同，教育模块仅在卡库马、卡洛贝伊、达达布难民层以及Turkana和达达布东道国社区层实施。针对‘代表受访者’的针对性模块也实施了，该受访者是从家庭中随机抽取的成年人，涵盖就业、移民、储蓄和心理健康。问卷以PDF格式作为外部资源提供。问卷是通过Ipsos开发的iField软件制作的。