Transnational Perspectives on Migration and Integration – TRANSMIT Surveys of Syrian Nationals and Their Neighbors in Lebanon and Turkey (2019–2023)

2.1 Baseline Sampling Lebanon

The initial survey in Lebanon took place in 2019. The number of respondents was 1,252 (see Table 1). As there are no publicly accessible population registers in Lebanon, respondents were selected using area sampling combined with the random walk method. This is a common procedure for hard-to-reach population samples (Bauer 2014; Thomson et al. 2020). For this purpose, five hierarchical lists were created on five geographical levels (governorates, zones within the governorates, sub-areas within the zones, clusters within the sub-areas, and neighborhoods within the clusters). In order to create a sample that is as representative as possible, information on the demographic and socio-demographic composition of the population (religious affiliation, socio-economic status), as well as the degree of urbanization and geographical characteristics, were considered when selecting the sub-areas. From the entirety of the neighborhoods, 200 neighborhoods were then randomly selected as Primary Sampling Units (PSU). In order to achieve the target size of the Syrian sample while remaining economically viable, only neighborhoods with an above-average proportion of Syrians are included in this selection. For this purpose, data from previous surveys and the field expertise of our implementation partners were used. Within the selected neighborhoods, Syrian and host households were recruited in equal numbers via random walk, i.e., interviewers walked the streets in the selected neighborhood according to predetermined walking rules to select households for participation. Within the households that agreed to participate the choice of the interviewee was also randomized (within age requirements). If the random walk in the respective neighborhoods did not yield sufficient observations, the survey continued in previously defined alternative neighborhoods until the respective quota of Syrian and host respondents was met.

2.2 Baseline Sampling Turkey

The baseline survey in Turkey took place in 2020. The number of respondents was 3,045. As in Lebanon, this survey was also stratified according to the country of birth of the head of household, so that 50 percent of the respondents came from Syrian households and 50 percent from non-Syrian households. In Turkey, too, no registry data on the Syrian population in the country is publicly available, so the respondents were also selected via area sampling combined with random walk. Based on publicly available aggregate data on the distribution of the Syrian population, the districts (cf. counties) with the highest and second highest Syrian population shares were selected in each region (cf. states) of Turkey. Within these districts, neighborhoods were again randomly selected as PSUs, and respondents were recruited via random walk (see 2.1).

2.3 Follow-up Strategy

Following the baseline (Wave 1) survey, follow-ups were conducted on an annual basis (see Table 1). In up to three attempts, the participants from the baseline surveys were contacted again on the basis of the addresses where the previous interview took place. If the respondent could not be located, the agents tried via phone to arrange an in-person interview appointment, in cases where respondents had provided a telephone number in the previous wave. If this also proved unsuccessful or if the respondent refused to participate in the study again, the contact attempts were cancelled and the respondent’s contact details were permanently deleted. To compensate for the dropout of participants, a refreshment sample was recruited, following the target group requirements and baseline sampling strategies described in 2, 2.1, and 2.2.

If this number of participants or the quota of Syrian and non-Syrian households could not be met through interviews within the neighborhoods of the dropped-out participants, additional neighborhoods were selected for the recruitment of participants. The selection of these additional neighborhoods in Turkey and Lebanon was equivalent to the area sampling strategy of the baseline surveys.

2.4 Deviations

Due to unforeseen events, there were instances where the survey deviated from the outlined sampling and follow-up strategies. First, data collections in 2020 and 2021 took place when restrictions due to the COVID-19 pandemic were in place. These included nightly curfews and lockdowns on the weekends. With appropriate hygiene and social distancing protocols fieldwork was still possible, but the restrictions caused major disruptions and fieldwork delays. These delays in turn lead to a funding shortfall in 2021, so that the third wave data collection in Lebanon was limited to a follow-up of 1,000 panelists. Panelists where only re-contacted up to the point that this quota was realized. No comprehensive follow-up nor refreshment-sampling took place.

Second, a series of earthquakes in February 2023 caused wide-spread devastation in the border region between Syria and Turkey; a region from which roughly half of the TRANSMIT sample was drawn. Consequently, the Turkey fieldwork planned for 2023 was deemed unfeasible under ethical and logistical standpoints. Instead, a limited phone follow-up was conducted to inquire about the well-being of the survey participants. The phone follow-up is not described in this article. See Ruhnke et al. (2024) for details.

3.1 Interview Process

Interviews were conducted using Computer-Assisted Personal Interviews (CAPI), based on questionnaires designed by the project team for the contexts of Lebanon and Turkey. Wherever possible, the items were aligned with the IAB-BAMF-SOEP Survey of Refugees in Germany to facilitate country comparisons. The interviews were conducted by private, for-profit research institutes, namely by the Research & Consulting House (REACH) in Lebanon and Yöntem Research Consultancy Ltd. (Yöntem) in Turkey. Both institutes are members of the European Society for Opinion and Market Research (ESOMAR) and had previously conducted surveys for international research projects (e.g., the United Nations and the International Organization for Migration).

The interviews in Lebanon were conducted in Arabic, while in Turkey, they were held in either Arabic or Turkish according to the participants’ preference. All interviewers were native speakers of the respective language. The questionnaires and participant information were prepared in English by the research team and translated by the contracted survey companies. They also checked the questionnaires for coherence and piloted the baseline questionnaire among a pre-selected population.

Potential participants were approached by employees of the implementation partners at their place of residence, upon explanation of the study’s purpose and content, asked to participate. Upon recruitment, interviewers provided participants with an information sheet containing key information about the study, data protection measures, and contact information and confirmed the participant’s informed consent. Interviews were conducted in a preferably private location within the respondent’s premises, chosen by the respondent. The aim was to create a safe environment that would allow respondents to share private and potentially sensitive information without interference from third persons.

Interviews on average took 55 min in Lebanon and 47.6 min in Turkey (Table 2). In some cases (e.g., for respondents with a more complex migration or family history), this time was exceeded. To reduce the burden on participants, the questionnaire was shortened starting in 2021/22 (wave 3 for Lebanon and wave 2 for Turkey), lowering median interview lengths. Participants were free to pause the interview at any time, skip individual questions, or terminate the interview. Splitting the interview over several days was not permitted. No financial compensation was offered for participation in the study to avoid creating any coercion to participate.

3.2 Questionnaire Design

The surveys cover a wide range of demographic (age, gender, family structure) and socioeconomic characteristics (education, income, employment, housing). In addition, questions about individual migration biographies and intentions, societal participation (networks, discrimination), and (mental) health and well-being were included. Additional modules on events such as the Covid-19 pandemic or the Beirut Port explosion were added on short notice. In addition to the main questionnaire collecting individual information on the respondent, basic demographic information was collected on selected household members and family members outside the household (see 2 for details).

3.3 Ethical Review

Ethical considerations were an integral part of the survey design process, including ensuring the informed consent of all survey participants. But waves 1 through 3 of the Lebanon survey and waves 1 and 2 of the Turkey survey did not undergo independent ethical review prior to fieldwork, as no responsible review body was established at the onset of the project. Starting in 2022 all subsequent survey waves (Lebanon waves 4 and 5, Turkey wave 3), which employed near identical procedures, data protection measures, and similar questionnaire designs as the previous waves, were approved by the Ethics Commission of the Faculty of Humanities and Social Sciences at Humboldt Universität zu Berlin.