The online survey Migration and Citizenship after Brexit was conducted as the first phase of the MIGZEN research project. It had a broad target population consisting of those moving to and from the UK before the end of the Brexit transition period, and specifically: (a) British citizens or nationals who were currently living/have lived in an EU/EEA member state (excluding UK); (b) Foreign-born and citizens of an EU/EEA member state who were currently living/have lived in the UK, and (c) Foreign-born and non-EU/EEA persons who were currently living / have lived in the UK. It was designed to understand whether and in what ways Brexit and the COVID-19 pandemic had affected respondents’ perceptions, plans and decisions on whether to stay put, migrate, or repatriate, and how these events had how these events have changed, if at all, attitudes towards the EU, country of residence and origin, understandings of citizenship, identity and belonging.
It was administered via Qualtrics (an online survey platform) and was open for five weeks (13 December 2021 - 16 January 2022), during which we collected 2,024 unique and valid responses.
Survey Design
The survey contained 96 questions organised into six modules, exploring: current residency and migration/legal status; citizenship and migration trajectories; relationships; identities and belongings; social, political and community participation; and socio-demographic information. These were formulated as either multiple choice or open-ended questions, the latter allowing participants to describe their circumstances, position and identities in their own words.
At the start of the survey, participants were asked to confirm their eligibility (i.e., age and inclusion in one of the target populations groups) and to provide informed consent.
Questions in modules 1-5 developed the project’s research questions, while we designed module 6 drawing from existing survey tools and in particular the UK census (2021); the European Social Survey (2020); the ILO Labour Force Surveys and ‘how to survey’ guides prepared by organizations and groups doing anti-discrimination work.
Most modules included a mix of closed and open-ended questions. Closed questions consisted of: (i) multiple choice; (ii) Likert scales; (iii) matrixes; open-ended questions consisted of: (iv) requests of optional additional information; (v) text boxes; (vi) ethnic and/or racial self-identification. Cognizant of the constrained nature of multiple- choice questions, we systematically included in each an open text option, thereby maximizing the survey’s overall inclusiveness and context-specificity.
Pilotting and cognitive interviewing
Before launching the survey, we asked representatives of our project partners The3Million, British in Europe and Migrant Voice, to review the survey and suggest any adjustments. We revised the survey taking on board their comments and suggestions.
To pilot and test the survey, we conducted a round of cognitive interviewing and field testing. This helped us to further develop the questions’ clarity and effectiveness and identify any branching failures in the survey logic. This testing was conducted with seven volunteers, drawn from across the target population who took part in a videocall with a member of the research team. Each volunteer was invited to: (i) respond to the survey; (ii) report in real time any doubt, problem or barrier they may have encountered; (iii) offer their advice on how to improve the formulation of unclear or ambivalent questions. Insights gained through this phase were subsequently incorporated in the survey’s final version.
Sampling and distribution
Participation was by self-selection. Our recruitment strategy was designed to reach out as many different voices as possible among the target populations. We thus circulated our call for participants as widely as possible, investing energies in encouraging the participation of typically underrepresented nationalities and social groups. To this end, and particularly to diversify the profile of respondents, we translated recruitment materials in the languages of the ten most numerous groups of foreign nationals living in the UK – as per the Home Office Immigration Statistics – and distributed calls for participation to migrant and diaspora organizations.
The survey was administered in the form of an anonymous link that we shared across multiple means of online communication: (i) project and partners’ websites; (ii) social media (Facebook and Twitter); (iii) emails to relevant organizations and associations at the local, national and European levels.
Read our original call for participants here.
Analysis
Upon conclusion of the survey, we exported the data into a .CSV file and cleaned it for use with STATA MP4 and NVivo12 to undertake quantitative and qualitative data analysis respectively. We addressed the open-ended responses to the multiple-choice questions (see ‘Structure and design’) by recoding them manually, as relevant and appropriate, either under existing or new responses. The question on ethnic/racial self-identification was open-ended and it required creating categories anew, which we did inductively and with the goal to offer a broad versus granular picture of respondents’ identities.1 The demographic characteristics of the overall sample are described in Section 2 of this report. Using STATA, we undertook a descriptive analysis of the responses to the closed questions, while we used NVivo to undertake inductive thematic analysis of the open-ended questions.
Limitations
Individual vs. relational information. The survey was designed for and administered to individuals, and it did not include a household roster, which means that we do not have granular information on the composition of their household. We also do not know whether and how many respondents were related to each other and/or lived in the same household.
Incomplete sociodemographic data. Approximately one fourth of the respondents (26%) dropped out before completing the survey, meaning that we do not have their sociodemographic information. Very few of those who did reach the end of the survey provided information on their individual and (if applicable) household income, which compelled us to drop this variable from the analyses due to its weak statistical power.