Papers, cards, and perilous encounters with the state 27 experiences across time, I opted to use a multi-site model from which to select participants (see Marcus 1995; Hannerz 2003). The field research took place in Kuala Lumpur in Malaysia; New Delhi in India; Chittagong, the refugee camps near Coxes Bazar and surrounding areas in Bangladesh; and (via signal or in person) with Rohingya in diaspora in Europe and Australia. Diasporic Rohingya communities beyond refugee settlements in Bangladesh have played a vital part in building Rohingya social movements across the years (Cowper-smith 2019). Since this research was qualitative and focused on meaning and experience rather than broad general findings, in-depth analysis of the narratives of a smaller rather than larger sample group yield better data (Jørgensen and Phillips 2002:120). There were 104 people who participated in the research. I used snowball sampling, which is common in research related to work with refugees or people with precarious legal status, especially where access can be difficult. Snowball sampling involves asking existing participants to recruit further participants for a study (Weiss 1995). This usually involves the help of one or more ‘gatekeepers’ (Atkinson et al. 2001:29). However, sometimes snowball sampling can slant the data in particular ways, for example, the first participant may reach out to people of a similar social status, age, or political positioning (Weiss 1995). Although it is impossible to eliminate bias in a study such as this, I tried to mitigate such potential slants by using multiple entry points and ‘gatekeepers.’ For example, in Bangladesh, I drew on the help of four different gatekeepers, one from the 1970s displacements, one from the registered camps from the 1990s expulsions, and two from the 2017 expulsions who had different political allegiances. I sought a range of participants across different age groups, from different regions of Rakhine State and Myanmar, with a range of political and family-based affiliations. The ages of participants ranged from their late teens to their nineties. Rohingya identities are characterised by regionalisms—first between those within the zone administered as North Rakhine State, who have been subject to segregation and a separate set of persecutory policies, and those residing in other areas of Rakhine State and Myanmar (Amnesty International 2017); second between Maungdaw, Buthidaung, and Rathedaung within North Rakhine State. These regionalisms intersect with different family-based loyalties and political affiliations. I therefore interviewed Rohingya originally from within Rakhine State—Maungdaw, Buthidaung, and Rathedaung; from elsewhere in Rakhine including Sittwe, the state capital, and southern Rakhine; and those who grew up in Yangon and elsewhere in Myanmar. I made efforts to ensure that both women and men were represented in the research. There were 32 women participants and 72 men. This imbalance reflects several factors: First, gendered norms in Rohingya society include the widespread practice of purdah whereby many women remain in private domestic settings rather than public ones, or often there is a separation of women and men in public spaces (Ripoll 2017; Asia Foundation and Centre