Map 6: Greater Mekong Subregion transportation
3. Study design: collaborative research for action
3.2 Sample design
This dissertation makes use of multiple overlapping samples. The largest sample for this study was purposively developed from among the population of Burmese migrant workers living in Mae Sot and Phob Phra. But I also consider the co-researchers with whom I worked a sample as I took extensive field notes of our group discussions,
conducted in-depth interviews with most of them, and rely on such information to address the primary questions behind the dissertation.7 In addition, I conducted in-depth semi- structured interviews with people in leadership roles from nine different Burmese CBOs working in Mae Sot and Phob Phra, representing three of the member organizations from the advisory group (see Table 4 below for a complete list of participating organizations). This sampling strategy enables me to critically engage with both migrant narratives and
one of the Burmese organizations had moved away from Mae Sot in the middle of the assessment. These interviews were all recorded and then transcribed.
the research process that accessed such narratives. In this section, I devote particular attention to the largest sample.
Given that I cannot hope to write about a “migrant community” in Mae Sot without glossing over numerous differences that divide up the migrant population into numerous and quite distinct groups, I pay particular attention to the differences of the four locations where we collected data and interrogate the notion of community as it is used by civil society, both local CBOs and international NGOs. In some ways, this study relied on the notion that through collaborative work, it would be possible to make some headway in understanding important differences within and between communities. However, it would be essentialist to presume that my co-researchers would be able to articulate Mae Sot’s many different communities devoid of the kinds of power relations or politics that would cause some differences to be privileged while submerging others. As Fine and her colleagues write (2000: 110), it is important to realize that “profound fractures, and variation, cut through lives within…communities,” reminding us that even when being considerate about differences from one community to another, it is important not to forget about differences and power dynamics within communities as well. Thus, I made sure to ask questions about community when I interviewed co-researchers in order to understand how our research might have confronted or perpetuated marginalizing discourse about community dynamics.
The advisory group chose the four sites of data collection via a systematic analysis that took into consideration questions of access, acceptance and interest by residents of the site, familiarity, existing resources in place, and concrete plans for
a possible district to conduct the assessment because their previous work there highlighted numerous social problems facing the Burmese population working in the agricultural sector, including threats to their safety, labor issues, violence, poverty, and a lack of freedom of movement, which precludes access to healthcare and justice. In deciding where in this vast hilly landscape filled with remote labor camps to focus this assessment, the above-mentioned criteria proved useful in identifying five locations. The advisory group then narrowed it to two with a vote. In terms of Mae Sot, all of the organizations represented in the advisory group worked and had strong relationships in the majority of the town’s Burmese neighborhoods or settlements. Thus, in addition to the selection criteria, the group considered where previous research had been done and where they could identify the most “need” for assistance.8
As Figure 3.3 above shows, we chose a sample that would account for gender and that would provide a variety of different perspectives. In each site, the advisory group elected to recruit a sample of residents divided by gender and age; “community workers,” a category that includes anybody considered by the residents there to be a service
provider, accepting a broad meaning of “service” that included para-professional or non- professional health workers (such as “community health volunteers”), teachers,
organizers—effectively those migrant workers who provide some assistance and who are part of a network that links them to the work of NGOs and CBOs; Thai community workers, which included teachers at official schools and border health workers from the Thai government district health offices adjacent to migrant settlements; Burmese
8
Mae Sot s Burmese communities are well-studied, particularly by research with a public health focus, making the risk of over-saturation high.Our goal was not to find an unstudied population; rather we aimed to balance information already generated elsewhere and to engage with migrant workers with greater
community leaders—anybody who the Burmese migrants there consider to be in a leadership or decision-making role (this included religious leaders, staff from CBOs living in that neighborhood who wield authority, or “section leaders” appointed by landlords or CBOs to help manage the migrant population there); and local Thai leaders, such as village heads (phuyai baan) or their deputies, local administrative authorities, or members of security forces who live in the area and have influence (including police or border patrol, for example). While this does not build a representative sample, it enabled us to access the sensitive local knowledge related to violence and safety that might not emerge so easily in larger quantitative studies and from a variety of perspectives.
Sampling was done purposively, as this study’s respondents were recruited via what is known as a “snowball,” “chain-referral,” or “respondent-driven” sampling method, which involves relying on participants to help introduce us to and recommend subsequent participants, ideally helping to break the ice with potential new respondents in terms of familiarity and trust (see for example, Heckenthorn 2002). It was necessary to have a purposive snowball sample strategy rather than a simple or stratified random sample for three main reasons. First, I was working with a group living on the margins of society. It is not always possible to locate individuals with addresses, phone numbers, or other markers of identity and location and to randomly knock on doors may inspire a level of suspicion difficult to dispel within the duration of the interview. Second, in part because of the marginal status of Burmese migrants in Southeast Asia, it is necessary to build relationships with research participants to the extent possible and to move from one trusted contact to another in putting together a sample of individuals willing to speak in- depth. Third, I have tried alternative sampling methods for qualitative research in Mae
Sot with the migrant worker population and found that missing even the slight familiarity that comes with the snowball method made for problematic and uncomfortable interview experiences. While I recognize the many drawbacks associated with relying on snowball sampling (some particularly associated with doing research with forced migrant
populations), as noted in Jacobsen and Landau’s (2003) important piece on methodology, ethics, and research on forced migrant populations, as an exploratory study this project does not claim to capture a representative picture of migrants on the Thai-Myanmar border.
With all this in mind, the sampling of residents started with the links between advisory group members and co-researchers and the Burmese “community workers” in their networks. The latter were crucial entry points into the settlements, though cannot be considered as objective or neutral actors capable—or interested in—generating a
representative sample. However, because they are in many cases a kind of first-responder when violence takes place near them, they were aware of who might be the most willing and interested to discuss the issues facing their community. This reflected the goal of being as representative as possible within the more paramount aim of accessing a set of narratives expressing a partial glimpse of local knowledge.
Between focus group discussions and in-depth interviews, we hoped to build a sample of 35-45 people in each community. Respondents ranged in age from 16 to 64, and 67% percent were women. The vast majority of the migrant sample who self-
identified did so as ethnically Burman. We did not inquire about ethnicity or religion. In terms of the latter, co-researchers and I felt that with ongoing religious tensions between Buddhists and Muslims in Myanmar, inquiring about religious identity might cause
discomfort, though the topic was not taboo and participants felt free to bring it up often. Co-researchers also felt that they did not want to give interviews a formal feeling by requesting bio-data such as ethnicity.9 They preferred to let the topic arise naturally in the flow of in-depth interviews. The FIC study found that among migrant participants, 49% identified Burman, 21% as Karen, 18% as “Muslim”, and 12% as Shan, Pa-O, Mon, Rakhine, Chin, Bangladeshi, Chinese, and Indian (Saltsman 2011).10 The ethnic population of Phob Phra is more heavily Karen, especially in the border area, though labor camps are more diverse. Though we did not ask participants about their legal status (for the same reasons), a majority of those who self-identified did so as undocumented in Thailand. See Table 4 for a detailed demographic breakdown of participants.
Table 4: Participant demographics
Community Gender Total
Male Female
Htone Taung 14 31 45
Kyuwe Kyan 11 30 41
Pyaung Gyi Win/Rim Nam 13 22 35
KM 48 13 20 33
Total 51 103 154
9
This does not presume that the category of ethnicity would have been particularly contentious, especially given the diversity of the co-researcher team (see below). However, like elsewhere ethnicity in Myanmar is a heavily politicized term that is, in many ways, a political and social construct that privileges certain differences and ignores others. During the FIC study, it became clear that participants interpreted
ethnicity as nationality, given the contextual link in Myanmar s current social environment as well as the historical construction of ethnic difference (see for example, Walton 2008).
10 Though Muslim refers to a religion and not an ethnic group, Muslim respondents self identified as
ethnically distinct in this way in the FIC study, a reflection of the fact that that this group finds itself primarily categorized in Myanmar by their religion. There are in fact multiple ethnic groups in Myanmar who are all or majority Muslim (such as Rohingya) or who have a significant Muslim minority, including part of the Chinese-Burmese population. It is possible that some of the participants in the FIC study who
Community Age Total
Under 18 18-40 41-60 61 and up
Htone Taung 3 26 14 2 45
Kyuwe Kyan 4 24 12 1 41
Pyaung Gyi Win/Rim Nam 3 21 10 1 35
KM48 - 21 12 - 33
Total 10 92 48 4 154