DIVIDING THE RESPONSIBILITIES IN ADVANCE

Division of responsibilities and tasks among team members is the key to any successful collaborative effort. We arrived at an initial division of labor in both projects by following the comparative advantages of team members. The foreign scholars focused on the empirical analysis and framing of survey findings in line with previous literature and I focused on the successful execution of the survey.

This initial division of labor provided each member a “point of entry” into the project. After completion of the survey, I had the opportunity to work on other parts of the project—including the theoretical and conceptual development as well as the empirical analysis—with my foreign partners. It is conceivable that local scholars in survey research have a role limited to data collection and therefore do not feel like an equal partner. This was not the case for us, for two reasons. First, my collaborators and I have similar levels of methodological expertise, which allowed me to engage with the entire project. Second, the theoretical and conceptual framework of the project had to be tied to the specific case we were studying. As a local scholar, I naturally had relatively more extensive knowledge of the Turkish case, which also allowed me to engage with the theoretical and conceptual framework of our research.

IMPORTANCE OF PRELIMINARY FIELDWORK

We started both projects with field trips to the southeastern provinces of Turkey, where many Syrian refugees live. However, our decision to take a field trip as an entire team of scholars before conducting the survey in 2014 was not an easy one. Whereas my coauthors were excited to take the trip to learn about the local dynamics of the Syrian-refugee issue in Turkey, I was hesitant. I thought that it would be challenging to interview locals on a politically sensitive topic with my foreign colleagues who are from the United States and Israel. I was concerned that the interviews could easily go in an undesired direction about the role of these countries in Turkish politics and the broader Middle East, thereby putting my colleagues at potential risk. In fact, during our interviews with locals, interviewees were interested in the nationalities of my coauthors and were willing to engage and challenge them on various political topics, including the Syrian civil war and the involvement of external actors in it.

I hesitantly agreed to participate in the field trip and interviews with my foreign colleagues, but we followed a set of rules to make the interviews as productive and as noncontroversial as possible. We first decided to visit only a single province in the region and then determine whether we would expand our trip to other provinces depending on our initial experience. Second, during our interviews, I introduced my foreign scholars as colleagues working with me on this project, without immediately specifying their nationalities. Third, we decided beforehand that even if an interviewee did not understand English, I would translate all of our conversations with my coauthors during the meetings into the local language. Our interviews with locals proved to be productive and we decided to expand our interviews to other provinces.

BENEFITS OF FIELDWORK

We benefited from this fieldwork in two ways. First, fieldwork can explain findings of the survey that might seem counterintuitive in its absence. This is especially true when the findings do not match well with the expectations of foreign scholars trained in the previous literature. For example, our survey experiment conducted in 2014 in Turkey presented different messages about the potential effects of hosting Syrian refugees in Turkey based on previous research of other scholars (Salehyan and Gleditsch Reference Salehyan and Gleditsch2006). One of our treatments was a positive message emphasizing that Turkey’s open-door policy toward Syrian refugees enables saving the life of innocent Syrian women and children. To my colleagues’ surprise, this positive message turned out to have a negative impact on Turkish locals’ attitudes toward refugees.

As we looked for explanations, the findings of our fieldwork in the Turkish border provinces hosting Syrians provided a possible answer: we repeatedly heard that local Turkish men were marrying Syrian refugee brides as a second or third wife, including child marriages in that region. The increased divorce rate in the region since the arrival of Syrian refugees also was blamed on marriages between Turkish men and Syrian women. This explains how a seemingly positive message about saving Syrian women and children also may have reminded respondents of specific negative externalities that come with hosting young women and children refugees. This interpretation is supported by other scholars’ work on this topic (Erdoğan Reference Erdoğan2015; Orhan and Gündoğar Reference Orhan and Gündoğar2015).

Second, presurvey interviews can shape the survey instrument. For instance, when we were drafting the instrument for the survey of Syrian refugees in Turkey, we realized that several refugees did not feel close to any group involved in the civil war; their preferred position was to remain neutral toward all groups. This was not simply a matter of interviewing Syrians who did not want to be involved in politics. It was true also for those who were active at the initial stages of the uprising in 2011 and had participated in the peaceful protests against the Syrian regime. We realized that our initial preconceptions about Syrians having hardened attitudes and strong support for one of the groups involved in the war were simply wrong, and there were many people who did not feel close to any group. As a result, we decided to add an open-ended question about respondents’ support for neutrality and which group involved in the conflict represented their interests. A significant percentage of respondents stated that none of the groups involved in the conflict represented their interests. Indeed, our observation about refugees’ support for neutrality became a major theme explored in our paper based on this survey.

FROM PRELIMINARY FIELDWORK TO PREPARING THE INSTRUMENT

This preliminary fieldwork typically is followed by finalizing the survey instrument. The initial instrument was drafted in English. The key task of the local academic at this stage was translation of the survey instrument from English into the local language. Translating is certainly not a new task, and excellent guides are available as a resource to scholars who engage in survey research in multinational or multicultural contexts (Harkness et al. Reference Harkness, Braun, Edwards, Johnson, Lyberg, Mohler, Pennell and Smith2010). However, producing a new survey instrument with questions that have not been asked before in the native language of a particular country can be especially challenging for two reasons. First, when drafting the instrument, questions are shaped by the terminology of the extant literature. However, “literally” translating specific terms may not accurately reflect circumstances in the survey country.

For instance, in our survey on Turkish citizens’ attitudes toward Syrian refugees, our aim was to contribute to the literature on potential links between refugees and the likelihood of civil conflict in host countries. However, it would not be factually correct to use the word refugee to refer to Syrians in Turkey in the Turkish version of the survey. This is because in terms of their legal status, Syrians who escaped to Turkey are not refugees. Turkey signed the 1951 Refugee Convention and the 1967 Additional Protocol to the Convention on Legal Status of Refugees with a geographic limitation such that it accepts only people from Europe as refugees (United Nations High Commissioner for Refugees 2016). A term that better captures Syrians’ status in Turkey is sığınmacı (which roughly translates as asylum seekers). Therefore, we used this term in the Turkish version of the survey instrument when referring to Syrians in Turkey.

Second, local scholars are central in formulating questions on politically sensitive topics. Whereas a foreign scholar may want to ask about a sensitive topic, the local scholar can formulate the question in a way that respects local sensitivities and also allows the survey to gauge the respondents’ attitudes. For instance, in our survey on Turkish citizens’ attitudes toward Syrians, there were two closely related issues that were considered to be sensitive in the Turkish context. One was asking about the ethnic identity of the respondents. Our goal was to identify those respondents who belong to minority ethnic groups in Turkey, especially Kurdish citizens. This is because one of our initial hypotheses was that Kurds’ minority status and past experience with political conflict in Turkey may affect their attitudes toward refugees fleeing from conflict in Syria. Although questions about ethnic identity in surveys conducted in Turkey are becoming increasingly widespread, at the time of our survey and especially in the areas we surveyed, a direct question about ethnic identity may have led to nonresponses or outright refusals to participate. Alternatively, we opted to ask about the languages that respondents spoke. Those who told us that they spoke Kurdish were coded as Kurds in our survey sample.

A second sensitive topic was the question about our respondents’ views on political violence in general and the Kurdistan Workers’ Party (PKK) insurgency in particular. The topic of the Turkish–Kurdish conflict is deeply polarizing in Turkish society. Hence, instead of directly asking about their views of the Kurdish insurgency PKK and its use of violence, we asked whether respondents agreed with the use of violence in general as a method to advance political goals. In addition, at the time of our survey, negotiations known as the Peace Process between the Turkish government and the PKK were underway to end the 30-year-long conflict in exchange for political reforms. As an alternative method to gauge our respondents’ views on political violence and the Kurdish insurgency, we also asked them whether they would agree to sign a petition that supports the Peace Process.

The sensitivity of the Kurdish conflict also influenced our sampling decisions in the first survey. Although my foreign colleagues were keen on including all provinces heavily affected by the Turkish–Kurdish conflict, we had to exclude the provinces of Hakkari, Şırnak, and Tunceli from our sampling frame because of the security situation. This was not an easy decision but we did not want to put our enumerators at risk. Instead, we opted to be fully transparent about excluding these provinces in our description of the research design in our published work.

FIELDING THE SURVEY

For both surveys, we received Institutional Review Board approvals from my institution as well as those of my foreign coauthors. However, these permissions may not be sufficient to smoothly conduct surveys in the field. It is likely that as enumerators walk around neighborhoods and knock on doors, especially in small towns, they may be received with suspicion and asked questions by the respondents or others in the area, such as the local police. To alleviate these suspicions, we had a few additional measures in place. First, we did not mention the names or schools of my foreign coauthors in the survey instrument. Second, I received a letter of permission from my school that authorized me to conduct the surveys. Third, we informed local authorities about the surveys beforehand (i.e., either the governors or district governors of the provinces where we conducted our surveys). Our enumerators had copies of these permission and notification forms, which they could show whenever someone asked about their purpose.

Another critical decision that we had to make at the point of survey implementation was the choice between using paper questionnaires and a tablet computer. In both surveys, my foreign colleagues insisted on using the latter because it would be easier to import the collected data and track locations of the enumerators. However, despite these potential advantages, we decided to use paper questionnaire forms for three reasons. The use of tablets would draw extra attention to the enumerators and reinforce locals’ suspicion that some unknown figures were collecting data on them. Second, if local police or security officers questioned enumerators about what they were doing, everything was written on paper on the form. In the case of a tablet, however, there might be hidden material , which could increase suspicion and risk confiscation of the tablet and surveys. Third was the concern for enumerator safety because locals might attack the enumerators to steal the tablets for personal gain.