Motivation and opportunity for conflict-induced migration: An analysis of Syrian migration timing

Abstract

How do civilians decide when to leave their homes during conflict? Existing research emphasizes the role of violence in driving civilian migration decisions. Yet, migration timing often does not correspond with the timing of violence. To explain this discrepancy, I argue that violence fits within broader considerations of motivation and opportunity to migrate. Witnessing violence triggers post-traumatic growth that delays narrative ruptures and the subsequent migration that they motivate. Civilians who have ‘wasta’ – an advantaged social position resulting from some combination of money and connections – have the opportunity to migrate safely. Civilians who possess both motivation and opportunity migrate earlier. I use over 170 structured interviews with Syrian refugees in Turkey to test this argument. Descriptively, respondents who did not witness violence (early motivation) left their homes seven months earlier, on average. Respondents with wasta (opportunity) left their homes one full year earlier, on average. Respondents who both did not witness violence (early motivation) and had wasta left their homes approximately one and a half years earlier, on average. Cox proportional hazard models reveal that respondents only migrated earlier in the conflict if they had both early motivation and opportunity. Open-ended responses from the interviews support the quantitative results and help explain their causal mechanisms. These findings contribute to understandings of conflict-induced migration, civil war, and the Syrian conflict.

Keywords

armed conflict civil war migration narratives refugees Syria

Introduction

By the end of 2016, civil war in Syria had produced roughly six million internally displaced persons and six million refugees, including about one million in Europe. These numbers have provoked global alarm, strained host country resources, and contributed to fears of the spread of conflict within and outside Syria (Salehyan & Gleditsch, 2006). Syrian conflict-induced migration skyrocketed in 2013, just as pro-government and anti-government violence increased and became more indiscriminate (Lynch, 2013). Violence levels, violence types, structural conflict conditions, or other factors could drive this migration timing. Meanwhile, at the end of 2016, about half of Syria’s pre-conflict population of approximately 24 million people had not migrated (McHugo, 2015). If we accept that some people will just never move, then a crucial question remains: for people who migrated, why did they leave when they did?

Explaining migration timing will highlight civilian agency amidst the difficult circumstances of armed conflict and clarify how civilians respond to violence. Rather than migration timing being the outcome of binary choices of whether to fight or whether to flee – fight or flight – people select from large repertoires of protection strategies such as daily movement, bribery, or protest (Baines & Paddon, 2012; Jose & Medie, 2016; Kaplan, 2017). These choices may lead civilians to stay in violent areas as they find alternative protection strategies to migration.

Existing research suggests that perceived physical threat from violence has the strongest effect upon conflict-induced migration (Adhikari, 2013; Davenport, Moore & Poe, 2003), but this relationship exists specifically for violence in residential areas (Schon, 2016). Violence has different kinds of effects upon migration depending on the targeting patterns and location of violence, as well as whether people are considering their awareness of violence or personal experiences of violence (Balcells & Steele, 2016; Blattman, 2009; Schon, 2016; Steele, 2019). Across these violence types, violence timing does not have a clear correlation with migration timing (Melander & Oberg, 2006; Schon, 2015). For a more effective explanation of migration timing, violence must be incorporated within a broader set of considerations.

These broader considerations involve motivation and opportunity. Building on frameworks of motivation and opportunity from contentious politics and international relations, motivation refers to the preferred choices for civilians and opportunity refers to the ‘total set of environmental constraints and possibilities’ (Most & Starr, 2015: 23; Tilly, McAdam & Tarrow, 2001).

As people consider when to migrate, I expect that civilians are aware of violence during conflict, which creates some level of perceived physical threat (Davenport, Moore & Poe, 2003; Moore & Shellman, 2004). Perceived threat motivates responses, which are driven by narratives – discrete stories that encapsulate a given set of information (Olsen, 2014). Narratives are guides for what is happening and how best to respond. New information and experiences that are not consistent with existing narratives strain belief in those narratives. Some people break under this strain and suffer narrative ruptures – breaks in understanding of ongoing events due to fundamental contradictions between existing narratives and new information and experiences (Janoff-Bulman, 1992; Pemberton & Aarten, 2018). When people suffer narrative ruptures, they often become motivated to migrate (Rosen, 2017).

Different types of violent experiences – violence to family, violence received, and violence witnessed – can increase or decrease the ability to understand how to reconcile new experiences and information with narratives depending upon whether they trigger post-traumatic growth (PTG) or post-traumatic stress disorder (PTSD). Social support helps people understand violent experiences, thereby helping people undergo PTG and not PTSD, but it only helps up to a certain threshold of trauma. Since violence witnessed is unlikely to cross that threshold, it is the only type of violent experience that is correlated with PTG without having a clear correlation with PTSD (Blattman, 2009; Kulkarni et al., 2011; MacDonell, 2012). Therefore, people who witness violence tend to undergo PTG, believe existing narratives for a longer duration and suffer narrative ruptures later, and develop motivation to migrate later than those who do not witness violence.

Motivation to migrate, however, does not independently affect migration timing. Civilians also need opportunity to migrate safely. This opportunity comes from the security environment and resources available to respond to that environment. This security environment is shaped by the types and spatial distribution of violence. Specifically, civilians must monitor whether violence involves selective or indiscriminate targeting and whether the mix of violence targeting is similar within residential areas (home violence) and along migration routes (road violence) (Kalyvas, 2006; Schon, 2016). Often, home violence contains a higher share of indiscriminate targeting and road violence contains a higher share of selective targeting (Balcells & Steele, 2016; Lombard, 2013; Pottier, 2006).

Civilians who have an advantaged social position – resulting from some combination of money and connections – can protect themselves from the selective targeting of road violence. Advantaged people protect themselves by leveraging their own power and relationships with powerful individuals and institutions (De Smedt, 2009; Peleg & Waxman, 2011; Utas, 2012). This protection does not help in the face of indiscriminate violence in residential areas. Since conflict security environments tend to include a larger share of selective violence along migration routes, advantaged people can protect themselves more effectively while moving. As a result, civilians who hold an advantaged socio-economic status should possess greater opportunity to migrate safely. If civilians already have motivation to migrate, then holding an advantaged status allows them to act on that motivation.

I test this argument using structured interviews with over 170 Syrian refugees in Turkey. With Syria producing more refugees than any other country and Turkey hosting the largest number of those refugees, this is an especially important case to study. Syria’s conflict began with protests in March 2011, but the regime of Bashar Assad implemented a violent crackdown that, rather than quashing dissent, escalated tensions into all-out conflict (Pearlman, 2016). This conflict presents a case that is extremely salient for its high levels of conflict-induced migration. On a practical level for researchers, the fact that Syria’s conflict is ongoing and began relatively recently in 2011 contributes to the case’s value. Respondents were still able to recall the precise and detailed information.

My interviews contained open-ended and closed-ended questions. To test the motivation hypothesis that people with experiences of violence migrate later, I use indicators of violent experiences based on the Survey of War Affected Youth (Blattman, 2009; Blattman & Annan, 2010). To test the opportunity hypothesis, I operationalize advantaged social status with an indicator of whether or not people had ‘wasta’ when they were inside Syria. I define ‘wasta’ as having sufficient money, connections, or both to enjoy special influence or increased ability ‘to find jobs and obtain government services, licenses, or degrees that would otherwise be out of reach or would take a long time or effort to obtain’ (Barnett, Yandle & Naufal, 2013; Ramady, 2015: vii). Then, I add an interaction term to test for the conditional effects of motivation and opportunity, as well as whether both motivation and opportunity are needed to influence migration timing. Cox proportional hazard models yield quantitative evidence in support of the motivation and opportunity hypotheses when there is no interaction term. With the interaction term, the models show that respondents only migrated earlier, on average, if they did not witness violence (early motivation) and had wasta (opportunity). Open-ended responses from the interviews support the quantitative results and help explain their causal mechanisms.

The next section reviews recent research on conflict-induced migration, particularly the role of violence as the strongest driver of migration. I then discuss the factors behind motivation and opportunity to migrate. In the empirical section, I test the hypotheses. I conclude with a discussion of the results and potential avenues for future research.

The complexity of the violence–migration relationship

The most robust finding from research on conflict-induced migration is that violence causes migration. There are direct and indirect channels through which violence influences migration during conflict. Violence can directly affect migration by causing civilians to perceive a higher level of threat than they are willing to tolerate (Adhikari, 2013; Davenport, Moore & Poe, 2003; Moore & Shellman, 2004, 2006). This argument produces the expectation of a linear relationship between violence and migration, where more violence leads to more migration. Other work contends that violence has a non-linear relationship with migration – low to moderate levels of violence actually reduce conflict-induced migration and high violence levels increase conflict-induced migration (Massey & Bohra-Misha, 2011). Other factors, including social networks, economic factors, membership in community organizations, transportation options, and attachment to one’s homeland, can mediate the effects of violence, but violence remains the strongest influence upon conflict-induced migration (Adhikari, 2012, 2013; Czaika & Kis-Katos, 2009; Massey & Silva, 2014; Williams, 2013).

Violence can also indirectly affect migration by reducing economic opportunities. Without sufficient income, many civilians choose to move. Debate is ongoing over whether civilian migration is fundamentally different when violence or economics are the direct causes of migration (Betts, 2013; Engel & Ibáñez, 2007; Massey & Bohra-Misha, 2011). Regardless, physical and economic insecurity are strong drivers of conflict-induced migration (Arias, Ibáñez & Querubin, 2014).

While this research has offered valuable explanations of the causes of migration, migration timing remains difficult to explain. If violence alone causes migration by creating threat perceptions and damaging the economy, then migration timing should correspond with the timing of violence. Yet, in both cross-national and within-country analysis, migration timing does not correspond with violence timing. Cross-national analysis from Melander & Öberg (2006) argues that conflict always threatens civilians, so people will move as soon as possible. This produces high levels of migration at the beginning of a conflict that taper off over time. Within-country analysis from Schon (2015) argues that migration levels fluctuate, with migration increasing when civilians observe or expect changes in the nature of the conflict.

A key limitation of both of these studies is that they examine macro-level evidence and rely upon assumptions about individual behavior. I address this shortcoming by providing individual-level evidence to match my argument about individual decisions of when to migrate. For this argument, it is essential to disaggregate violence by location, type of targeting, and whether civilians have awareness of violence or personal experiences of violence. The following motivation and opportunity sections will lay out the argument with an explanation of the key variables and causal mechanisms through which civilians select their migration timing.

Motivation to migrate: How witnessing violence delays migration

Threat perceptions alone are not enough to explain when people become motivated to migrate. Explanations must also account for why people respond differently to similar threats. In particular, different people may respond differently to similar violent experiences.

While some violent experiences produce negative psychological responses such as PTSD, personal experiences with violence may actually yield more positive than negative effects due to post-traumatic growth (Joseph, 2013).¹ Post-traumatic growth (PTG) involves improvements in five components: appreciation of life, relationships with others, new possibilities in life, personal strength, and spiritual change (Tedeschi & Calhoun, 1996). For civilians living through armed conflict, a key implication of PTG is that it produces increased creativity in finding coping mechanisms (Calhoun & Tedeschi, 2014). The positive psychological outcomes from PTG after violent experiences coexist with the negative psychological effects of PTSD.

The relationship between violent experiences and PTSD or PTG is different for different types of violent experiences. Researchers have had difficulty precisely identifying the specific effects of each type of violent experience due to their frequent co-occurrence for the same individuals, but there is a consensus that the effects of each type of violent experience are different and that they should not be aggregated into a single indicator of violent experiences (Barbarin, Richter & DeWet, 2001; Eitle & Turner, 2002; Shields, Nadasen & Pierce, 2009; Uehara et al., 1996). Violence to family members indirectly affects an individual through the characteristics of the relationship between the individual and family member, as well as individual traits like empathy. Considering more direct types of violent experiences, receiving violence – such as being beaten or shot – can trigger PTSD, but there are mixed findings about its relationship with PTG. Witnessing violence can trigger PTG, but it may not have a significant relationship with PTSD (Kulkarni et al., 2011; MacDonell, 2012; Paxton et al., 2004). Since it has the potential to create positive psychological changes without the negative changes, violence witnessed should demonstrate a significant relationship with PTG and the resultant increased creativity in finding coping mechanisms (Blattman, 2009; Blattman & Annan, 2010; Werdel & Wicks, 2012).

To cope with the dangers of armed conflict, people need to understand security threats and how to respond to those threats (Allport & Postman, 1947; Shibutani, 1966). Narratives, which can be defined as discrete stories, provide this understanding. They can be personal stories about an individual’s experiences, or they can be broader stories about how one side in a dispute understands the events that have taken place (Olsen, 2014). Jose & Medie (2016) categorize the options to respond to perceived threat into violent engagement, nonviolent engagement, and non-engagement strategies. Violent engagement involves fighting. Nonviolent engagement involves protests, demonstrations, and other public nonviolent methods to change armed group behavior or protect people from it (Kaplan, 2013a,b). Non-engagement includes migration, along with a wide range of actions such as hiding, paying bribes, daily movement, or sharing information (Baines & Paddon, 2012; Corbett, 2011).

Among these strategies, civilians tend to prefer alternatives to migration due to its costs and dangers (Schon, 2016). They can identify alternatives as long as they believe a narrative about their security. Yet, civilians may have violent experiences that seem inconsistent with existing narratives. In these situations, people look to their communities to understand how to reconcile the violent experiences with a narrative (Pemberton & Aarten, 2018). If their communities provide enough support to help them understand the violent experiences, then they often undergo PTG (Randolph, Koblinsky & Roberts, 1996; Rizkalla & Segal, 2017). This is because social support helps people understand how to reconcile violent experiences with existing narratives. Individuals can then learn how to do this on their own and be prepared for it as they undergo PTG. If community support is insufficient, then violent experiences may overly contradict existing narratives, leading to a ‘narrative rupture’. When narrative ruptures occur, existing understandings of home and the events occurring there shatter (Janoff-Bulman, 1992). People lose their ability to explain their own experiences through the lens of popularly accepted narratives, creating substantial uncertainty over what is happening and how best to respond (Pemberton & Aarten, 2018). In this uncertainty, many people choose to migrate (Rosen, 2017).

Therefore, the timing of narrative ruptures should drive the timing of motivation to migrate. That timing is influenced by whether people receive sufficient social support after violent experiences. Higher social support increases the likelihood of PTG, while lower social support increases the likelihood of PTSD (Foster & Brooks-Gunn, 2015; Salzinger et al., 2002; Shields et al., 2010). With PTG, people continue believing narratives for a longer period of time, delaying narrative ruptures and the resultant motivation to migrate. PTSD yields the opposite effect and facilitates earlier narrative ruptures.

Beyond a certain threshold of severity, however, social support becomes ineffective. Receiving violence may cross this threshold, so I expect it to lead to PTSD and earlier migration. This expectation is tentative, however, since receiving violence could lead to PTG and later migration in some cases. Witnessing violence is unlikely to be so traumatic that social support cannot mediate its psychological effects, so it is likely to be more closely related to PTG and later migration (Kaminer et al., 2013). The causal mechanism of PTG and delayed narrative ruptures is difficult to observe, but evidence consistent with its observable implication – witnessing violence delays migration – supports the causal mechanism (McAdam, Tarrow & Tilly, 2008).² This yields the following hypotheses:

H1: Civilians who witness violence will migrate later than those who do not witness violence.

H2: Civilians who receive violence will migrate sooner than those who do not receive violence.

Opportunity to migrate: How advantaged social status provides opportunity to migrate

Alternatively, conflict may produce so much violence that people generally want to migrate, and they are just evaluating whether they have the opportunity to migrate safely. This idea parallels the notion from contentious politics that there are always people with sufficient grievances to motivate political action, leaving political opportunity structures as the real source of variation that drives participation (Meyer, 2004). During conflict, the opportunity structure for migration is created by the combination of locations and targeting of violence (Balcells & Steele, 2016; Kalyvas, 2006; Schon, 2016). In this structure, civilians who hold an advantaged social status have greater opportunity to migrate. Civilians can leverage advantaged status for protection during migration if the security environment created by the combination of locations and targeting of violence contains selective violence along migration routes.

Syria’s security environment is consistent with this scope condition. Violence along migration routes in Syria tends to occur at checkpoints. A report by the research group Caerus found 1,462 checkpoints in Aleppo from 16 September 2013 through 6 January 2014 (Caerus, 2014). Checkpoints proliferate throughout all of Syria’s major cities, controlled by pro-government and anti-government groups. Respondents almost unanimously indicated that they encountered checkpoints in Syria. To illustrate the prevalence of checkpoints, one person from the town of Afrin, which is near Aleppo, explained:

At first, there were not a lot of checkpoints. It was just one and then they were erected every 100 meters and 200 meters. There were 30 to 40 regime checkpoints from just Afrin to Aleppo [about 40 miles] by the time I left in 2015. (Hall, 2016)

Checkpoint violence, while harming many civilians, still has defined targets. It is a form of selective violence (Kalyvas, 2006; Schon, 2016). As armed groups select their targets at checkpoints, civilians with advantaged status can leverage their status for protection. Soldiers know that they should not harass or threaten advantaged people (Debos, 2016). This is similar to using vertical ties for protection (Harpvicken, 2009).

In Syria, advantaged status does not also affect motivation to migrate because violence in residential areas is often indiscriminate. If it were selective, then advantaged status could also influence motivation to migrate. Instead, civilians face indiscriminate violence at home from barrel bombs, chemical weapons, and pro-government and anti-government militias who are undisciplined, poorly controlled, and abusive (Byman, 2016; Lister, 2016; Lynch, 2013). For example, from March 2013 to March 2017, the United Nations documented 25 chemical weapons attacks in Syria.³ Physicians for Human Rights also documented 830 medical personnel killed in 478 attacks on 323 separate hospitals from March 2011 through August 2017 (Physicians for Human Rights, 2017). This makes everyone, including advantaged people, feel threatened.

As they decide how to respond to information about violence, civilians consider whether they have advantaged social status. Advantaged social status in Syria can be operationalized by whether people have wasta (Lust, 2009). Linguistically, the word ‘wasta’ comes from the Arabic root for ‘middle’ or ‘medium’. It can be translated into English to refer to connections or resources, which extends to money.

Sometimes wasta allows civilians to pass through checkpoints without incident. In other instances, civilians are threatened and escape safely thanks to wasta. One person explained:

They took us at a roadblock. They said they were with the government, but it was clear that they were not government soldiers. We were held one night and released the next morning. The soldiers were Shabiha. They asked for ransom. We were released because my friend’s relatives knew about it and contacted people who were high-up in the government. They knew the kidnappers and forced them to release us without a ransom. (Respondent T085)

Instead of being able to rely upon direct personal connections, some people must pay for their wasta. For example, one person paid a man from the mokhabarat (secret police) to help him escape the Syrian government:

I was wanted by the military, so I paid a mokhabarat man to take me to Hamaa. He used his military ID to get us through checkpoints […] I went through Hamaa because it was the only way to get to Idlib […] Hamaa was controlled by the regime […] I passed through 9 checkpoints between Hamaa and Idlib. (Respondent T116)

I use this insight to examine how wasta affects civilian behavior during conflict. Wasta’s ability to protect civilians – thereby providing opportunity to migrate – should allow those with wasta to migrate earlier. This expectation yields the following hypothesis:

H3: Civilians who perceive that they have wasta will migrate sooner than those who do not.

Another possibility is that civilians consider both whether they want to migrate and whether they have the opportunity to migrate. It may not be a given that some number of people will always be motivated to migrate. Following Berry, Golder & Milton (2012), I consider the possibilities that migration timing is influenced by motivation without opportunity, opportunity without motivation, and both motivation and opportunity. Considering international migration, de Haas (2014) uses an aspirations–capabilities framework to argue that motivation alone is not enough to produce migration. During conflict, motivation without opportunity produces people who want to migrate but do not believe they can do it safely. Lubkemann (2008) identifies this group of people

Table I.

Interactive expectations

	Wasta	No wasta
No violence witnessed (Hypothesized early motivation)	Migrate earliest	Migrate earlier than reference category
Violence witnessed (Hypothesized late motivation)	Migrate earlier than reference category	Reference category

as being stuck in a state of ‘involuntary immobility’. Checkpoint violence, as discussed earlier, may deter so many people from migration that motivation alone would not tend to produce migration. On the other hand, opportunity alone cannot explain why everyone with opportunity does not immediately leave. I test the possibility that both motivation and opportunity are needed to influence migration timing. For analysis with an interaction term, I also consider the conditional effect of motivation and opportunity, respectively, without the presence of the other.⁴ Motivation alone or opportunity alone should lead to earlier migration than neither motivation nor opportunity. People should migrate earliest when they have both early motivation and opportunity. Table I illustrates the expectations for the interaction between wasta and witnessing violence. The expectation for receiving violence is more tentative, so I examine its interaction with wasta in the Online appendix.

Sample selection

To gather data on civilian migration timing in Syria, I conducted interviews with 179 Syrian refugees in Turkey from July to November 2016. Practical and ethical concerns prevent researchers from interviewing civilians living in active conflict zones who do not migrate, which is why I do not address the causes of migration. For the study of migration timing, I acknowledge that people who have not migrated could be people who have not yet migrated or people who will never migrate.

Sampling biases may deflate estimates of the effects of witnessing violence, receiving violence, and wasta. Based on my argument, people who have not yet migrated may have witnessed violence, received social support that helped them undergo PTG, and then not experienced a narrative rupture. Or, they may not have wasta. Out of those who received violence, I can only include those who did not die. Usually, these individuals received minor forms of violence.⁵ People who will never migrate could fit into any of these scenarios, or they may have altogether different characteristics that this article cannot address.

My analysis thereby carries the caveat that I only assess variation in migration timing out of people who migrated to Turkey between the start of the Syrian conflict in March 2011 and July 2016. Despite the biased sample, Turkey hosts more refugees in total, and Syrian refugees in particular, than any other country. Analyzing the migration timing of Syrians who migrated to Turkey can therefore still contribute to understanding migration behavior.

This focus on Syrian refugees in Turkey, as well as the sampling procedure to identify respondents, lead to the emphasis of a specific subset of Syrian refugees. Consistent with most research involving Syrian refugees, the sample primarily includes secular Sunni Arabs who oppose Bashar Assad (Pearlman, 2016). Syrian refugees in Turkey also remained contiguous to Syria. Contiguity often indicates that these refugees were either unable or unwilling to move further. Many respondents stated that they chose Turkey because they want to remain in a Muslim-majority country. While the language barrier of switching from Arabic to Turkish is substantial for many Syrians, the shared predominant religion often eases the cultural transition. Moreover, Turkey is commonly perceived as one of the few countries actually willing to welcome Syrian refugees.

Syrian refugees who successfully reached Europe or other Western countries have different characteristics than those who remain in Turkey or other Middle Eastern countries. Most importantly, they tend to be wealthier, more educated, and have better networks (Erdogan, 2016). For Western host countries, this is often intentional. Turkey attempted to force this dynamic to change in 2015 and 2016, as it attempted to negotiate a deal with the EU that would obligate the EU to accept more refugees with disabilities, criminal records, less wealth, and less education. Its efforts have extended at times to blocking resettlement to the United States for Syrian refugees with high education and skill levels, even after the United States accepted them for resettlement (Kingsley, 2016).⁶ Syrians in Turkey were therefore able to exit Syria, but they often did not migrate further due to cultural reasons or they lack the educational, economic, and social network resources to do so.

During the fieldwork period from July to November 2016, Turkey faced severe political turmoil from a coup attempt on 15 July 2016, and a harsh post-coup crackdown. Turkey also suffered several terrorist attacks and escalations in conflict with the Kurdistan Worker’s Party (PKK) in the southeast. These developments made it unsafe to visit border regions in the south and southeast that host large numbers of refugees.⁷ Instead, I conducted interviews in Istanbul and Izmir, where safety concerns could be balanced with the practical research value of cities hosting large numbers of Syrian refugees. Figure 1 displays the fieldwork regions and regions where fieldwork could not safely take place.

Istanbul hosts the largest number of Syrian refugees in Turkey, roughly four times more than Izmir, so the majority of interviews took place in Istanbul. In total, 145 interviews occurred in Istanbul and 34 were in Izmir. Within Istanbul, there is substantial variation within Syrian refugee communities, so interviews were conducted across five university campuses and six districts. This sometimes required travelling as much as two or three hours to meet with respondents. By spreading interviews across the city, it was possible to meet a wider range of respondents. This ranged from the relatively prosperous Syrian refugees in Fatih district to the poorer Syrian refugees, including several of the Kurdish Syrian refugees who completed interviews, in districts like Esenyurt. Figure 2 displays the field sites and distribution of Syrian refugees across districts within Istanbul as of July 2015 (Kaya & Kirac, 2016).

Working through these multiple points of insertion into Syrian refugee communities and then snowball sampling, I was able to include respondents with variation in the values of key explanatory variables (Bloch, 2007). With multiple points of insertion, I minimized the bias in my sample that is a source of substantial methodological concern in refugee research (Jacobsen & Landau, 2003). In my sample, 100 out of 156 people witnessed violence and 53 out of the 154 people had wasta when they were inside Syria.⁸

These interviews were structured with both open-ended and closed-ended questions, and they typically

Figure 1.

Turkey field sites

Figure 2.

Istanbul interview locations

lasted between 60 and 90 minutes. A few interviews were just over two hours, and one interview lasted over four hours. There was one particularly short interview that lasted about five minutes, but that is an outlier. I was able to meet with some respondents multiple times, thereby providing the opportunity to follow up on questions and increase the reliability of interview responses (Fujii, 2011). Interview methods are summarized in Table AI in the Online appendix, following the guidelines of Bleich & Pekkanen (2013).

Variable operationalization

This section will explain how questions from the interviews yield variables for the analysis and present descriptive statistics for those variables.⁹ These descriptive statistics demonstrate that there is a diverse sample of respondents that captures variation along all key explanatory variables (Bloch, 2007). Furthermore, there is wide variation in migration timing, ensuring that the sample captures civilian experiences during all stages of the Syrian conflict from March 2011 to July 2016.

While not a variable included in the hypotheses, Armed group ties is a potential indicator of access to information about armed group activities. It is a count variable of the number of armed groups with members that the respondent knows. There are ten armed group categories under consideration, covering major pro-government, secular opposition, the Kurdish opposition group YPG, and Islamist groups. While Syria has hundreds, if not thousands, of armed groups, these ten categories were selected in order to identify major armed groups and ‘other’ options where respondents could highlight armed groups not specifically identified in the questionnaire.¹⁰

Since violence has different kinds of effects when it occurs in residential areas than when it occurs along migration routes (Schon, 2016), I also include indicators of threats along migration routes: Obstacles and Facilitators. Obstacles captures direct threats to civilians along migration routes. In addition to capturing migration-deterring factors, I capture factors that encourage migration with Facilitators. Facilitators captures fears of waiting too long to migrate, whether any armed groups generally made movement easier, and whether respondents received any offers of protection during movement. Obstacles and Facilitators measure opposite ends of the spectrum from migration encouragement to deterrence, so I do not include them in the analysis independently. Instead, I create indicators to isolate the combinations of High obstacles and low facilitators, Low obstacles and high facilitators, and High obstacles and high facilitators. High and low values were determined based on whether the respondent experienced more or fewer obstacles and facilitators than the average.

The variable Wasta is a dichotomous variable indicating whether the respondent believes that they had wasta when they were inside Syria. Despite a potential for Syrians to underreport whether they had wasta, the fact that 53 out of 154 respondents reported that they had wasta suggests that many respondents were not deterred from sharing this self-perception. In addition, respondents sometimes expressed a belief that wasta was necessary to live comfortably and were not embarrassed to admit that they had it. This is consistent with the trend that despite general disdain for the wasta system, individual Syrians are willing to use wasta to their benefit (Ramady, 2015). Nuances in respondent understandings of wasta were detected by asking respondents how they would define wasta. Respondent definitions of wasta yielded the opportunity to explore the variation in how respondents understood the concept, but they also highlighted the common thread that no matter how respondents defined wasta, they all felt that wasta provided security.

Violent experiences are measured as three variables, based on Blattman (2009) and Blattman & Annan (2010). Violence received is a count variable that captures eight kinds of violence that respondents may have suffered when they were inside Syria. This includes hearing gun fire, being beaten or attacked, having property stolen, and suffering injuries. Out of the 152 respondents who provided data for this variable, there were 29 individuals who only received one form of violence. All of those individuals reported having heard gunfire. Out of the 50 people who had received two forms of violence, 46 heard gun fire and lost property. Then, 36 of the 47 people who had received three forms of violence had heard gun fire, lost property, and had bullets shot at them or at their home. Therefore, as respondents received more forms of violence, they were usually receiving more severe forms of violence. Since the least severe forms of violence were also the most common in my sample, my respondents are less likely to receive sufficiently severe violence to systematically undergo PTSD rather than PTG. As a result, I am less likely to find support for H2.

Violence to family is a count variable that captures whether family members were killed, abducted, or suffered forced disappearances. Violence witnessed is a dichotomous variable derived from whether respondents witnessed killings, beatings, or torture. Following the argument on migration motivation, respondents who did not witness violence had early motivation, while those who did witness violence had late motivation. Since 100 out of 156 respondents did witness violence, there were 56 respondents with hypothesized early motivation and 100 respondents with hypothesized late motivation. These variables measure the type, not frequency, of violence exposure. Frequency of violence exposure, however, does not appear to have a significant psychological effect on victims (Başoğlu, 1993). Since motivation to migrate is heavily influenced by the psychological effects of violence exposure, a focus on violence types is even more justified.

Table II.

Descriptive statistics for continuous and categorical variables

Variable	Observations	Mean	SD	Min	Max
Armed group ties	158	2.987	2.102	0	9
Violence received	152	2.520	1.409	0	8
Violence to family	153	0.843	0.919	0	4
Neighborhood trust	174	3.592	1.447	1	5
Neighborhood alert	174	2.299	1.455	1	5
Total income (Syrian pounds)	161	39,596.270	41,619.310	6,000	400,000
Age	176	31.125	9.682	18	65
Obstacles	161	3.81	1.57	0	6
Facilitators	166	1.03	0.89	0	3

Table III.

Descriptive statistics for binary variables

Variable	Observations	Absolute frequency	Relative frequency
Wasta	154	53	0.344
Male	178	120	0.674
College educated	176	94	0.534
Betrayal	155	23	0.148
Computer	178	141	0.792
Ties to international company	168	43	0.256
Ties to police	168	67	0.399
Violence witnessed	156	100	0.641
High facilitators, high obstacles	159	29	0.182
High facilitators, low obstacles	166	9	0.054
Low facilitators, high obstacles	159	71	0.447
Low facilitators, low obstacles	159	50	0.314

Finally, additional demographic information was collected on gender, monthly income, age, education, and whether the respondent owned a computer in Syria. Income, Age, Education, and Computer ownership are control variables. Descriptive statistics are presented in Tables II and III.

I measure migration timing using the month respondents first left their homes in Syria. March 2011 is coded as the first month because that is when the Syrian conflict began. While war did not unfold evenly across Syria, existing research does show that civilians attempt to predict violence patterns in order to keep themselves safe (Rubin & Moore, 2007; Schon, 2015; Uzonyi, 2014). This means that civilians throughout the country were likely to have begun considering their responses as soon as the conflict began in March 2011. Respondent migration timing ranged from April 2011 until July 2016. Figure 3 displays the resultant distribution in migration timing.

Figure 3.

Respondent migration timing

In case civilians actually begin considering migration at different times based on subnational variation in conflict onset, I also created an alternative migration timing variable as a robustness check. This variable codes the first month of conflict in each Syrian province as the month when the Violations Documentation Centre (VDC) documents the 25th fatality.¹¹ I chose this threshold because the Uppsala Conflict Data Program uses 25 fatalities as its threshold for civil conflict onset at the country level (Gleditsch et al., 2002). Adapting that threshold to subnational conflict onset in Syria, 25 fatalities for a province was a logical extension. For example, Dera’a’s conflict onset is March 2011 and Aleppo’s conflict onset is October 2011. I matched respondents to provinces based on their origin location within Syria. Results did not change for this robustness check. For simplicity, I only discuss the results of analysis using March 2011 as the month of conflict onset for all respondents.

Testing the hypotheses

Descriptive statistics from these interview data suggest that there is strong support for H1 and H3. For the motivation variable of witnessing violence, respondents who witnessed violence left their homes in June 2014, on average. Respondents who did not witness violence left their homes in December 2013, on average. Given the high violence levels in Syria, this seven-month difference is striking. As one respondent explained, violent experiences, which includes witnessing violence, were increasing the desire of many people to help others, which is consistent with PTG:

When people have many violent experiences, they stop caring whether they will die. They just talk to whoever they want […] People in general just assume that they face danger no matter what they do […] Why not help the people you care about? (Respondent T001)

Regardless of whether respondents experienced PTG and delayed narrative ruptures, they described clear narrative ruptures that had occurred at some point. Many respondents described narrative ruptures as ‘everybody going crazy’. In one example, ‘[f]ew people have the same political beliefs as me […] I think in a logical way. Everyone is going crazy. In Islam, it is forbidden to kill without reason. Both are wrong, and they destroyed Syria. Now they realize that they were wrong’ (Respondent T166). For another respondent, the government specifically went crazy. He explained, ‘I didn’t see how anything would get better. The government became crazy. They started to arrest people randomly. They would arrest you even if you helped them. After investigating, they would realize that they had their own people, but they would continue anyway’ (Respondent T023). For other respondents, narrative rupture materialized in the form of disappearing trust. For one person:

Trust disappeared during the conflict. For example, if one person had no water, he will try to stall your water to get his own water […] 30% will help […] 30% will trust with money […] I was living on a street that was very simple. It changed during the conflict. You can’t live there without money […] These bad circumstances led people to do bad things. (Respondent T157)

Another respondent discussed a very similar process, ‘[w]e had a besieged area where the opposition was. We knew many of these guys. It was easier to trust. Now it is big and widespread. I don’t know who is fighting who anymore. I don’t trust anyone in Syria now’ (Respondent T144). Some people may even be able to pinpoint a precise event that ruptured their narrative of home. In one case, a respondent recounted:

My brother was arrested in 2013. He was killed in prison after being tortured. I had been working with my brother in an organization that provided aid for people in need. After that, the mokhabarat came to my home and asked me about my brother. I was afraid. From that time, I prepared myself to travel. (Respondent T021)

The opportunity variable of wasta is even more striking, with a difference of one full year. Respondents who had wasta when they were inside Syria left their homes in July 2013, on average. By contrast, respondents without wasta left their homes in July 2014, on average.

The combination of motivation and opportunity produces the largest difference in average migration timing. Respondents with both early motivation and opportunity left their homes in March 2013, on average. Respondents with late motivation and without opportunity left their homes in August 2014, on average.

To build on these descriptive findings, the hypotheses are tested with Cox proportional hazard models that can directly assess the determinants of migration timing. Due to violations in the proportionality assumption of Cox proportional hazard models (Box-Steffensmeier & Jones, 2004), gender is used as a time-varying covariate. With Cox models, this means that I interact gender with time. This is necessary because men and women exhibit different migration timing patterns. I lack sufficient evidence to specify exactly why men and women have different migration timing patterns, but it is possible that this difference exists due to the timing of forced conscription in Syria, which almost exclusively targets men (Carpenter, 2006). I provide some suggestive evidence for this possibility in the Online appendix. Additional research could clarify how the timing of gendered violence, both violence against men and violence against women, influences the migration timing of men and women.

There are models with and without an interaction term between motivation and opportunity to migrate. In addition, there are models with and without control variables: Neighborhood trust, Neighborhood alert, Total income, Age, College educated, Betrayal, Computer, Ties to international company, and Police ties. Table IV

Table IV.

Determinants of migration timing

	(1)	(2)	(3)	(4)
Wasta	0.447*		0.656*
	(0.21)		(0.27)
No violence witnessed	0.497*		0.430
	(0.23)		(0.27)
No wasta, no violence witnessed		0.213		0.285
		(0.27)		(0.30)
Wasta, violence witnessed		0.118		0.390
		(0.28)		(0.36)
Wasta, no violence witnessed		1.188***		1.254**
		(0.31)		(0.37)
Violence received	0.063	0.031	0.063	0.042
	(0.08)	(0.09)	(0.11)	(0.11)
Violence to family	–0.010	–0.069	–0.078	–0.070
	(0.12)	(0.13)	(0.13)	(0.14)
Armed group ties	–0.066	–0.081	–0.050	–0.056
	(0.05)	(0.05)	(0.06)	(0.06)
High obstacles, high facilitators	0.124	0.036	0.220	0.164
	(0.29)	(0.30)	(0.33)	(0.34)
High obstacles, low facilitators	–0.112	–0.113	–0.199	–0.191
	(0.23)	(0.23)	(0.28)	(0.28)
Low obstacles, high facilitators	0.478	0.353	0.273	0.202
	(0.47)	(0.47)	(0.52)	(0.53)
Time-varying covariate
Male	–0.006	–0.004	0.002	0.002
	(0.01)	(0.01)	(0.01)	(0.01)

Other controls	No	No	Yes	Yes
N	122	122	111	111

Cox model coefficients are reported. Standard errors are in parentheses.

* p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 4.

Survival curve for violence witnessed

Figure 5.

Survival curve for wasta

displays the results. Since the violence witnessed and wasta variables are the only variables with statistically significant coefficients, Figures 4 and 5 display their survival curves. In the Online appendix, I include a correlation matrix of Cox model coefficients from Model 4.

Without an interaction term, H1 and H3 are supported. H2 is not supported. When I add the interaction term, I find that respondents migrated earlier when they both had not witnessed violence and had wasta. Respondents who did not have wasta and did not witness violence did not have significantly different migration timing than those who did not have wasta and did witness violence. In addition, respondents who did have wasta and did witness violence did not have significantly different migration timing than those who did not have wasta and did witness violence.

It is also important that Armed group ties, Violence to family, and Violence received are not statistically significant. The null result for Armed group ties supports a view that more information about armed group activities does not have a systematic relationship with motivation or opportunity for migration. Then, the insignificant coefficients for Violence to family and Violence received support an argument that the counter-intuitive effect of Violence witnessed is not just driven by an endogenous relationship with migration timing. If witnessing violence were just more likely because somebody had stayed home for a longer duration and would have more violent experiences generally, then receiving violence or family members receiving violence should also be more likely as people stay home for a longer duration.

These results support the argument that migration timing requires the combination of motivation and opportunity to facilitate earlier migration. Motivation without opportunity and opportunity without motivation are not enough to significantly affect migration timing.

Conclusion

This article has shown that civilians must have both early motivation and opportunity in order to migrate at an earlier time during conflict. Motivation without opportunity and opportunity without motivation are not enough. This is a key insight for migration research that often only emphasizes one or the other. There are some exceptions, such as work that uses an aspirations–capabilities framework, but it has not yet become standard in migration research.

I also contribute to research on the relationship between violence and migration. Violent experiences trigger crucial psychological (PTG or PTSD) and social processes (narrative ruptures) that influence the timing of motivation to migrate. The combination of locations and targeting types of violence creates the opportunity structure for conflict-induced migration. In Syria, I show that civilians with wasta have more opportunity to safely navigate this opportunity structure and migrate safely.

Future work would benefit from developing methods to obtain individual-level data from both migrants and non-migrants during conflict. Expanding the sample could allow findings to generalize to a broader population. In addition, it could yield a sample with a wider variety of combinations of violent experiences, which could improve comparison of the effects of violence to family, violence received, and violence witnessed. Skype interviews and collaborations with in-country journalists and enumerators present possible solutions to obtain this kind of data (Lund, 2016). Extensions of this work to other contexts should also consider how civilians in those other contexts obtain advantaged positions. Syrians refer to those with advantaged positions as having wasta. Comparable systems include blat in Russia, Big Man Politics in sub-Saharan Africa, and good ol’ boy networks in the United States. Examining these systems could advance understandings of how social systems interact with armed conflict.

Understanding conflict-induced migration requires explaining more than why people migrate. It is also important to explain when they migrate. Improving explanations of migration timing will bring substantial academic and policy benefits for civilians living through the challenges and dangers of armed conflict.

Footnotes

Replication data

The dataset and do-file for the empirical analysis in this article, along with the Online appendix, can be found at http://www.prio.org/jpr/datasets or . All analyses were conducted using Stata 15.

Acknowledgments

I would like to thank Alex Braithwaite, Idean Salehyan, and Burcu Savun for organizing this special issue and providing valuable advice. I would also like to thank the anonymous reviewers. Participants at the Workshop on Violence, Trauma, and Refugees at the University of Hamburg’s Institute of Law and Economics provided valuable feedback. Also, comments from Karen Rasler, Lauren M. MacLean, Armando Razo, Michael McGinnis, Cyanne Loyle, Yehuda Magid, Jennifer N. Brass, Anthony J. DeMattee, and the Indiana University Contentious Politics Workshop were extremely valuable. All errors and omissions are my own.

Funding

For financial support, I thank the Ostrom Workshop at Indiana University and the Project on Middle East Political Science.

ORCID iD

Justin Schon

Notes

References

Adhikari

Prakash

(2012) The plight of the forgotten ones: Civil war and forced migration. International Studies Quarterly 56(3): 590–606.

Adhikari

Prakash

(2013) Conflict-induced displacement, understanding the causes of flight. American Journal of Political Science 57(1): 82–89.

Allport

Gordon

Postman

Leo

(1947) The Psychology of Rumor. New York: Russell & Russell.

Arias

María Alejandra

Ibáñez

Ana María

Querubin

Pablo

(2014) The desire to return during civil war: Evidence for internally displaced populations in Colombia. Peace Economics, Peace Science and Public Policy 20(1): 209–233.

Baines

Erin

Paddon

Emily

(2012) ‘This is how we survived’: Civilian agency and humanitarian protection. Security Dialogue 43(3): 231–247.

Balcells

Laia

Steele

Abbey

(2016) Warfare, political identities, and displacement in Spain and Colombia. Political Geography 51(March): 15–29.

Barbarin

Oscar A

Richter

Linda

DeWet

Thea

(2001) Exposure to violence, coping resources, and psychological adjustment of South African children. American Journal of Orthopsychiatry 71(1): 16–25.

Barnett

Andy

Yandle

Bruce

Naufal

George

(2013) Regulation, trust, and cronyism in Middle Eastern societies: The simple economics of ‘wasta’. Journal of Socio-Economics 44(June): 41–46.

Başoğlu

Metin

(1993) Prevention of torture and care of survivors: An integrated approach. Jama 270(5): 606–611.

10.

Berry

William D

Golder

Matt

Milton

Daniel

(2012) Improving tests of theories positing interaction. Journal of Politics 74(3): 653–671.

11.

Betts

Alexander

(2013) Survival Migration: Failed Governance and the Crisis of Displacement. Ithaca, NY: Cornell University Press.

12.

Blattman

Christopher

(2009) From violence to voting: War and political participation in Uganda. American Political Science Review 103(2): 231–247.

13.

Blattman

Christopher

Annan

Jeannie

(2010) The consequences of child soldiering. Review of Economics and Statistics 92(4): 882–898.

14.

Bleich

Erik

Pekkanen

Robert

(2013) How to report interview data. In: Mosley

Layna

(ed.) Interview Research in Political Science. Ithaca, NY: Cornell University Press, 95–116.

15.

Bloch

Alice

(2007) Methodological challenges for national and multi-sited comparative survey research. Journal of Refugee Studies 20(2): 230–247.

16.

Box-Steffensmeier Janet

Jones

Bradford S

(2004) Event History Modeling: A Guide for Social Scientists. Cambridge: Cambridge University Press.

17.

Byman

Daniel

(2016) Understanding the Islamic State: A review essay. International Security 40(4): 127–165.

18.

Caerus (2014) Mapping the conflict in Aleppo, Syria (https://s3.amazonaws.com/fmg_static/aleppo/Caerus_AleppoMappingProject_SecurityConditions.pdf).

19.

Calhoun

Lawrence G

Tedeschi

Richard G

(2014) Handbook of Posttraumatic Growth: Research and Practice. London: Routledge.

20.

Carpenter, R Charli (2006) Recognizing gender-based violence against civilian men and boys in conflict situations. Security Dialogue 37(1): 83–103.

21.

Corbett

Justin

(2011) Learning from the Nuba: Civilian resilience and self-protection during conflict. Copenhagen: Local to Global Protection (https://www.local2global.info/wp-content/uploads/SK_Nuba_L2GP_report_final.pdf).

22.

Czaika

Mathias

Kis-Katos

Krisztina

(2009) Civil conflict and displacement: Village-level determinants of forced migration in Aceh. Journal of Peace Research 46(3): 399–418.

23.

Davenport

Christian

Moore

Will

Poe

Steven

(2003) Sometimes you just have to leave: Domestic threats and forced migration, 1964–1989. International Interactions 29(1): 27–55.

24.

Debos

Marielle

(2016) Living by the Gun in Chad: Combatants, Impunity and State Formation. London: Zed.

25.

de Haas

Hein

(2014) What drives human migration? In: Anderson

Bridget

Keith

Michael

(eds) Migration: A Compas Anthology. Oxford: COMPAS.

26.

De Smedt

Johan

(2009) ‘No Raila, No peace!’ Big man politics and election violence at the Kibera grassroots. African Affairs 108(433): 581–598.

27.

Eitle

David

Jay Turner

(2002) Exposure to community violence and young adult crime: The effects of witnessing violence, traumatic victimization, and other stressful life events. Journal of Research in Crime and Delinquency 39(2): 214–237.

28.

Engel

Stefanie

Ibáñez

Ana María

(2007) Displacement due to violence in Colombia: A household-level analysis. Economic Development and Cultural Change 55(2): 335–365.

29.

Erdogan

Murat

(2016) Study: Educated Syrians in Europe, illiterate ones in Turkey (https://www.dailysabah.com/nation/2016/03/05/study-educated-syrians-in-europe-illiterate-ones-in-turkey).

30.

Foster

Holly

Brooks-Gunn

Jeanne

(2015) Children’s exposure to community and war violence and mental health in four African countries. Social Science & Medicine 146(December): 292–299.

31.

Fujii

Lee Ann

(2011) Killing Neighbors: Webs of Violence in Rwanda. Ithaca, NY: Cornell University Press.

32.

Gleditsch

Nils Petter

Wallensteen

Peter

Eriksson

Mikael

Sollenberg

Margareta

Strand

Håvard

(2002) Armed conflict 1946–2001: A new dataset. Journal of Peace Research 39(5): 615–637.

33.

Hall

Natasha

(2016) Waiting for No One: Civilian Survival Strategies in Syria. Washington, DC: Center for Civilians in Conflict.

34.

Harpvicken

Kristian Berg

(2009) Social Networks and Migration in Wartime Afghanistan. New York: Palgrave Macmillan.

35.

Jacobsen

Karen

Landau

Loren

(2003) The dual imperative in refugee research: Some methodological and ethical considerations in social science research on forced migration. Disasters 27(3): 95–116.

36.

Janoff-Bulman

Ronnie

(1992) Shattered Assumptions: Towards a New Psychology of Trauma. New York: Free.

37.

Jose

Betcy

Medie

Peace A

(2016) Civilian self-protection and civilian targeting in armed conflicts: Who protects civilians? Oxford Research Encyclopedia of Politics. https://doi.org/10.1093/acrefore/9780190228637.013.216.

38.

Joseph

Stephen

(2013) What Doesn’t Kill Us: The New Psychology of Posttraumatic Growth. New York: Basic Books.

39.

Kalyvas

Stathis N

(2006) The Logic of Violence in Civil War. New York: Cambridge University Press.

40.

Kaminer

Debra

Hardy

Anneli

Heath

Katherine

Mosdell

Jill

Bawa

Umesh

(2013) Gender patterns in the contribution of different types of violence to posttraumatic stress symptoms among South African urban youth. Child Abuse & Neglect 37(5): 320–330.

41.

Kaplan

Oliver

(2013a) Nudging armed groups: How civilians transmit norms of protection. Stability: International Journal of Security and Development 2(3). https://doi.org/10.5334/sta.cw.

42.

Kaplan

Oliver

(2013b) Protecting civilians in civil war: The institution of the ATCC in Colombia. Journal of Peace Research 50(3): 351–367.

43.

Kaplan

Oliver

(2017) Resisting War: How Communities Protect Themselves. Cambridge: Cambridge University Press.

44.

Kaya

Ayhan

Kirac

Aysu

(2016) Vulnerability assessment of Syrian refugees in Istanbul. Support to Life (https://data2.unhcr.org/en/documents/details/54518).

45.

Kingsley

Patrick

(2016) Turkey blocks Syrian refugees from resettlement in the US – for having degrees. Guardian 19 September (https://www.theguardian.com/world/2016/sep/19/turkey-syrian-refugees-resettlement-blocked-us-degrees).

46.

Kulkarni

Madhur R

Graham-Bermann

Sandra

Rauch

Sheila AM

Seng

Julia

(2011) Witnessing versus experiencing direct violence in childhood as correlates of adulthood PTSD. Journal of Interpersonal Violence 26(6): 1264–1281.

47.

Lister

Charles R

(2016) The Syrian Jihad: Al-Qaeda, the Islamic State and the Evolution of an Insurgency. Oxford: Oxford University Press.

48.

Lombard

Louisa

(2013) Navigational tools for Central African roadblocks. Political and Legal Anthropology Review 36(1): 157–173.

49.

Lubkemann

Stephen

(2008) Involuntary immobility: On a theoretical invisibility in forced migration studies. Journal of Refugee Studies 21(4): 454–475.

50.

Lund

Aron

(2016) Into the Tunnels: The Rise and Fall of Syria’s Rebel Enclave in the Eastern Ghouta. New York: Century Foundation (https://tcf.org/content/report/into-the-tunnels/?agreed=1).

51.

Lust

Ellen

(2009) Competitive clientelism in the Middle East. Journal of Democracy 20(3): 122–135.

52.

Lynch

Marc

(2013) The political science of Syria’s war. In: The Political Science of Syria’s War. POMEPS Briefing 22. Washington, DC: POMEPS, 3–7.

53.

MacDonell

Kathleen Watson

(2012) The combined and independent impact of witnessed intimate partner violence and child maltreatment. Partner Abuse 3(3): 358–378.

54.

Massey

Douglas

Bohra-Misha

Pratikshya

(2011) Individual decisions to migrate during civil conflict. Demography 48(2): 401–424.

55.

Massey

Douglas

Silva

Adriana Carolina

(2014) Violence, networks, and international migration from Colombia. International Migration. https://doi.org/10.1111/imig.12169.

56.

McAdam

Doug

Tarrow

Sidney

Tilly

Charles

(2008) Methods for measuring mechanisms of contention. Qualitative Sociology 31(4): 307–331.

57.

McHugo

John

(2015) The roots of Syria’s tragedy. Aljazeera (https://www.aljazeera.com/indepth/opinion/2015/10/roots-syria-tragedy-151004075956712.html).

58.

Melander

Erik

Öberg

Magnus

(2006) Time to go? Duration dependence in forced migration. International Interactions 32(2): 129–152.

59.

Meyer

David S

(2004) Protest and political opportunities. Annual Review of Sociology 30: 125–145.

60.

Moore

Will

Shellman

Stephen

(2004) Fear of persecution: Forced migration, 1952–1995. Journal of Conflict Resolution 48(5): 723–745.

61.

Moore

Will

Shellman

Stephen

(2006) Refugee or internally displaced person? To where should one flee? Comparative Political Studies 39(5): 599–622.

62.

Most

Benjamin A

Starr

Harvey

(2015) Inquiry, Logic, and International Politics: With a New Preface by Harvey Starr. Columbia, SC: University of South Carolina Press.

63.

Olsen

Kristine A

(2014) Telling our stories: Narrative and framing in the movement for same-sex marriage. Social Movement Studies 13(2): 248–266.

64.

Paxton

Keisha Carr

LaVome Robinson

Shah

Seema

Schoeny

Michael E

(2004) Psychological distress for African-American adolescent males: Exposure to community violence and social support as factors. Child Psychiatry and Human Development 34(4): 281–295.

65.

Pearlman

Wendy

(2016) Narratives of fear in Syria. Perspectives on Politics 14(1): 21–37.

66.

Peleg

Ilan

Waxman

Dov

(2011) Israel’s Palestinians: The Conflict Within. Cambridge: Cambridge University Press.

67.

Pemberton

Antony

Aarten

Pauline GM

(2018) Narrative in the study of victimological processes in terrorism and political violence: An initial exploration. Studies in Conflict & Terrorism 41(7): 541–556.

68.

Physicians for Human Rights (2017) Anatomy of a Crisis: A Map of Attacks on Health Care in Syria (https://s3.amazonaws.com/PHR_syria_map/web/index.html, accessed 30 November 2017).

69.

Pottier

Johan

(2006) Roadblock ethnography: Negotiating humanitarian access in Ituri, Eastern DR Congo, 1999–2004. Africa: Journal of the International African Institute 76(2): 151–179.

70.

Ramady

Mohamed

(2015) The Political Economy of Wasta: Use and Abuse of Social Capital Networking. Basel: Springer.

71.

Randolph

Suzanne M

Koblinsky

Sally A

Roberts

Debra D

(1996) Studying the role of family and school in the development of African American preschoolers in violent neighborhoods. Journal of Negro Education 65(3): 282–294.

72.

Rizkalla

Niveen

Segal

Steven P

(2017) Well-being and posttraumatic growth among Syrian refugees in Jordan. Journal of Traumatic Stress 31(2): 213–222.

73.

Rosen

Eva

(2017) Horizontal immobility: How narratives of neighborhood violence shape housing decisions. American Sociological Review 82(2): 270–296.

74.

Rubin

Jacqueline

Moore

Will

(2007) Risk factors for forced migrant flight. Conflict Management and Peace Science 24(2): 85–104.

75.

Salehyan

Idean

Gleditsch

Kristian Skrede

(2006) Refugees and the spread of civil war. International Organization 60(2): 335–366.

76.

Salzinger

Suzanne

Feldman

Richard S

Stockhammer

Tanya

Hood

Julie

(2002) An ecological framework for understanding risk for exposure to community violence and the effects of exposure on children and adolescents. Aggression and Violent Behavior 7(5): 423–451.

77.

Schon

Justin

(2015) Focus on the forest, not the trees: A changepoint model of forced displacement. Journal of Refugee Studies 28(4): 437–467.

78.

Schon

Justin

(2016) The centrality of checkpoints for civilians during conflict. Civil Wars 18(3): 281–310.

79.

Shibutani

Tamotsu

(1966) Improvised News: A Sociological Study of Rumor. New York: Bobbs-Merrill.

80.

Shields

Nancy

Fieseler

Carlie

Gross

Cassie

Hilburg

Meredith

Koechig

Nick

Lynn

Randy

Williams

Brandi

(2010) Comparing the effects of victimization, witnessed violence, hearing about violence, and violent behavior on young adults. Journal of Applied Social Science 4(1): 79–96.

81.

Shields

Nancy

Nadasen

Kathy

Pierce

Lois

(2009) A comparison of the effects of witnessing community violence and direct victimization among children in Cape Town, South Africa. Journal of Interpersonal Violence 24(7): 1192–1208.

82.

Steele

Abbey

(2019) Civilian resettlement patterns in civil war. Journal of Peace Research 56(1): 28–41.

83.

Tedeschi

Richard G

Calhoun

Lawrence G

(1996) The posttraumatic growth inventory: Measuring the positive legacy of trauma. Journal of Traumatic Stress 9(3): 455–471.

84.

Tilly

Charles

McAdam

Douglas

Tarrow

Sidney

(2001) Dynamics of Contention. Cambridge: Cambridge University Press.

85.

Uehara

Edwina S

Chalmers

Deborah

Jenkins

Esther J

Shakoor

Bambade H

(1996) African American youth encounters with violence: Results from the community mental health council violence screening project. Journal of Black Studies 26(6): 768–781.

86.

Utas

Mats

, ed. (2012) African Conflicts and Informal Power: Big Men and Networks. London: Zed.

87.

Uzonyi

Gary

(2014) Unpacking the effects of genocide and politicide on forced migration. Conflict Management and Peace Science 31(3): 225–243.

88.

Werdel

Mary Beth

Wicks

Robert J

(2012) Primer on Posttraumatic Growth: An Introduction and Guide. Hoboken, NJ: John Wiley & Sons.

89.

Williams

Nathalie

(2013) How community organizations moderate the effect of armed conflict on migration in Nepal. Population Studies: A Journal of Demography 67(3): 353–369.

90.

Wood

Elisabeth Jean

(2000) Forging Democracy from Below: Insurgent Transitions in South Africa and El Salvador. New York: Cambridge University Press.

91.

Wood

Elisabeth Jean

(2003) Insurgent Collective Action and Civil War in El Salvador. New York: Cambridge University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.05 MB