Investigation of the Effects of Violence Experience During Political Demonstrations

In this study, predictors of post-traumatic stress symptom levels (PTSSL) and post-traumatic growth levels (PTGL) resulting from the experience of violence were investigated. The sample of the study consisted of 514 Gezi Park demonstrators. Participants completed measures assessing stress symptoms, post-traumatic growth, social support and beliefs about the world as well as the open-ended event specific questions. Results showed that being politically active, psychologically prepared and experiencing mild levels of violence were related with decreased PTSSL individually but not in combination as the literature suggested. The two hierarchical regression analyses showed that: (1) PTSSL were predicted by violence exposure levels, perceived social support from significant others and “randomness” and “self-worth” beliefs about the world; (2) PTGL were predicted by violence exposure levels, total amount of time spent at the demonstrations and “benevolence” and “justice” beliefs about the world. These findings suggest that psychological preparedness might be an important variable in violence experience regarding human masses. Also, violence exposure levels and duration of participation seems to be important event- specific variables. Lastly, political activism needs to be more precisely operationalized and measured in future studies.

ideological commitment mediates between political violence and psychological wellbeing (Muldoon, Schmid, & Downes, 2009). In terms of psychopathologies, subjective meaning is an important predictor, whereas weakness in that meaning, severe violence, and captivity were related with PTSD symptoms (Paker, 2000).
After a violent experience, a non-activist group's political commitment may increase (Paker, 2000).

World Assumptions
First systematically studied by Janoff-Bulman (1989), world assumptions refer to the unquestioned assumptions of people on themselves and the world. Traumatic events cause people who assume that the world is predictable and safe to question their beliefs and realize their vulnerability. According to Janoff-Bulman (1989), people have three fundamental assumptions: (1) the overall benevolence of the world, (2) the meaningfulness of the world, and (3) self-worthiness. The two sub-assumptions of the benevolence of the world refer to the benevolence of the world as an entity, including the benevolence of the people in it. The meaningfulness of the world contains justice, controllability, and randomness as sub-assumptions. Lastly, self-worthiness comprises self-worth, self-control, and luck (Janoff-Bulman, 1989). After a traumatic event, these assumptions are shattered; acknowledging their vulnerability, survivors are believed to question their positive beliefs of their self and the world (Janoff-Bulman, 1989). In the aftermath of trauma, these assumptions are rebuilt, forming new assumptions that integrate the traumatic experience in lieu of previous assumptions (Janoff-Bulman, 1989, 1992. Janoff-Bulman (1992) further reported that after natural disasters, the assumptions benevolence of the world, randomness, and control are affected the most, whereas after human-made disasters, the assumptions benevolence of the world and the people and self-worthiness are affected the most. Meanwhile, research conducted with tsunami survivors revealed changes in their justice assumptions to be related to elevated PTSD symptoms and decreased life quality (Nygaard & Heir, 2012).

Social Support
Throughout literature, "perceived" social support has gained increased attention (Eker, Arkar, & Yaldız, 2001;Lazarus, 1990;Yap & Devilly, 2004) compared to "received" social support, which are the two types of social support suggested (Heller, Swindle, & Dusenbury, 1986). Perceived social support protects trauma survivors against depression, anxiety, and stress; whereas its deficiency has been related to elevated distress and plays an important role in trauma symptomatology (Yap & Devilly, 2004). Moreover, perceived social support has an effect on the relationship between the traumatic event and post-traumatic stress, buffering the effects of the traumatic event, although the contribution of social support and its relation with other variables remain unexplained (Haden, Scarpa, Jones, & Ollendick, 2007).
In collective traumas, social support is an important buffer at all societal levels (Hoffman & Kruczek, 2011). A study on political violence revealed that apart from the importance of social support, the individual's level of satisfaction with the perceived social support buffers the impacts of political violence on mental health (Punamäki et al., 2005). Moreover, studies suggest that collective traumas facilitate higher social support and broaden coping opportunities (Luszczynska, Benight, & Cieslak, 2009). However, there is still controversy regarding findings on the buffering effects of social support on mental health after collective traumas (Grills-Taquechel, Littleton, & Axsom, 2011). For example, a study on the 9/11 attacks found no relationship between social support and physical and mental health (Adams, Boscarino, & Galea, 2006).

Gezi Park Demonstrations as a Traumatic Experience
The Gezi Park demonstrations began as an ordinary environmental protest and spread all over Turkey, becoming a social movement with the active participation of large groups of citizens. Security forces' use of deterrents, such as pepper gas, batons, rubber bullets, and pressurized water, to disperse the crowds shifted the environmental focus of the demonstrations to an anti-government stance. Amid the demonstrations, the  (Ete & Taştan, 2013). The psychological effects of the physical and psychological violence inflicted, however, have not been explored comprehensively. Violent acts throughout the demonstrations have been described as components of a human-made disaster (Kaptanoğlu, 2013). The complexity of state violence, which includes human rights violations, is related to its political context and consequences. The body of literature on human rights violations in Turkey (e.g., Başoğlu et al., 1994Başoğlu et al., , 1997Paker, 2000) has taken the contexts into account, in terms of the visibility of the violence to which protesters were exposed.

Procedure
The Ethics Board of the Hacettepe University Senate evaluated and approved the study. Data were gathered through an online survey system by snowball sampling. To reach the people who could have possibly participated in the Gezi Park demonstrations, social media channels, various e-mail groups, political gatherings/groups/ parties/solidarity networks, and post-demonstrations groups were contacted. Anonymity was guaranteed for participants in all phases of the study. The participants, who agreed to participate voluntarily and declared this by signing the voluntary participation form, filled out the demographic information form followed by the event information form and the scales. The total amount of time needed to complete the entire study was about 15 minutes; however, no time limit was established. No question could be skipped. The last window included a brief appreciation message for participating in the study.

Measures Demographic Information Form
To obtain a detailed profile of the participants, a comprehensive demographic information form was designed.
In this form, participants were asked to state their age, gender, marital status, education level, occupation, residence, people they lived with (household sharing), previous psychiatric diagnoses, physical illness history, and whether they had a paid job. The main aim of asking this information was to see if there was a certain pattern of demographic features belonging to the demonstrators. However, no unique pattern was found and the characteristics was parallel with the commonsense opinion that the demonstrators were young adults with education level above the average, exhibiting no gender difference.

Event Information Form
The event information form was created to gather event-specific information. The participants were asked about the total amount of time they had spent at the demonstrations and the frequency of their participation in the demonstrations, as well as their behavior after police intervention. Only the former was used due to latter two not working properly as if the questions were not operationalized to cover the concepts. Throughout the literature, no scale was found to measure political activism and psychological preparedness for violence directly. In studies that dealt with psychological trauma and political activism, political activism was addressed within the demographics form with event-specific questions (Başoğlu et al., 1994;Kanagaratnam, Raundalen, & Asbjørnsen, 2005;Paker, 2000). In the present work, political activism and psychological preparedness for violence were approached in a similar way but as a distinct section, using an event information form.
Participants were asked about their experience being in police custody, under arrest, in prison, or tortured for political reasons (1) before the Gezi Park demonstrations and (2) during the demonstrations. The participants were also asked to evaluate themselves in terms of political activity (1) before the Gezi Park demonstrations and (2) during the demonstrations. Moreover, participants were asked about their preparedness for the violence they faced during the demonstrations. At the end of this form, additionally, the first part of the Post-Traumatic Stress Diagnostic Scale (PTSDS) was added (Foa, Cashman, Jaycox, & Perry, 1997). This part identified Kırseven & Işıklı 483 traumatic event types (natural disaster, war, accident, etc.) as well as other possible traumatic events that participants might have gone through after the demonstrations.

Stress Symptoms Sub-Scale of the PTSDS
The scale was developed by Foa et al. (1997) to identify post-traumatic stress symptom levels in accordance with DSM-IV. The scale consists of 17 Likert-type items. Total scores range between 0 and 51; with higher scores indicating higher symptom levels. The internal consistency of the original scale was Cronbach's α = .92, and the test-retest validity coefficient for items was .83 (Foa et al., 1997). Isıklı (2006) completed a Turkish adaptation of the scale, and the Cronbach's alpha obtained for the total scale was .93. For this study, Cronbach's alpha for the total scale was .91, and the Cronbach's alpha for the "re-experiencing," "avoidance," and "hyperarousal" sub-factors was .83, .79 and .86, respectively.

PTG Inventory
Tedeschi and Calhoun developed the scale in 1996, with the aim of identifying growth in the aftermath of a traumatic event (Tedeschi & Calhoun, 1996). The scale consists of 21 items scored between 0 and 6. Total scores range between 0 and 105, with higher scores indicating a higher amount of growth. In the original study, internal consistency for the entire scale was α = .90, whereas for the sub-scales, it ranged between α = .67 and α = .85. The test-retest validity correlation coefficient was .71. Dürü developed a Turkish adaptation of the scale in 2006 (Dürü, 2006). The internal consistency coefficient obtained was Cronbach's α = .93. For the present study, the internal consistency coefficient for the total scale was Cronbach's α = .95. The internal consistency coefficients for the sub-factors obtained were Cronbach's α = .87 for "relating to others," α = .85 for "new possibilities," α = .83 for "personal strength," α = .90 for "spiritual change," and α = .63 for "appreciation of life."

Multidimensional Scale of Perceived Social Support
Zimet, Dahlem, Zimet, and Farley (1998) developed the scale in 1998 to measure perceived social support from three different sources: family, friends, and significant others. The Likert-type scale consists of 12 items scored between 1 and 7. Scores range between 12 and 84, with higher scores representing higher perceived social support. In the original study, the internal consistency coefficient for the "family" subscale was Cronbach's α = .91, for "friends," α = .87, and for "significant others," α = .85 (Zimet, Dahlem, Zimet, & Farley, 1988).
In the current study, dimensions of social support was taken separately with the aim of understanding specific sources of support that might be more salient in collective violence experience.

The World Assumptions Scale
The scale was developed by Janoff-Bulman in 1989 with the purpose of identifying and measuring the effects of traumatic stress on fundamental assumptions. The total scale is composed of 32 items scored between 1 and 6 (Janoff-Bulman, 1989). The original form consisted of seven sub-scales: benevolence of the world,  (Yılmaz, 2008). For this study, the total scale internal consistency coefficient was Cronbach's α = .84, whereas those for the subscales were .88, .37, .83, .68, .61, and .90, respectively.
In the current study, sub-scales of WAS was taken individually throughout the analysis. The reason behind doing so was that, having assumed the world assumptions as a whole would be shattered, each dimension was thought to be referring to a different concept that might be emphasizing on a different aspect of world assumptions specific to the event. Thus, it was done to understand the specific world assumptions that might be shattered with the experience of violence during political demonstrations.

Data Analysis
The data were analyzed using SPSS For the political activism variable, participants were asked to mark the best-fitting option (being an activist, being a member of a political group but not being politically active, being politically active but not being member of a political group and being politically inactive). Before starting the analysis, the first two answers were merged due to both groups being parts of a political group and ideology. New groups were activists (members of a political group), politically active (not member of a political group) and politically inactive. For the preparedness for violence variable, participants were directly asked psychologically how prepared would they define themselves to the violence witnessed. The 5 options were well-prepared, prepared, neither prepared nor unprepared, not prepared and not prepared at all.
The total amount of time spent at the demonstrations was asked directly to the participants. They were expected to answer in days and hours; and the maximum hours one can participate in the demonstrations was 6 hours a day where the duration of demonstrations was taken as 20 days (maximum number of days one can participate in the demonstrations). So, the maximum amount of time that could be spent at the demonstrations were 120 hours.
In the following step, to examine the effects of political activism and psychological preparedness for violence on PTSSL, a 3 (activist / politically active / politically inactive) x 5 (well-prepared / prepared / neither prepared nor unprepared / not prepared / not prepared at all) analysis of variance (ANOVA) was conducted. Lastly, hierarchical regression analyses were carried out to identify the predictors of PTSSL and PTG. As psychological preparedness was found to have a significant effect on PTSSL, it was included as a control variable in the first step of the regression analysis together with the pre-event and post-event indices.

Effects of Political Activism and Psychological Preparedness for Violence on PTSSL
No significant interaction effect between political activism and psychological preparedness for violence was found, F(8, 514) = 1.06, p = .388, η p 2 = .02; Table 1. However, the main interaction effects for political activism, F(2, 514) = 3.31, p = .037, η p 2 = .01, and psychological preparedness for violence, F(4, 514) = 12.47, p < .001, η p 2 = .09, on PTSSL were found to be significant. Between-group differences were examined through the Tukey test, using Bonferroni correction (Table 3). The PTSSL of the activist group (M = 32.8) were found to be significantly higher than that of the politically inactive group (M = 28.7). However, further analysis were carried through splitting the data in terms of political activism levels and on none of the variables did political activism indicate meaningful results. Even though the main effect of political activism was significant, with the data in hand, no further inference could be made. Thus, political activism variable was not included in the regression models.
Violence During Demonstrations 486 Note. Different subscripts refer to the significant differences between means.
The main interaction effect for psychological preparedness for violence on PTSSL was found to be significantly lower for the "well-prepared" (M = 24.1) and prepared groups compared with the "not prepared" (M = 34.0) and "not prepared at all" (M = 38.7) groups, separately. Similarly, the PTSSL of the "neither prepared nor unprepared" (M = 29.2) group were found to be significantly lower compared with the "not prepared" (M = 34.0) and "not prepared at all" (M = 38.7) groups.

Hierarchical Regression Analysis for the Predictors of PTSSL
In the first step of the regression equation, traumatic experience levels before and after the demonstrations and psychological preparedness for violence were entered as control variables (Table 3). In the second step, violence exposure levels and total amount of time spent at the demonstrations were entered. Lastly, sub-scales of MSPSS and WAS were entered as separate blocks into the regression equation.

Discussion
The study found no significant interaction effect between political activism and psychological preparedness for violence. However, the main effects for political activism and psychological preparedness for violence on PTSSL were significant. Hierarchical regression analyses revealed that after statistically controlling for traumatic experience levels before and after the demonstrations and psychological preparedness for violence, (1) PTSSL were predicted by violence exposure levels, the MSPSS sub-scale "significant others," and the WAS sub-scales "randomness" and "self-worth"; (2) PTGL were predicted by violence exposure levels, time spent at the demonstrations, and the WAS sub-scales "benevolence", "luck," and "justice."

Kırseven & Işıklı 489
In the current study, results on political activism and psychological preparedness for violence, the two variables frequently studied together were found to have a significant effect on PTSSL individually but not together, conflicting with previous studies (Başoğlu et al., 1994(Başoğlu et al., , 1997Paker, 2000). Previous research proposed that possible repetitive traumatic stress throughout political struggles, political activists might acquire immunity to traumatic stress (Başoğlu et al., 1994). However, in the present study, the opposite tendency of this immunity against stress was found: higher PTSSL were seen in the politically active group than the inactive group.
This finding may be the result of the politically active group's higher exposure to violence before and after demonstrations. Prior studies on political activism studied the effects of being in police custody or jail for political reasons (Başoğlu et al., 1994(Başoğlu et al., , 1997Paker, 2000). However, state violence during the Gezi Park demonstrations took place at a more public sphere than the violence in prison or under surveillance. We assume that this public nature brought about higher social support, like social sharing and cooperation. In addition, during the demonstrations, deterrents were largely used as tools for violence, whereas more systematic ways of violence are applied in prisons and under surveillance (Paker, 2000). Thus, event-specific factors seemed to affect the results and explain the failure in covering some concepts. For the political activism variable, it is possible that the operationalization and measuring has failed to grasp the concept. It is highly recommended for future studies to well-operationalize and standardize the concept and measurement of political activism with its all dimensions. This study has mostly focused on the 'being part of a political group and ideology' but it is possible to say that there was missing aspects and levels of it.
Meanwhile, as psychological preparedness for violence tended to increase, PTSD symptoms tended to decrease. Previous research proposed that psychological preparedness relates to the severity of distress during torture and other severe psychological issues. Thus, psychological preparedness functions like a "vaccination" (Başoğlu et al., 1997). The present finding on preparedness might confirm this "vaccination" effect, although the phenomenon needs to be understood more profoundly. Findings on the predictors of PTSSL showed that as the violence index scores increased, PTSSL increased. As the violence index contains many physical violence statements, literature (North et al., 1999;Paker, Paker, & Yüksel, 1992) on violence experiences indicated that physical damage would result in higher PTSSL.
The MSPSS sub-scale "significant others" was found to be another predictor of PTSSL; higher social support perceived from significant others related with lower PTSSL. This result may relate with factors specific to the sample, mainly age. The participants were mostly in their early adulthood (M age = 27.80, SD = 7.50), in which the need for close relationships is significant, as pointed out in studies (Erikson & Erikson, 1998). The WAS sub-scales "randomness" and "self-worth" likewise predicted PTSSL. This finding may be related with the attributions on the meaningfulness of the world, which is encompassed in people's beliefs on the distribution of outcomes (Janoff-Bulman, 1989). A sub-dimension of this belief is randomness, in which people believe that specific events and consequences happen randomly; people with strong randomness beliefs have the belief of having no control over negative events (Janoff-Bulman, 1989). Viewed from this perspective, the Gezi Park demonstrations were a collective movement against the political authority and action to change the perceived injustice (Doğan, 2014); thus, demonstrators with high randomness belief and low PTSSL may have less belief in their potential to change the injustice. Moreover, the "self-worth" sub-scale indicates people's beliefs about themselves; namely, to what extent they see themselves as good, moral, and worthy, how properly they behave, and to what extent they are able to protect themselves from bad luck (Janoff-Bulman, 1989). This study showed that a decrease in demonstrators' belief of self-worth related with PTSSL.

Violence During Demonstrations 490
Elevated violence exposure indicating PTG was parallel with the literature. PTG and PTSSL could be present simultaneously, with some amount of stress necessary for growth . Meanwhile, the total amount of time spent at demonstrations predicting growth may be related to the severity of the traumatic event. The total amount of time spent at the demonstrations predicted PTG, but not PTSSL, which might be explained by the findings that PTG occurs after a long period (Tennen & Affleck, 1998). Those with present PTG findings might have suffered from elevated PTSSL in the past, indicating that PTG is not the result of post-traumatic stress but the process occurring over time.
The first sub-scale of the WAS predicting PTG was benevolence of the world. According to benevolence assumption, people are good, kind, and helpful, and the world is a good and safe place (Janoff-Bulman, 1989).
Findings on attenuated benevolence assumption predicting PTG may point to the shattered assumptions owing to trauma being rebuilt.
Another sub-scale was luck found to predict PTG, in which the world is thought to work according to chance and only the lucky people could get good outcomes (Janoff-Bulman, 1989 Ete & Taştan, 2013). As widely discussed in the days of demonstrations, these findings might be indicating to a generation's political behavior. These young cohorts seemed to express their political demands and attitudes in a more expressive way than the older cohorts. It is a matter of interest to explore this difference through a longitudinal method to see if this is a result of age or generational difference in terms of political attitudes and behaviors.
The psychological viewpoint on political traumas is based on contextual assumptions (Montiel, 2000). The first assumption is on traumatic events being intense but short. In the present study, when referring to the experience of violence, participants mostly mentioned a short time frame in which the experience was intense.
The second assumption is on the effects of the political context on traumatic event survivors; it was assumed that the political context was evil itself according to the survivors (Montiel, 2000). During the demonstrations, the political context might have been perceived as so evil that despite the massive violence, it took almost one and a half months for street demonstrations to subside (Ete & Taştan, 2013). The last assumption is on the relationship between survivors and the political context (Montiel, 2000). As mentioned by Suedfeld (1997), survivors are not only victims, but also have the potential to be modifiers of the political context. When studying trauma, especially political trauma, the contextual perspective needs to be considered as well. Being active subjects in the political context might have recuperative effects and might relate with PTG. Humor, a collective way of coping during demonstrations, is thought to have a healing role. In this sense, Garrick (2006) mentioned that humor had healing effects in the aftermath of trauma.

Kırseven & Işıklı 491
Limitations Firstly, when the data collection of the present study was completed, almost two years had passed since the Gezi Park demonstrations. Although subjects like war, migration, and torture are frequently examined retrospectively throughout literature, the current results might have been influenced with the elapsed time.
Meanwhile, as previously mentioned, for PTG to be observed, the passing of time is seen to be necessary, making longitudinal studies or repeated measures studies more reliable when studying PTSSL and PTG together.
Although the study mostly relied on analysis from quantitative data, the qualitative data gathered were rich and contributed significantly to the study. Approaching multidimensional subjects both through quantitative and qualitative forms could make the information richer. However, the standardization of the qualitative data is challenging. When working with political activism and psychological preparedness, it is recommended that data be collected in a way that is detailed, systematic, and suitable to the nature of the trauma.
During the Gezi Park demonstrations, many people were exposed to violence through the media or were secondary victims in some other ways. Focusing on the secondary traumatization aspect might be enlightening.
As Becker (1995) mentioned, the notion of PTSD might have deficiencies in grasping traumas within the sociopolitical context. In this sense, taken as "post" traumatic, PTSD assumes trauma occurring at a certain time interval with a beginning and an end; however, in the case of many traumatic events, trauma might be perpetual. Moreover, the tendency to diagnose survivors of traumas with a "disorder" has been criticized, as the symptoms of the survivors may be the results of fundamental ethics violations (Becker, 1995). Thus, studying the effects of the traumatic event solely through PTSD might engender a narrow understanding of the phenomenon. In addition, it has been suggested that reactions to the traumatic event have a culture-specific aspect (Shannon et al., 2015). PTSD studies can grasp the symptoms beyond the cultural factors; however, culture-prone dimensions might be left out.

Funding
The authors have no funding to report.