Palestinian Solidarity Action: The Dynamics of Politicized and Religious Identity Patterns Among Student Activists

This study attempted to explain the factors that lead Muslim student activists to participate in Palestinian solidarity actions by testing the Social Identity Model Collective Action model (SIMCA, van Zomeren, Postmes, & Spears, 2008). A survey of 303 student members/administrators of Islamic organizations was conducted. The sample was obtained from more than seven Islamic-based student organizations. Collective solidarity actions were comprised of peaceful actions such as demonstrations, protests, and petition signings. The model involved two identities (politicized and religious) and two mediators (group efficacy and group-based anger). Results of the Structural Equation Modeling (SEM) analysis suggest that politicized identity, as indicated by strength of participants’ affiliations with Islamic movement organizations, predicts solidarity action intention more effectively than religious identity. Other study findings demonstrated that group efficacy is a significant partial mediator of the interaction between politicized and religious identities, and collective action. Religious identity has a stronger interaction with collective action than politicized identity within the partial mediating effect of group efficacy. Meanwhile group-based anger does not influence the desire to engage in collective action either directly or as a mediator. Aksi solidaritas Palestina: Dinamika Pola Identitas Terpolitisasi dan Keagamaan pada Aktivis Mahasiswa


Introduction
In recent years, demonstrations in the name of religion have often occurred in Indonesia in various ways and forms. One example are acts of solidarity with Palestine. Similar actions have been carried out in various regions throughout Indonesia, such as the speech of solidarity at the infamous Hotel Indonesia Roundabout in Jakarta (Ferdyansyah, 2017), the fundraiser and mass march of thousands in West Sumatra (Tirta, 2017), and a long march action in Banjarmasin (Ramadhani, 2017). In social psychology, these actions are classified as collective action. The operationalization of collective action includes attitudes, intentions, and tendencies of participatory behavior, reports of past actions, and real behaviors in collective action settings (van Zomeren & Iyer, 2009).
There are three main theories that explain collective action (Shi, Hao, Saeri, & Cui, 2015;van Zomeren & Iyer, 2009), namely, Relative deprivation theory (RDT; Smith, Pettigrew, Pippin, & Bialosiewicz, 2012), resource mobilization theory (RMT; McCarthy & Zald, 1977), and social identity theory (SIT; Tajfel & Turner, 1979).  attempted to combine the three theories in a model called social identity model of collective action (SIMCA), comprising three causative factors in the form of group-based anger, group efficacy, and social identity, which are included in an integrative model. The focus of SIMCA is on the centrality of social identity factors, especially politicized identity.
According to Simon and Klandermans (2001), politicized identity is a form of identity that comprises an individual's membership of a group that consciously fights for power on behalf of the group in a broad social context.  simplified this concept, defining it as the identity of an "activist" who uses it to carry out collective action. A meta-analysis by  also discovered that politicized identity has a stronger influence (effect size value) than non-politicized identity upon collective action. On the other hand, the role of religious identity can explain a lot of radical collective action and violence (Basedau, Strüver, Vüllers, & Wegenast, 2011;Duffy & Toft, 2007;Hirsch-hoefler, Canetti, & Eiran, 2016). Religious identity is also manifested as either normative (peaceful) or nonnormative (violent) collective action depending on the purpose of the action, its positive acceptance, and threats to identity (Phalet, Baysu, & Verkuyten, 2010). Another study dmeonstrated that although the identification of Muslims did not play a role in supporting democracy during the Gezi Turkey protest in 2013, identification as a Muslim did indicate a slightly different pattern in adopted political attitudes (i.e., secular, liberal, conservative, or moderate) (Baysu & Phalet, 2017).
Politicized and religious identities are types of single identity -a concept that has become the focus of various studies on collective action. However, the concept of multiple identities can also be taken into consideration when attempting to explain collective action. For example, Klandermans (2014) applied a concept of a dual identity and discovered that members of immigrant groups who are also members of civil society (political) organizations are more likely to be involved in protests. In addition, other findings suggest that when religious and ethnic identities overlap with one another, participation in acts of violence becomes more likely (see Basedau, Pfeiffer, & Vullers, 2014;Basedau et al., 2011). We are thus faced with two debates about the role of identity in the study of collective action. First, as with politicized identity, can religious identity explain peaceful (normative) collective action? Although it is claimed that religious identity leads to radical and violent action, does the same argument stand when applied to Indonesian Muslims' support for Palestine? Second, what is the pattern of interaction between politicized and religious identities when the SIMCA model (with group efficacy and group-based anger as mediators) is used to explain peaceful collective action? The basis of this question is rooted in the notion that identity plays a dual role in collective action, which prompted us to use the concept of dual identity (politicized and religious) to address this second debate.

Identity Factors.
Many studies on collective action have focused on politicized identity (Alberici & Milesi, 2013Blackwood & Louis, 2012;. Becker and Tausch (2015), conducted a literature review of collective action studies and concluded that a politicized identification in both direct and indirect (through group efficacy and group-based anger) could strengthen the desire for peaceful action. Meanwhile, religious identity was found to be more strongly related to radical action (Basedau et al., 2014;Duffy & Toft, 2007;Hirsch-hoefler et al., 2016). For example, religious identity has been found to encourage armed conflict if the identity happens to overlap with ethnic identity (Basedau et al., 2014(Basedau et al., , 2011. In a slight variation of this finding, Phalet et al. (2010) discovered that when their religious identity is prominent, Turkish Muslims in the Netherlands appeared less eager to engage in collective action (normative or non-normative) than Moroccan Muslims in the Netherlands (except for superordinate purposes such as human rights), which is presumably due to Turkish Muslims experiencing more positive acceptance and less of a threat to their identity.
As mentioned earlier, in addition to single identities like a politicized or religious identity, the concept of dual identity has sometimes been used to explain collective action (Klandermans, van der Toorn, & van Stekelenburg, 2008;Klandermans, 2014). The dual identity theory was proposed as a response to CIIM, or the Common In-group Identity Model. CIIM suggests that to reduce bias between groups, different group members should be included through the use of superordinate identities (e.g., national identity) (Gaertner & Davidio, 2000). However, there are also findings that a common identity can threaten subordinate identities (e.g., ethnic identity) and will ultimately lead to adverse effects on inter-group relations (see Hornsey & Hogg, 2000). That is, people with dual identities are more satisfied with their situation (González & Brown, 2003), and dual identities reduce the likelihood that members of the majority will respond to injustice (Banfield & Dovidio, 2013;Saguy & Chernyak-Hai, 2012). People with dual identities are more likely to engage in collection action if they are dissatisfied with a situation (Klandermans et al., 2008). People with multiple identities (i.e., having two identities, one belonging to a benefited group and another to a disadvantaged group) who are involved in an alliance group movement are more able to understand intergroup relations and the social inequalities that the group wants to change (Curtin, Kende, & Kende, 2016). Hypothesis 1: Political identity plays a larger role than religious identity in the intention to participate in collective solidarity action.
Several other studies have attempted to explain the role of group efficacy factors (Alberici & Milesi, 2013;Shi et al., 2015;Stewart et al., 2016;Tausch & Becker, 2013;van Zomeren et al., 2011). In particular, high group efficacy was found to better predict normative collective action, while low group efficacy leads to violent action . Both group-based anger and group efficacy served as mediating factors in some research (Becker & Tausch, 2015;Miller, Cronin, Garcia, & Branscombe, 2009;Stewart et al., 2016). Hypothesis 2a: group-based anger mediates the effect of politicized identity (and religious identity) on collective solidarity action. Hypothesis 2b: group-based efficacy mediates the effect of politicized identity (and religious identity) on collective solidarity action.
Integrative Model. Some research findings and theories have suggested that collective action should be investigated using an integrative model, in which each variable interacts with another to explain collective action (see Becker & Tausch, 2015;Thomas, McGarty, & Mavor, 2016;van Stekelenburg, Klandermans, & van Dijk, 2011;van Zomeren, 2015).Van Zomeren (2015) argued that a "flat" model that involves only one variable would be less comprehensive than an integrative model in its ability to explain collective action. This argument is based on the finding that involvement in collective action is caused by various factors and each inter-factor relationship cannot, by itself, adequately explain collective action, as certain interactive patterns also play a role.

Current study.
We used the SIMCA model to explain the interaction of politicized and religious identities with other mediator variables, such as group efficacy and group-based anger. The decision to use this model was based on the fact that our study focuses on the dynamics of identity as a central factor in SIMCA . SIMCA offers an integrative approach to analyzing the motivational factors of collective action. To address the dual identity concept in our study, we slightly modified SIMCA by adding a religious identity variable (in addition to politicized identity) with group efficacy and group-based anger acting as mediators.
The type of action that will be examined in this study consists of peaceful actions in the context of solidarity with Palestine issues. Issues involving religious and political beliefs were chosen based on Louis, Chapman, Chonu, and Achia's (2017) explanation that while political and social issues often take center stage in studies of collective action, religion is a context that needs to be examined further. In this study, Palestinian issues are of special concern because Indonesia is known for providing consistent support to the Arabs and Palestinians in their fight against Israel (Sukma, 2003). In fact, the Palestinian issues often become the focus of attention among Indonesians, with various supportive actions such as donations, petition signings, speeches, protests, marches, among others, being commonplace.

Methods
Design and Sample. This study employed a survey method and the data was analyzed using a correlation technique. This particular method was selected to answer explorative questions about the dynamics of the factors that predict collective solidarity action intention. The said predictive factors, functioning as research variables, were (1) religious identity (as Muslims); (2) politicized identity (as member of a movement); (3) group efficacy; and (4) group-based anger. Meanwhile, the dependent variable was (5) collective solidarity action intention.  (Gravetter & Forzano, 2012).

Measurement.
All research variables were measured using self-reporting methods entailing individuals' responses to distributed questionnaires (Stangor, 2011). The instruments were adapted from previous studies and adjusted to the context of the current research. All items were measured using a 7-point Likert scale ranging from 1=extremely unlikely/disagree to 7=extremely likely/agree, except for negative items that were scored in reverse. The instruments were tested using Confirmatory Factor Analysis (CFA), with validity being determined by the standardized loading factor score of each item. The estimated reliability of the measures was established using internal composite reliability (CR) (Brown, 2006;Kline, 2015), which considers both errors of measurement and errors of covariance. We also calculated the average variance extracted (AVE), which is equivalent to the value of communality construct (Hair, Hult, Ringle, & Sarstedt, 2014).
The measurement of collective solidarity action intention was adapted from Tausch et al. (2011), with 8 statements related to normative or nonviolent actions (CR = 0.88). Some items were changed to match the conditions in Indonesia, particularly by an expert judgment to formulate types of action. Before a social psychologist (as expert) formulated the action, eight leaders of Islamic student organization answered openended questions aimed at confirming the types of action taken by organization members. An example of a normative action item was, "participating in a discussion meeting on the Palestinian conflict," with the response indicating the extent of agreement with the statement. Each of the instruments measuring politicized and religious identities consisted of four items, and both instruments were adapted and developed from instruments constructed by van Zomeren et al. (2010) and van Zomeren, Postmes, and Spears (2012). An example of an item measuring religious identity (CR = 0.86) was, "I feel a strong bond with fellow Muslims," while an example of a politicized identity item (CR = 0.90) was, "I see myself as a member taking part in the Islamic movement." Group efficacy was measured by three items adapted from Tausch et al. (2011) and Tausch and Becker (2013), with a reliability of CR = 0.87. A sample item measuring group efficacy was, "In my view, the power of Muslims can stop the conflict in Palestine." The measurement of group-based emotion, particularly anger, comprised two items with a reliability of CR = 0.88 with the following sample item, "Muslims are furious with the Palestinian conflict situation." Procedure. Data was collected using a paper-and-pencil method administered both directly to members of Islamic organizations as well as online through a Google Form filled out by Islamic student participants, who were spread throughout several regions of Indonesia. After all the data was collected, the researchers performed a check of participant data to identify and eliminate data that did not derive from Muslim student activists.

Statistical Analysis. The researchers performed a statistical analysis by running a Structural Equation
Modeling (SEM) on LISREL to test the research hypotheses. We tested a full model that involved indicators of each variable (measurement structure) and the structure of the proposed model. The criteria used for the model fit were as follows (Hooper, Coughlan, & Mullen, 2008;Ghozali & Fuad, 2012;Schumacker & Lomax, 2016): (1) Chi Square (χ2) close to zero, and significance > 0.05. A large chi square is attributed to a large number of participants, yet as an increase in data size corresponds to an increase in chi square (Wijanto, 2008), it is difficult to explain the fit in a study of 303 people using only a chi square value; (2) the Root Mean Square Error of Approximation (RMSEA) close to zero, or between 0.05 and 0.08. A RMSEA confidence interval (CI) of 90% was also included for all model fit analysis results. The criteria for a good model fit is a low CI that is close to 0 and no higher than 0.08 (Hooper et al., 2008); (3) Comparative Fix Index (CFI), Normed Fit Index (NFI), and Non-Normed Fit Index (NNFI) greater than 0.90; (4) Standardized Root Mean Square Residual (SMRS)< 0.08.

Results
Confirmatory Factor Analysis (CFA). During the initial stage of the study, a CFA was performed simultaneously on all variables prior to SEM analysis (see Kline, 2015). religious identity with a value of 0.597, and another indicator of collective attention with a value of 0.646. A loading factor > 0.50 suggests that the item quality still falls within an acceptable range (Ghozali & Fuad, 2012;Wijanto, 2008). All instruments had internal composite reliability (CR) scores between 0.70 and 0.90, thus demonstrating satisfactory reliability (Nunally & Bernstein, 1994;Hair et al., 2014). Moreover, the average variance extracted (AVE) values in the current research ranged from 0.551 to 0.782, while AVE > 0.50 indicated that each construct was able to account for more than half of the indicator variance (Hair et al., 2014).

Preliminary Analysis. Normality is one of the requirements for testing Structural Equation Modeling
(SEM), yet results of the multivariate normality test showed that the data was not normally distributed, as indicated by a skewness and kurtosis significance of less than 0.05 (p = 0.000). Based upon mean value calculations, all variables tended to have means larger than 5.5 on a scale of 1 to 7 (see Table 1). The univariate normality test revealed that almost all indicators were negatively skewed or skewed to the left (i.e., scores tend to be high). The mean values of all variables were also relatively high and ran in parallel with the results of the univariate normality test (see Table 1). Results of the univariate normality test showed three indicators belonging to the moderate normality category, namely indicators 1, 2, and 3 of collective action, with skewness and kurtosis values that ranged between ± 1.0 -2.3 (Beauducel & Wittmann, 2005). We additionally performed a check for outliers, which resulted in the elimination of data from 30 participants whose scores were identified as outliers. Thus, data from a total of 273 of 303 participants were analyzed.
Several solutions exist to manage data that is not normally distributed, one of which involves removing outliers, which we opted to perform. Another solution is to use the Maximum Likelihood (ML), Generalized Least Square (GLS), or Weighted Least Square (WLS) estimates (Ghozali & Fuad, 2012). We estimated the model using ML estimation via LISREL 8.80. The decision to use ML for the model estimation was due to the number of participants as well as the findings of Beauducel and Wittmann (2005), which suggested that in practice, using ML and GLS estimates with abnormal data would produce parameter estimates and standard error estimates that are not significantly different.
In SEM testing, items must be free from multicollinearity; that is, the inter-correlations among items should not be larger than 0.90 (Ghozali & Fuad, 2012). Correlation matrix tests using LISREL revealed only one correlation that indicated multicollinearity (r = 0.935), namely between indicator 1 and indicator 2 of religious identity. The two items that have multicollinearity were retained because Schumacker and Lomax (2016) recommended paying attention to the theoretical rationale of a model that contains several multicollinear indicator variables. Another consideration for maintaining the multicollinear variables was due to the reliability of the instrument used to measure social identity as a Muslim (0.86).

SEM Test.
We tested the strength of social identity as a Muslim and politicized identity of Indonesian Muslim activists using a structural model with ML estimation conducted by LISREL 8.80. We selectively implemented the suggestions for modification indices among the error variances in one measurement (variable), which was performed to meet the model fit criteria. In general, several criteria of the model fit were fulfilled, although the chi square value did not meet the fit criteria. To replace the chi square parameter in a data set that is not normal, NFI, NNFI, and CFI are used on samples < 500 (Beauducel & Wittmann, 2005). Using data from the 273 participants who remained after the elimination of outliers, the model  The coefficient of determination of the structural equation (R2) was found to be 0.447. In other words, religious identity, politicized identity, group efficacy, and group-based anger unitedly account for 44.7% of collective solidarity action intention. In the structural equation, the effect of politicized identity on collective action intention is significantly stronger than the effects of other predictors (β = 0.588; SE = 0.084; t = 7.047; p <0.01). Group efficacy was shown to have a significant relationship with collective action intention (β = 0.155; SE = 0.073; t = 2.123, p <0.05). However, religious identity (β = -0,055; SE = 0,090; t = -0,567; p = ns) and group-based anger (β = 0,077; SE = 0,066; t = 1,165; p = ns) did not significantly influence collective solidarity action intention. A corresponding path diagram is presented in Figure 1.
Results of the LISREL analysis suggest a simultaneous indirect effect of both mediators (group efficacy and group-based anger). To examine their specific effects, we tested the mediating effect of each identity factor separately using LISREL 8.80. A summary of these tests can be found in Table 2. We found that group efficacy functioned as a significant partial mediator (B= 0.168; SE= 0.048; t= 3.497; p < 0.01) on the interaction between religious identity and collective solidarity action intention. β indicates a standardized effect size.
Despite a smaller effect size, group efficacy also played a significant role as partial mediator of the effect of politicized identity on collective solidarity action intention (B= 0.080; SE= 0.033; t= 2.558; p < 0.05). On the other hand, group-based anger had no significant mediating effect on the interaction of religious identity (B= 0.063; SE= 0.033; t= 1.931; p= ns) or politicized identity (B = 0.036; SE= 0.018; t= 1.931, p=ns) with collective action intention.

Discussion
The SEM test results demonstrated that politicized identity has a significant direct effect on collective solidarity action intention, while the effect of religious identity was not significant (i.e., support for Hypothesis 1). The politicized identity measured in the current study was rooted in social identity theory, the main premise of which is that when individuals are faced with various situations, they tend to view themselves and others as part of a group more than they emphasize their own individual uniqueness (Ellemers & Haslam, 2012). Results of prior studies on collective action support our findings (see Tausch & Becker, 2013;Thomas, McGarty, & Mavor, 2016;, in that the more a person identifies themselves as part of a group movement (i.e., politicized identity), the more willing they are to be involved in collective action. Specifically, it was found that a political social identity is a better predictor of collective action than a non-political identity (Alberici & Milesi, 2016;van Zomeren, Spears, & Leach, 2008).
Self-categorization into a group was one of the prerequisites for an individual's participation in collective action (Wright, 2001;2009). In our study, individuals were more likely to identify themselves as members of a movement defending the Palestinian cause, which indicated that their religious identity did not directly motivate their intention to engage in collective action. This finding seems to illustrate that religious identity is viewed as an aspect of an individual's personal life (see Nagel & Staeheli, 2011). Among Muslims in the West, Islam is perceived as a belief (faith) and personal identity, while political involvement typically occurs through a separate identity that is more secular (Nagel & Staeheli, 2011). Evidence for this argument comes from astudy by Klandermans (2014) on the role of dual identity in protest actions. In their research, Klandermans (2014) and Klandermans et al. (2008) found that immigrants who simultaneously identify with a movement (i.e., a politicized identity) show a greater tendency to participate in collective action. Alongside our findings, this implies that religious identity is not the main reason Indonesian Muslims, especially Islamic organization activists, engage in collective action in solidarity with Palestine. When responding to a conflict involving their fellow Muslims in another part of the world, politicized identity is more likely to be the factor that drives their desire to be involved in collective action.
According to their demographic data, all of the current study's participants were affiliated (either as members or administrators) with an Islamic organization. This indicates that the resources provided by a religious organization can result in the formation of social or political movements (Burns & Kniss, 2013). According to Resource Mobilization Theory (RMT; McCarthy & Zald, 1977), resources are at the root of group efficacy. Our findings revealed that despite its small effect size, as shown by the mediation test, group efficacy was found to significantly mediate the relationship between politicized identity and collective solidarity action (i.e., support for Hypothesis 2b).
Additionally, the structural equation revealed that group efficacy had a significant direct effect on collective solidarity action intention. Group efficacy can be defined as a group's feeling about their capacity to achieve group goals (Whiteoak, Chalip, & Hort, 2004). Results of several past studies are consistent with our findings on the role of group efficacy in collective action (Alberici & Milesi, 2013;Saab, Tausch, Spears, & Cheung, 2015;Shi et al., 2015;Stewart et al., 2016;Tausch & Becker, 2013;van Zomeren et al., 2011). In particular, the stronger an individual's belief in the group, the more willing the individual is to engage in normative or peaceful collective action van Zomeren, Spears, Fischer, & Leach, 2004).
According to the specific mediating effect tests, we found that group efficacy was a significant mediator of the relationship between religious identity and collective action intention (i.e., support for Hypothesis 2b). Our findings contribute to the explanation that religious identity predicts the desire to engage in collective action (Baysu & Phalet, 2017). We suspect that the source of group efficacy pertaining to the Palestinian issue is largely attributable to religious identity, which helps explain the reason Muslim religious identity -through the mediating effect of group efficacy -has a stronger effect on collective action intention. The conflict in Palestine is narrated in the Muslim scriptures, which purports the belief that one day at the end of the era, Muslims will once again be able to win the conflict. In addition, the Palestinian territory is closely related to the historical and cultural contexts of religion. God is said to have chosen Palestine as the land of divine missions for both Islam and Christianity in particular (Litvak, 1998).
Group-based anger was not found to significantly affect collective solidarity action intention directly in the main model, yet such a finding seems to contradict prior discoveries (see Shepherd, Spears, & Manstead, 2013;Shi et al., 2015;Stewart et al., 2016;van Zomeren et al., 2011). This particular emotional factor was also not found to have a significant mediating role in the two measured types of identity (i.e., Hypothesis 2a not supported), again conflicting with the findings of past studies that social identity affects the desire to act through group-based anger (Tausch & Becker, 2013;Thomas, Mcgarty, & Louis, 2014). A longitudinal study was conducted by Srour, Mana, and Sagy (2016) to investigate the perceptions of Arab and Jewish teens in Israel about the collective narratives of the prolonged Israeli-Palestinian conflict. The researchers found that Arab Israelis did not feel much anger about the collective narrative of their ingroup compared to Jewish Israelis. We suspect that Indonesian Muslims similarly feel less anger about the collective narrative of the Israeli-Palestinian conflict, the reason being that the emotion measured in the current study is related to group-based anger, where the group has no direct experience in the conflict. Group-based emotions are believed to emerge when individuals consider their own emotional experiences as a response to group-related events (Goldenberg, Saguy, & Halperin, 2014).
In studies of collective action, group-based anger was found to stem from responses to injustice van Zomeren, 2015;van Zomeren et al., 2004). Brown, Abernethy, Gorsuch, and Dueck (2010) explained that Muslim anger can arise when it is associated with the perception of political events as a form of holy violence and injustice. We assume that the inability of group-based anger to explain collective action in the current research is due to the characteristics of our participants (i.e., from Indonesia), who do not directly experience the injustices suffered by Palestinians. This contrasts with the studies of Tausch et al. (2011) andvan Zomeren et al. (2004) where participants' were students who had directly experienced Germany's policy on educational costs, which revealed that perceptions of injustice led to group-based anger in peaceful collective actions.
Finally, this study has some limitations that should be noted. One limitation of the present research is its design, which is only able to explain the interactive effects between variables due to its correlational nature. To explain causality, an experimental design is needed. We suggest that future researchers test the central role of social identity using an experimental design. We also recommend that further studies consider the concept of dual identity, which may include both national identity (e.g., Indonesian nationality) and ethnic identity (e.g., membership in a particular ethnic group), especially in ethnically diverse countries like Indonesia.

Conclusion
This study corroborates the finding that politicized identity is a stronger direct predictor of peaceful collective action than religious identity. The function of politicized identity is especially robust among members of group movements. Moreover, the role of religious identity is only prominent when mediated by group efficacy. This finding adds to the existing literature by suggesting that religious identity functions more effectively if there is a strong belief in the success of the movement, and that this type of group efficacy is very likely to be associated with religious beliefs. Our SIMCA test of collective solidarity action discovered that the model does not explain the dynamics of the consistent motivation behind collective action. Additionally, group-based anger was not a causal factor involved in individuals' collective action, either directly or through its mediating effect. The rationale for this finding is that the source of anger, which in this case takes the form of experiences of injustice, was not experienced directly by participants in solidarity actions in the current study.