Cultures of conflict: Protests, violent repression, and community values.

What are the cultural origins of societal conflicts that revolve around democratization, women's rights, and modern libertarian values? We propose that deep-seated differences in community-based collective values (at the micro-level) may be related to why people support anti-government protest and why they support repression of such protests (at the macro-level). The hypothesis was examined among residents of Turkey (N = 500). Cultural values, measured at the individual level and community level with the community collectivism scale, correlated with political orientation and emotions, as well as with subsequent support for anti-governmental protest or its repression. The main conclusions are that both support for protest and support for repression are related to the cultural values people hold and their subsequent political orientations and emotions. Micro-level cultural values in local communities may thus play a role in explaining macro-level socio-political divides.

Background

Since around 2010, many countries in the world have experienced a sharp increase in major nationwide protests (Banks & Wilson, 2016). The Arab Spring, mass protests in India against sexual violence, the women's march in the United States triggered by Donald Trump's policies and statements, and the Gezi protests in Turkey are examples of a subset of these protests that primarily seem to revolve around democratization, women's rights, and modern libertarian values. The present study's starting point is the observation that protests often coincide with deep‐seated societal divides (Kriesi, 2010) which may revolve around cultural values. We take this one step further by proposing that the cultural values associated with social and political conflicts stem from local communities and that they are potent influences because they perform an important function in within‐community group dynamics. The present study therefore focuses on values of individualism and collectivism in relation to one's proximate community (so, at the micro‐level) and examines to what extent these values are related to societal protest against a conservative government and to the violent repression of those progressive protests (at the macro‐level). The question is whether micro‐level cultural values are related to macro‐level political support.

We study this by focusing not just on support for protests, but also on support for the (violent) repression of these protests. Traditionally, studies of protest and movements focus on disadvantaged groups in intergroup settings (Kelly & Breinlinger, 1996; Van Zomeren, Postmes, & Spears, 2008). Only few studies (Cakal, Hewstone, Schwar, & Heath, 2011; Milesi & Alberici, 2018; Osborne, Jost, Becker, Badaan, & Sibley, 2019) have focused on what might motivate support for activism on both sides of a political protest – both in support of progressive or libertarian principles (in the current study: protests that challenge the status quo) and in support of authoritarian or conservative principles (in the present study: the repression of those protests). In the present study, we focus on what we conceptualize are proximate determinants of such support (political orientations, emotions) as well as more distal predictors (community‐based values of individualism and collectivism). We thus focus on cultural divisions within society that, we assume, may be related to proximal predictors. We examine these processes in Turkey where we expect such community‐based values to be associated with large‐scale clashes. The specific protest, the so‐called Gezi protest, as well as the context of Turkey, we believe makes an interesting case study of a within‐society conflict with cultural origins.

Distal determinants of conflict: clashing cultural values

In studies of political conflict and collective action, culture has often been identified as a relevant factor. For example, cultural characteristics can sometimes enhance the likelihood of intercultural conflicts, by encouraging intercultural misunderstandings (Triandis, 2000; Williams, 2004). Moreover, speaking to the relevance of studying ‘culture’ in relation to collective action (Van Zomeren & Louis, 2017), cross‐national differences have been shown to affect the relation between determinants of collective action and collective action itself (Stewart et al., 2016), as well as beliefs about social change (Bain, Kroonenberg, & Kashima, 2015). These perspectives implicitly build on the assumption that cultures are broadly shared within a society and not themselves contested. However, in the current study we take the perspective that cultural values may also become a source of disagreement within a society: The question of which culture is or should be dominant may itself be the subject of political contestation (see also Jasper, 2017). Indeed, various societal tensions currently witnessed across the globe (e.g., Brazil, Iran, Turkey, perhaps also the United States) can be explained as intra‐societal conflicts over basic cultural values: women's rights, democracy, and self‐determination. This is indirectly confirmed by research that shows that liberal voting tends to coincide with cultural values such as universalism, benevolence, and self‐direction, whereas conservative voting is associated with more conservative cultural values of security, power, achievement, tradition, and conformity (Caprara, Schwartz, Vecchione, & Barbaranelli, 2008). Therefore, there is reason to believe that differences in cultural values may have political consequences.

Assessing within‐country cultural variation

It is widely accepted that culture can be considered the template for socialization for (new) members of a society (Schwartz & Ros, 1995) and as such serves to instil a shared understanding of how a society works and should work. There is, however, debate about at what level such common understanding is shared.

Many existing measures and conceptualizations of cultural values such as individualism versus collectivism (Minkov & Hofstede, 2012), tightness versus looseness (Gelfand et al., 2011), or basic human values (Schwartz, 1999) are measures of individually held values. With respect to examining the differences between cultural groups, these measures have one shortcoming: One can only make inferences about the group by aggregating across individual‐level measures (cf. McCrae et al., 2013), which conceptually seems at odds with the commonly accepted definition of culture as shared meaning system (Fischer, 2012). When examining within‐culture differences (which is particularly important in heterogeneous societies), this problem is compounded by not knowing what particular community or subcultural group an individual's responses should be attributed to.

A recent approach attempts to solve this problem by focusing on the intersubjectivity of values (Chiu, Gelfand, Yamagishi, Shteynberg, & Wan, 2010), suggesting that people attach special importance to cultural values that they perceive to be widely shared within their cultural group. In most implementations thus far, researchers have tended to measure what individuals perceive to be the cultural values held by the majority of ‘their country’ or ‘their culture’. This approach is well suited to examining between‐national or between‐cultural differences in intersubjective values; however, it is less well suited to examining within‐nation differences, such as the clashing cultural values between subcultural groups or communities that the present study is focused on.

Community collectivism

Extending the intersubjective methodology, (Akkuş, Postmes, & Stroebe, 2017) recently developed the concept of community collectivism. The core idea is that some cultural values serve the purpose of regulating social interactions within communities such as one's (extended) family.1 These tend to be tight‐knit communities in the sense that members are connected by strong ties such as family bonds and they tend to be relatively close proximity, characterized by frequent contact and longevity (e.g., networks of close kin, cf. Bian, 1997). Within such tight‐knit communities, we propose there are values that serve the purpose of keeping the group's structure intact and managing intra‐group interactions.

Community is defined in the community collectivism scale as: ‘your core family, your extended family and other families (in your environment) that matter to you’.

These values help the community to achieve clarity about expectations, obligations, group interactions, and transactions. These core values perform four functions that keep the group tight‐knit: (1) provide loyalty and support, so bonds remain strong; (2) divide labour and responsibilities, so group members know their roles and position; (3) enforce norms and expectations, so group members know what is honourable and shameful; and (4) encourage agency and autonomy, so group members know when to act and think for themselves (see also Postmes, Akkus, & Stroebe, 2015). Accordingly, the central collectivist values should be concerned with loyalty and support, hierarchy and structure, and honour and norm enforcement. Individualist values are concerned with agency and autonomy. Intersubjective agreement about these community‐based values benefits group efficiency and keeps social relations intact: They inform every member of the community about who they are and about how to act in relation to others.

This concept of community collectivism has implications for how values of collectivism and individualism should be measured. These values are held by individuals, but they originate from the community and are normative. In line with this concept of community collectivism, a 30‐item community collectivism scale was recently developed and validated (CCS, Akkuş et al., 2017). The CCS measures collectivist and individualist values held at the personal level and perceived to exist within one's self‐relevant community (i.e., one's family and the immediate social network around it). In order to measure community values, for example, the scale assesses perceptions of common social practices within one's community (e.g., ‘In my community, members of the family feel responsible for preserving and protecting another family member's honor’). The assumption is that individual values are grounded in the values of one's community. Akkuş et al. (2017) validated the scale and showed that it distinguishes between personal collectivism and community collectivism, as well as between personal agency and community agency.

Prior research has shown that CCS explains behavioural intentions and attitudes within the community (e.g., whether one should care for the elders in one's family, or how people of different genders should behave, Akkuş et al., 2017). Relevant for the present study is that Akkuş et al. (2017) also showed that CCS predicted voting behaviour: Higher scores on collectivism were strongly related to voting for a conservative party.2 In the current research, we study whether values of community collectivism can also be related to support for collective action against a conservative government and in favour of it. We thus propose that values originating from local tight‐knit communities may play a role in the broader political context of society. In particular, the varying degrees to which communities promote collectivist values of honour and hierarchy and the subsequent subcultural differences can become intimately connected with national‐level politics. As Jasper (2017, p. 298) puts it: ‘Small groups are precisely the kind of interactive context that creates understandings, in which decisions are made and actions initiated, where emotions are generated and displayed, with impacts on others […] They are the kind of local setting where politics unfolds’. While we believe that this connection between community values and political movements is relevant in multiple societies today, we do not know of any research that has addressed this specific point. What is interesting and novel about this approach, we believe, is that it implies that macro‐level political conflict may ultimately stem from the desire to preserve and protect local communities.

It is important to note that this finding was a follow‐up study in which the sample used in the present study was resampled.

Support for activism and repression in conflict situations: proximal determinants

What are relevant proximate predictors of support for anti‐government protest and for government oppression? Given the constraints of conducting a short survey, we decided to focus on two key predictors that should, according to prior research, account for a considerable amount of variance: political orientation and negative emotional reactions (including moral condemnation of the outgroup).

In general, people's political orientation is the strongest predictor of whether they support a government policy or not (Lau & Heldman, 2009; Sears, Lau, Tyler, & Allen, 1980). In situations of a ‘hot’ conflict between government and its opponents, it makes sense that this effect of political orientation is also strong, if not stronger: Intergroup conflict tends to accentuate ingroup favouritism (Tajfel & Turner, 1979) and so in a conflict situation those who take the government's side should, if anything, support the governing party's actions to suppress a protest even more (Brewer, 2001). This reasoning can be extended to predictors of support for protests against the government. A large literature shows that commitment to activism is predicted by identification with the activist cause (Van Zomeren et al., 2008). By extension, defining oneself as a government opponent should coincide with support for government opposition. In the current study, since we are dealing with protest against a conservative government, we reason that this protest would be supported by those who categorize themselves as liberals (or progressives), while government repression would be endorsed by those who self‐categorize as conservatives.

In addition, we were interested in the role of moral emotions in support of protests or its repression. One reason for this is research showing that, particularly in conflict situations and particularly among those who are ideologically involved, collective emotions may run high and may play a role in motivating (support for) action (Mackie, Devos, & Smith, 2000; Sternberg, 2005; Tausch et al., 2011; Van Zomeren, 2015). Research on collective action and mobilization has often focused on injustice‐based anger – a prime motivator for disadvantaged groups. But for high status groups who support suppression feelings of entitlement and superiority may also play a role (cf. Postmes & Smith, 2009). Accordingly, we decided not to focus solely on specific emotions involved in politicized conflicts such as anger (Tausch et al., 2011) or hatred (Bar‐Tal, Halperin, & de Rivera, 2007), but on a range of emotions, including the overarching moral condemnation of the outgroup that is common to both feelings of injustice‐based anger (cf. Tausch et al., 2011; Van Zomeren, Postmes, & Spears, 2012) as well as superiority‐based emotions (Postmes & Smith, 2009).

The present research

Research context

The current study took place against the background of the Turkish protests of 2013 and its backlash. Amid increasing tensions between the conservative government and people with a more secularist outlook, a small and peaceful protest took place in Gezi Park, Istanbul. This escalated into unprecedented nationwide protests, initially against police brutality but later against the government's socially conservative politics. This development splits the country into supporters and opponents of the government (the conservative AKP of then Prime Minister Erdogan). Opponents of the government accused it of authoritarian leadership with disregard of libertarian values and civil rights. This political conflict echoes divisions within Turkey that historically stem from the country's founding secularist principles, but increasingly transcend socio‐economic class and ethno‐cultural differences (Gumuscu, 2010). The underlying cultural values are visible, to some extent, in neighbourhoods and communities which have very distinct lifestyles (Turam, 2013). In public life, they are signalled among others by women's involvement in society and style of dress (Arat, 2010; Vojdik, 2010).

In this context, the present research examines what variables are associated with support for anti‐government protests and support for repression of those protests. It identifies proximal variables (emotions towards the protests and towards protest repression and political orientation) and more distal variables (cultural values that are community‐based, see Figure 1). We tested this model in the context of an explosive political conflict between a conservative government and a sizable progressive opposition, in a situation where entire communities appear to be taking sides (Göle, 2013). We compared it with a closely related alternative model, in which these emotions were proximate predictors and in which cultural factors are both assumed to be distal predictors that are uncorrelated. We expected the fit of the alternative model to be inferior to the fit of the preferred model.

Figure 1

Two models of support for protests and support for repression of protests: theoretical model (top) and alternative model (bottom).

Choice of variables

We focus on political orientation because prior research showed that although the protests were initially quite diverse (Baysu & Phalet, 2017; Bilgic & Kafkaslı, 2013; Konda, 2014), they quickly developed into a conflict between liberals and conservatives (Öncü & Koçan, 2014; Ozkirimli, 2014). The choice of emotions with regard to the protest and to protest repression (i.e., hate, anger, and moral condemnation) was partially based on prior research but also informed by the way the press and social media legitimized the widespread (sometimes excessive) use of violence: Protesters were referred to as traitors and heathens (e.g., Corke, Finkel, Kramer, Robbins, & Schenkkan, 2014; Taştan, 2013, also cf. McLeod & Detenber, 1999; McLeod & Hertog, 1992).

We believe that this political conflict stems from opposing values with regard to the freedom to make personal choices in life versus traditional community‐based values. Rather than operationalizing this with personal‐ or national‐level values (Caprara et al., 2008; Vecchione et al., 2015), we operationalized this with the new measure of community collectivism which we believe meshes well with the grounding of these conflicts in local tight‐knit communities.

Predictions

Regarding proximate variables, we expected support for protest to be predicted by political progressiveness and negative emotions about government handling of the protests. Conversely, we expected support for the violent repression of protests to be predicted by politically conservative views and by negative emotions towards the protests. Regarding distal predictors, we expected conservative political orientation to be predicted by more community collectivism, at the level of individual‐ and community‐based values, and progressive political orientation to be positively correlated with agency. More specifically, we expected collectivism to be positively correlated with a more conservative political orientation and therefore also with repression, as a consequence of prioritizing social hierarchy. For agency, we expected the opposite, because of its expression of personal autonomy.

Method

Participants and procedure

Previous research (Van Zomeren et al., 2008; Vecchione et al., 2015) suggests a small‐to‐medium relationship among various predictors of collective action, that is, political orientation, emotional reaction, and cultural values. According to G*power, an a priori power of .90 can be achieved with 377 participants if a bivariate correlation r = .15 (or larger). We decided to recruit a slightly larger sample of 500 participants (50% women, M age = 34.47) because of concerns we might not attract enough conservatives. Respondents completed an online questionnaire programmed in Qualtrics. All lived in Turkey and were recruited from the Qualtrics panel (and compensated by Qualtrics). Six were identified as multivariate outliers (Mahalanobis distance, p < .001) and were replaced by six additional cases by Qualtrics, such that said numbers and proportions remained intact, that is 500 participants and 50% women. There was an oversampling of city residents and highly educated respondents. 52% of participants lived in Istanbul, Ankara, or Izmir although only 30% of the population does. Almost 70% of participants were highly educated or students, which is above the country average of 10.7% (TÜİK, 2015). Importantly, a sizeable number of conservatives participated: 38% indicated being right wing, 45% left wing, and 17% neither left nor right.

Measures

We administered a questionnaire (in Turkish) that measured both CCS and attitudes and emotions towards the Gezi protests, as well as general (societal) attitudes. In this section, we will describe the measures relevant for the present study.3

The questionnaire was part of a larger study and also included a number of exploratory questions not relevant for the present study. These were questions about: the communal level of the items (e.g., ‘My community fully supports the Gezi protests’), two questions to determine religiosity (‘What role does religion play in your life?’) and traditionalism (‘How important is it for you to lead a life in accordance with traditional norms and customs?’) on a 0–10 slider, from ‘Not important at all’ to ‘The most important thing in life’. Both items were also repeated for their community and the government; a question measuring attitude towards individual autonomy (‘Do you believe that everyone should be able to make their own choices in life without interference of others?’); four items concerning attitudes towards the 17–25 December corruption case (e.g., ‘I believe that the corruption charges were a plot against the government’), on a 5‐point scale from ‘fully disagree’ to ‘fully agree’, and repeated for community and government attitudes; the Subjective Life Satisfaction Scale (Diener, Emmons, Larsen, & Griffin, 1985) and Psychological Well‐being Scale (Ryff, 1989) to explore whether attitudes were related to general well‐being.

Community collectivism scale

The 30‐item community collectivism scale (CCS; Akkuş et al., 2017) is an intersubjective scale of cultural values. CCS measures both collectivism (the mean of the loyalty, hierarchy, and honour subdimensions) and agency. All items are assessed at the level of the person (to what extent do the following sentences apply to you personally) and the community (to what extent do the following sentences apply to your core family, your extended family, and other families (in your environment) that matter to you). This means that a personal‐level collectivism item such as ‘I would support my family members, even if I wouldn't want to’ is twinned with a community‐level item ‘In my community, people are expected to support their family members, even if they do not want to’. Similarly, the personal‐level agency item ‘I am responsible for the important choices in my life’ is twinned with ‘In my community, you are responsible for the important choices in your life’.

Community collectivism scale was validated in a Dutch version and with inhabitants of the Netherlands (Akkuş et al., 2017) and has been shown to discriminate between (sub)cultural groups. In the current study, CCS was translated to Turkish and maintained a high reliability: Cronbach's alpha for the honour subdimension was .88, for loyalty α = .71, for hierarchy α = .83, and for agency α = .75. We also replicated the factorial structure4 by means of CFA (using the Lavaan package for R). We therefore conclude that CCS's methodological characteristics are stable in this translation and with this sample from a different country.

CFA (using the Lavaan package for R) showed an excellent fit for the one‐factor model with CFI = .996, RMSEA = .040, and SRMR = .015 (with Satorra–Bentler correction).

Other measures

Four items assessed political orientation. Respondents were asked to indicate their position, on a 10‐point bipolar slider, on the dimensions liberal versus conservative, progressive versus traditional, left versus right, and egalitarian versus authoritarian. Higher values indicated a stronger endorsement of a conservative, traditional, right‐wing, and authoritarian political orientation. This operationalization of political orientation was found to be reliable (α = .79), and all four items converged on a single factor.5

The community‐level four‐factor model showed a good fit (with Satorra–Bentler correction and corrected for three cross‐loadings), with CFI = .975, RMSEA = .033, and SRMR = .041. For the individual‐level model, the fit (with two cross‐loadings) was also acceptable: CFI = .953, RMSEA = .052, and SRMR = .044.

Respondents were asked to answer one item determining opposition or support for the Gezi protests on a 0–10 slider, from ‘I fully oppose the protests’ to ‘I fully support the protests’. They were subsequently presented a question measuring support for violent government repression (‘Do you believe that authorities are entitled to use force and violence against protests they deem unjust?’), on a 5‐point scale, from ‘fully disagree’ to ‘fully agree’.

The respondents were then asked to indicate to what degree they felt six emotions (anger, grief, anxiety, fear, moral condemnation, and hatred), by responding to the question ‘what you felt (emotionally) about the protests and responses to it’ followed by statements, such as ‘I felt angry regarding to the government's reaction to the protests’ (1 = fully disagree, 5 = fully agree). Finally, we asked for age, gender, education, birthplace, and residence.

Analytic strategy

The aim of our study was to examine predictors of support for the Gezi protests and predictors of support for repression of protests (in general) in one integrated model. We predicted that both tendencies would be strongly related to proximate predictors (participants’ political orientation and negative intergroup emotions). These proximate predictors would, in turn, be predicted by CCS. To test these hypotheses, we performed a path analysis by means of structural equation modelling (SEM).

Results

Preliminary analyses

As preliminary analysis of the relations among the variables in the model,6 we inspected the correlations between all variables (Table 1). As expected, conservative political orientation correlates negatively with support for the Gezi protests and positively with the endorsement of violent repression. Unsurprisingly, the emotional reactions towards the Gezi protests and against the violent repression by the government covary as expected with support for either protest or repression. However, also clear is that the emotions of anger, moral condemnation, and hate correlate most strongly and consistently with both outcome variables (consistent with the literature on intergroup emotions, Smith & Mackie, 2015). In the Supporting Information, we further analyse the separate emotions and among others show that fear, sadness, and anxiety are more relevant with respect to support for government repression than for protests.

Table 1

Correlations between support for protests and for repression and various predictors: political orientation, emotional reactions, and CCS

	Support for Gezi protests	Endorsement of violent repression of protests
Political orientation	−.586**	.477**
Anger regarding protests	−.772**	.549**
Anger regarding government reaction to protests	.661**	−.527**
Moral condemnation regarding protests	−.738**	.565**
Moral condemnation regarding government reaction to protests	.620**	−.454**
Hate regarding protests	−.539**	.410**
Hate regarding government reaction to protests	.710**	−.528**
Sadness regarding protests	−.465**	.286**
Sadness regarding government reaction to protests	.599**	−.548**
Anxiety regarding protests	−.240**	.092*
Anxiety regarding government reaction to protests	.537**	−.452**
Fear regarding protests	−.265**	.111*
Fear regarding government reaction to protests	.461**	−.380**
Community collectivism score	−.205**	.240**
Personal collectivism score	−.303**	.287**
Community agency	.120**	.003
Personal agency	.135**	−.107*

N: 500; *p < .05; **p < .01; political orientation: Higher values indicate conservatism.

We controlled for gender, education, and place of residence and found some correlations and effects in regression analyses. However, these effects were all mediated by political orientation, emotional reaction, and CCS and were therefore not included in the integrated model. Traditionalism and religiosity were not included for a similar reason: These variables (which we combined into one) had such a high degree of covariance with political orientation that they did not contribute significantly to the model. In view of this and the fact that this protest turned into a political one, we chose to focus on political orientation and drop traditionalism and religiosity from the integrated model.

We then examined the question of how these three emotions should be included in the model. Confirmatory factor analysis assessed a two‐factor model predicting that emotions towards the protests and emotions about their repression loaded on two separate factors. In this model, the residual variances of the two hate items and of hate and anger towards the government were allowed to be correlated because this significantly improved fit. Model fit was excellent, χ2(6) = 9.80, p = .13, CFI = 1.00, RMSEA = .036, and significantly better than the single‐factor equivalent, , p < .001. In the final model, it proved not possible to include these emotions as latent variables, because the covariance between the negative emotions towards the protest and support for protests was very high indeed.7 To solve this problem, we ran the final model with the extracted factor scores for the emotions, instead of with the latent variables themselves.

The covariance table is included in the Supporting Information.

A final preliminary analysis considered the relation between CCS and action types (repression vs. protest). In line with expectations, support for the Gezi protests was negatively correlated with community collectivism and personal collectivism, and positively with community agency and personal agency. For the support for violent repression of protests, we see the same pattern of correlations but in the opposite direction. These results provide a first confirmation of our predictions that community‐based cultural values, as operationalized by CCS, may be closely related to political affiliation and societal stances.

Path analysis: linking CCS, political orientation, and emotions to support for protests and repression

On the basis of the preliminary analysis, we tested an integrated model in which both ‘Support for protests’ and ‘Support for violent oppression’ were outcome variables. The direct predictors were the factor scores of the two latent variables ‘Emotions towards protests’ and ‘Emotions towards the government response.’ Political orientation was included in the model as a predictor for support as well as for both emotions. The model further specifies that emotions and political orientation are predicted by personal collectivism and personal agency. In the last layer, personal collectivism and personal agency are predicted by community collectivism and community agency.

As becomes clear from the fitted model (Figure 2), the results are in line with our expectations: Support for the Gezi protests is positively related to negative emotions towards government repression, and it is negatively related (and more strongly) to negative emotions towards the protests. It is also negatively related, albeit to a lesser degree, to political orientation (i.e., higher degrees of conservatism).

Figure 2

Structural equation modelling path analysis of support for protests and support for repression of protests; political orientation: Higher values indicate conservatism; N = 500, CFI = .992, RMSEA = .047, SRMR = .031 with χ2(15) = 31.53 ; *p < .05; **p < .01. [Colour figure can be viewed at http://wileyonlinelibrary.com]

The model results also confirm that the negative emotions, either towards the Gezi protests or towards the repression of protests, are strongly related to political orientation. Higher scores on this predictor indicate higher degrees of conservatism, which are associated with less negative emotions towards government repression and more negative emotions towards the protests themselves. More interestingly, the emotions are also predicted by personal collectivism and personal agency, both directly and indirectly, via political orientation. This confirms our predictions: Political orientation is positively (and strongly) predicted by personal collectivism and negatively by personal agency.

The fit of the theoretical model was good: CFI = .992, RMSEA = .047, SRMR = .031 with χ2(15) = 31.53 (p = .007). The fit of the alternative model in which political orientation and cultural variables were specified as uncorrelated distal predictors was inferior, as becomes evident from its goodness‐of‐fit measures: CFI = .971, RMSEA = .081, SRMR = .092 with χ2(17) = 93.62, p < .001.

In sum, it is clear that there are strong relations between whether people support protests or their repression, emotions towards these events, and their political orientation. And in turn there are clear and strong relations between that political orientation and their (basic) cultural orientation, as expressed with the community collectivism scale. We can therefore conclude that the basic cultural values measured with CCS help explain which side people might pick in socio‐political conflicts between progressive protesters and a conservative government.

Discussion

What cultural processes underlie political divides that cause tensions within countries around democratization, women's rights, and modern libertarian values? Integrating insights from cross‐cultural research, political psychology, and collective action research, we proposed and found support for a conceptual model that is based on the idea that people's support for anti‐government protests and support for the government's repressive response to these protests are ultimately grounded in cultural values whose function is to preserve the integrity of local communities. As predicted, the results suggest that the relationship between community‐based values of collectivism and agency is mediated by proximal predictors: political orientation and emotions with regard to the conflict. Two things are novel about these findings. One is that it suggests macro‐level political conflict may stem from efforts to maintain or preserve local tight‐knit communities. The second is that the same factors that predict support for protest also predict support for governmental repression.

We took the Gezi protests in Turkey as an example of a mass protest that divided a country into those supporting the protests and those supporting its oppression by the government (Odağ, Uluğ, & Solak, 2016). Whereas most collective action research has looked at determinants of (peaceful) societal protest (for overviews see Van Zomeren et al., 2008; Wright, 2010), we reasoned that the same factors (emotions, political orientation) might also be related to support for (violent) repression of protests. This is in line with previous work showing that predictors of such societal protests are similar, even if the nature of the associations may differ, both for parties who would want to challenge versus defend the status quo (e.g., liberals vs. conservatives, Black vs. White students; Ayanian & Tausch, 2016; Cakal et al., 2011; Milesi & Alberici, 2018; Osborne et al., 2019). Indeed, we find that emotions directed respectively at government reactions to protest or at the anti‐government protest and political orientation, liberal versus conservative, were respectively related to support for societal protest and for its repression. This shows that, in the context of a political divide between a conservative government and non‐conservative protesters, the same factors that motivate action to achieve social change, such as people's political orientation and the emotions they experience, may also explain support for actions to maintain the status quo.

We also sought to understand what might underlie such emotions and political affiliations that have the potential to divide a nation. In the case of the tensions within Turkish society (and potentially others as well), we reasoned that one important factor may be that communities in Turkey lead their lives in very different ways and embrace very different basic cultural values. To capture this, the current research measured community‐based cultural orientations (CCS). Results show that as predicted, CCS is associated with support for protest as well as repression via political orientation and negative emotions: Endorsement of collectivist values is associated with conservatism and with negative emotions towards protests, which in turn is associated with greater support for repression. Conversely, endorsement of values of personal agency is related to more liberal political affiliation and more negative emotions about government actions. And this in turn is associated with greater support for protest. As predicted by the cultural collectivism perspective, the research also showed that personal values (of individualism as well as collectivism) were strongly related to perceived community values. Putting things together, the statistical model suggests that proximate predictors of support and opposition of government are related to personal values of collectivism and individualism, respectively, which in turn are related to community values of collectivism and individualism. In sum, the findings confirm that cultural values that are specific to local tight‐knit communities are a major source of personal values and thereby may also inform more macro‐level political views.

Theoretical implications

The present work illustrates the value of integrating (cross‐)cultural, political, and collective action perspectives. At the theoretical level, it indicates that predictors of action may be similar whether we consider the pursuit of social change or the preservation of the status quo (cf. Osborne et al., 2019). The fact that these predictors mirror each other underscores the dynamic nature of political divides: In order to understand the emergence and consolidation of such divides, research needs to take into account the attitudes of both parties. Being able to predict, both proximally, via political orientation and emotions, and distally, via cultural values, to what extent a more powerful group is willing to repress social protest is essential to understanding the mechanisms underlying political divides.

More generally, the present research shows that political divides that can instigate societal conflicts (e.g., Gezi Park protests; Arab Spring) are thus not only based on power imbalances and social class per se (cf. Kriesi, 2010; Saeri, Iyer, & Louis, 2015). To illustrate this with current tensions in the United States: The political divide between supporters of president Trump and (more liberal) opponents is not only based on social class and a power imbalance, but also on a fundamental disagreement on values, that are essentially cultural. Similarly, more right‐wing movements striving for change (e.g., PEGIDA in Germany) seem to be based on cultural rather than political values, on the clash between modern/liberal and traditionalist/conservative values. Such cultural differences can be described by means of personal values, such as the Schwartz values, but in many cases, they may also be anchored in the particular values that people share within their communities, as described by the CCS. As such, our proposed model provides the means to understanding political and societal divides from a collective action and cultural value perspective.

Limitations and future directions

Our model provides the means to unpack the cultural components of social and political conflict and to look into the contents of the values concerned. The model provides a template to study the origins of societal and political divides in other societies (e.g., the United States, Germany, India). However, replication across other political conflicts and different cultural settings would be of added value. Having said this, we would argue that the concepts we put forth, cultural values and more proximate determinants (emotions, political orientation), are likely to predict support for protest versus repression across cultural contexts. Yet, the relative strengths of these concepts and their subsequent associations may vary (see also Stewart et al., 2016): For example, in countries that are more liberal, personal agency and political orientation may be more strongly related than in the present study.

Also, we note that our analyses are all essentially correlational in nature. While it makes sense to consider cultural values as underlying political orientation, emotions, and political action, more dynamic models are also plausible: For example, engaging in or witnessing societal protests over time may affect communities and the cultural values these communities come to hold. In other words, cultural values and political orientations may also be dynamic and subject to change. Future research could speak to this question by taking a more longitudinal approach in which the dynamics of cultural values and political action are studied over time. At the same time, in line with the idea of culture as a shared meaning system (Fischer, 2012), it would be of interest to use multi‐level models as a means of modelling the impact of community level on individual‐level values and vice versa over time (see also Christ, Sibley, & Wagner, 2018).

Conclusion

The present work reveals that in order to understand the positions people take in political divides, such as supporting societal protest or, conversely, supporting the repression of such protest, it is important to consider not just individually held political attitudes but also the local community structure as a possible origin of intra‐societal conflict. Both support for protest and repression can be related to cultural values that originate in people's local tight‐knit community, which may inform their political orientations and emotional responses to political events. As such, the present work contributes to our understanding of why deep‐seated political views are held, maintained, and if necessary protected.

Data S1. Differences with regard to emotional reactions.

Table S1. Correlation Matrix of Dependent and Independent Variables.

Table S2. Covariances of emotional reactions to protests and repression.

Table S3. Parameter estimates of the SEM path analysis (without estimated variances).

Table S4. Descriptives of measured variables.

Click here for additional data file.