Beat, Ignore, Force to Conform: Development and Initial Validation of a Multidimensional Scale of Acceptance of Collective Violence

Abstract

Literature on collective violence usually treats an act of aggression as a unidimensional phenomenon-occurring or not. The social psychological perspective on intergroup relations shows, however, that different aspects of an intergroup situation (social context, differences in status, past relations) lead to different behaviors. This article describes the development and initial validation of the multifaceted concept of intergroup collective violence. In a series of three studies (N = 1,420, N = 1,000, N = 1,019) using mixed methodology, we constructed a scale for measuring acceptance of intergroup collective violence. Results show that it is a multidimensional phenomenon, dependent on 1) the ethnicity of the victims; 2) the perception of threat posed by them; and 3) the ideology. The results can have a substantial impact on the discipline, providing theoretical explanation of the differences in outbursts of violence in similar situations, such as pogroms.

Headnote

Literature on collective violence usually treats an act of aggression as a unidimensional phenomenon-occurring or not. The social psychological perspective on intergroup relations shows, however, that different aspects of an intergroup situation (social context, differences in status, past relations) lead to different behaviors. This article describes the development and initial validation of the multifaceted concept of intergroup collective violence. In a series of three studies (N = 1,420, N = 1,000, N = 1,019) using mixed methodology, we constructed a scale for measuring acceptance of intergroup collective violence. Results show that it is a multidimensional phenomenon, dependent on 1) the ethnicity of the victims; 2) the perception of threat posed by them; and 3) the ideology. The results can have a substantial impact on the discipline, providing theoretical explanation of the differences in outbursts of violence in similar situations, such as pogroms.

Keywords: collective violence, intergroup conflict, intergroup threat, collective narcissism, ideology

On 27 August 2018, the Kosher restaurant Shalom in Chemnitz, Germany, was raided by a group of neoNazis.· The group shouted, "Get out of Germany, Jewish pigs!" and threw stones and bottles at the restaurant. In this attack, the owner of the restaurant was wounded and the restaurant suffered substantial property damage (Aderet 2018).

In the summer of 2018, migrants and refugees traveling through southern Europe testified that while crossing Croatia they were subjected to systematic violence from the Croatian police. About 70 percent of the refugees reported some form of violence, frequently that they had been beaten and robbed, and their mobile phones had been smashed (Walker 2018).

At the end of March 2019, a series of anti-Roma incidents occurred in the suburbs of Paris. Gangs of young men armed with knives and batons attacked Roma people, beating them, burning cars, and damaging property. The attacks were triggered by false accusations of Roma people kidnapping children (Breeden 2019).

These events are examples of numerous cases of collective violence in Europe in recent years. We can clearly see commonalities between them. First, each of the incidents is violent: the perpetrators intend to do harm. Second, the victims are chosen because they belong to certain categories. Third, there are many perpetrators, they represent certain social categories, and, to some extent, their actions are coordinated.

1 Collective Violence

Collective violence can be characterized using Charles Tilly's (2003) definition as a social interaction that 1) inflicts injuries on people or objects; 2) involves at least two perpetrators; and 3) entails coordination of the perpetrators' actions. This definition excludes ac- cidental actions that can lead to injury and includes a wide range of behaviors. However, the social category of both the perpetrators and the victims is overlooked. From the perspective of social identity theory (Tajfel and Turner 1979), any interaction, even one between only two individuals, can have an intergroup character. It is enough for the actors to see themselves as representatives of a broader social category. Therefore, we can understand intergroup collective violence as a specific situation during which there is coordination of the actions of at least two perpetrators who, on behalf of their own group, commit acts of violence toward members of another group.

The events described at the beginning of this paper meet all of the above criteria. While their consequences were relatively small, history shows that intergroup collective violence can have far more devastating effects, from gunfights between gangs of cowboys in frontier towns in the nineteenth century in the United States (Courtwright 1996) to anti-Jewish pogroms in the Russian Empire (Klier and Lambroza 2004) and genocide (Horvitz and Catherwood 2014).

Direct causes and mechanisms that lead people to participate in acts of collective violence were among the fundamental problems underlying the rapid development of social psychology after the Second World War (Kruglanski and Stroebe 2011), and to this day much research on intergroup relations refers to this issue. Unfortunately, collective violence has two characteristics that hinder application of the scientific method to an in-depth analysis of this phenomenon: 1) the complexity of processes and the large number of factors leading to violence and impacting on its dynamics; and 2) limited options for testing theories related to the topic.

1.1Multidimensionality of Collective Violence

Regarding the first problem, researchers emphasize the complexity of processes leading to intergroup collective violence: genocide (Staub 1989), lynching (Tolnay, Beck, and Massey 1989), or anti-Jewish pogroms in the Russian Empire (Lambroza 1981). Multiple types of actor are involved in violent behavior, and multiple factors lie at its origin, including the personality, motivations, and past experiences of individual actors; processes within the groups of perpetrators, victims, and bystanders; and intergroup processes. Finally, contextual factors, such as the structure of the society, culture, and history (Staub 1989, Harff 2003), play an important role. The dynamics of the process come not only from the fact that a particular violent event is an outcome of a complex process, but also derive from the fact that with every outburst of aggression, the relationship between the actors and the situation changes (for example, duration or severity of violence). Importantly, while the complexity of the factors leading to aggression is analyzed by scholars, violence itself is usually treated as a binary phenomenon -that is, as an event that occurs or not. For example, research on genocide mainly focuses on the conditions leading to the occurrence of extreme violence (Harff 2003). Even the complex model proposed by James Waller (2007), describing how ordinary people participate in exterminatory violence in response to authority, is unidimensional in terms of the outcome. The model includes a complex multilevel interactional structure of factors, that shapes one of the two possible outcomes-participating or not. The few researchers who have addressed the differences in the mechanisms leading to violence make comparisons at the level of description and do not go into distinctions between the possible characteristics and functions of violence (Straus 2007). Meanwhile, preliminary archival data on pogrom violence in the eighteenth to twentieth century in Poland shows that the structure of intergroup relations and political factors has an impact on the type of violence (Winiewski 2016).

Within the framework of the studies on individual aggression, the differentiation of forms of aggressive behavior is well established (Buss and Perry 1992, Buss and Durkee 1957). Studies show that differences between forms of inter-individual violence are a function of personality (Kruh, Frick, and Clements 2005) and context (Steinke 1991). For instance, direct and indirect aggressive behavior in interpersonal conflict are related to the type of relationship (Richardson and Green 2006). A multidimensional approach to violence in intimate relationships, underlining the different dynamics and utilities of violence (Johnson 2006) has been an important voice in the long-standing discussion on gender asymmetry in violence (Dobash et al. 1992, Straus and Gozjolko 2014).

1.2 Measuring Collective Violence

The second problem, measurement, seems unsolvable. It is ethically impossible to develop direct measures or experimental manipulations of collective violence. Researchers must use proxy measures or historical data. Archival data are very useful in terms of identifying patterns; however, they do not provide enough control to test the hypotheses. Proxy measures are controversial and are used mainly in studies on aggression at the individual level (Tedeschi and Quigley 1996). The main problems with proxy measures are that a) they do not include aims of the perpetrators (for example, the intended outcomes of violent actions), and b) they ignore multidimensionality (they rarely include more than one form of response (Ritter and Eslea 2005)). Conscious of the aforementioned critique, we believe that in the case of intergroup relations, the measurement of violent intentions can be substituted by the acceptance of such behavior. This is not free of validity threats, such as the bias caused by social desirability (Saunders 1991, Selby 1984); however, it offers two kinds of benefit. First, measuring acceptance allows the inclusion of multiple types of violent behavior, thus allowing multidimensionality; second, this approach taps into an aspect often stressed by scholars -that social acceptance is one of the main factors leading to collective violence (Staub 1989, Harff 2003).

1.3 Factors Differentiating Collective Violence

In a preliminary attempt to validate the multifaceted concept of collective violence, we examine two groups of factors that can potentially differentiate acts of collective violence.

The first group is related to the way in which the perpetrators perceive the group of potential victims, recurring to cognitive categories-stereotypes and intentions-attributed to an outgroup. Most of the concepts describing cognitive appraisals of outgroups predict different behavioral intentions depending on outgroup perception. In addition, structure of intergroup relations is seen as source of these assessments, for example competition and status (Fiske et al. 2002), social context (Alexander, Brewer, and Livingston 2005), or perception and interpretation of the actions of outgroup members (Stephan, Ybarra, and Morrison 2009).

The second group of factors is ideologies - the key element that transforms group frustration into violence (Staub 2002). The role of ideology is crucial not only in explaining why the ingroup is in a negative situation, but also in the process by which solutions emerge. Studies in political psychology indicate that ideological orientations such as right-wing authoritarianism (RWA) or social dominance orientation (SDO) are involved in complex cognitive and emotional processes, leading to fundamentally different effects (Thomsen, Green, and Sidanius 2008).

1.3.1Threats and Intergroup Emotions

Outgroup stereotypes are usually quite complex, encompassing both negative and positive attributes (Fiske et al. 2002). Stereotypes are important in the context of intergroup collective violence and usually include references to threats that an outgroup may pose to the ingroup (Stephan and Stephan 2000). Researchers distinguish two main types of threat: realistic and symbolic (Stephan, Ybarra, and Morrison 2009). The former is a threat to the (broadly defined) wellbeing of the ingroup. It involves the threat of losing political and economic power, including the loss of material and other resources, such as health. The latter is related to perception of the outgroup as threatening to supplant and replace the ingroup's values, customs, traditions, or religion. It is worth emphasizing that both types of threat are based on perceptions and not on actual political or social realities. However, the consequences of the perceived threats are real. Threats evoke emotions, especially feelings of anger and fear (Stephan, Ybarra, and Morrison 2009). From an evolutionary point of view, these two emotions enable the ingroup to deal with threats and motivate it to action (Cottrell and Neuberg 2005).

Various emotions form a basis for specific behaviors aimed at an outgroup (Maitner, Smith, and Mackie 2016, Cuddy, Fiske, and Glick 2007). Research has shown that anger, hatred, and fear are all related to violent behavior, but the nature of the actions motivated by them is different. For example, anger triggers a readiness for direct physical or verbal attack, while fear results in attempts to avoid the outgroup (including physically moving away from it), ignoring it and rejecting any contact with its representatives (Mackie and Smith 2002).

1.3.2Ideological Factors

A variety of collective acts of intergroup violence can also be explained by ideological and social attitudes (Adorno et al. 1950). In this context, individual differences such as concepts of RWA (Altemeyer 1981) and SDO (Pratto et al. 1994) are especially useful. Individuals characterized by a high level of RWA tend to submit to authorities and rules, and react aggressively to all individuals and groups who deviate from the social norms (Altemeyer 1981). Those high on SDO perceive society and intergroup relations in a hierarchical manner, and believe that strong groups should dominate (Pratto et al. 1994). John Duckitt (2001) proposed a dual process model, pointing out distinct roles of RWA and SDO in intergroup prejudice. Authoritarianism motivates people to seek stability by following the existing standards. For this reason, the authoritarians feel threatened by groups that do not fit into the established social order. People high on SDO attach great importance to the status of their own group and its power over other groups. Awareness of intergroup differences in status arouses negative reactions to weaker and subordinate groups, while groups capable of competing become a threat (Duckitt 2006). We hypothesize that different motivations for intergroup aggression-namely, maintaining order for RWA and gaining dominance for SDO-can be related to different types of intergroup collective violent behavior.

1.4Aim of the Project

We argue that taking a social psychological perspective on intergroup collective violence allows variety of violent incidents to be explained. Different aspects of a situation-status differences (Cuddy, Fiske, and Glick 2007), perceived threats (Cottrell and Neuberg 2005), or ideologies (Duckitt 2006)-can lead to different outcomes of social interactions, among them several types of negative, hostile behavior.

In this paper we show that 1) intergroup collective violence is a multidimensional phenomenon; and 2) different forms of aggression are related to specific factors. In a series of studies, we develop a scale for measuring the acceptance of different types of intergroup collective violence. In a final study, we provide initial validation of the concept and the scale.

2Study 1

To explore the content and structure of behavior that participants perceive as an outcome of intergroup conflict, a concept map was constructed. In four independent studies, an open-ended question was asked about imagined intergroup conflict. The answers were pooled and analyzed jointly using the concept mapping method. As this was an exploratory study, no specific hypotheses were formulated.

2.1 Participants

The overall sample consisted of N = 1,420 participants: in study 1a, N = 76 participants (34 males, 42 females; Mage = 23.1, SDage = 2.39); in study 1b, N = 608 participants (128 males, 474 females, six people indicating other gender; Mage = 28.08, SDage = 9.72); in study 1c, N = 240 participants (106 males, 128 females; Mage = 24.24, SDage = 4.07); and in study 1d, N = 496 participants (174 males, 319 females, three people indicating other gender; Mage = 38.97, SDage = 15.55).

2.2 Procedure

All four studies followed a similar procedure (three used an online survey and one was a paper and pencil study). Online versions were distributed via Facebook and regional comment threads on newspaper forums. The paper and pencil version was conducted in a university library. Participants were asked to read a short story about an intergroup conflict, briefly describing a situation in which the presence of a minority group inhabits a middle-sized city somewhere in Poland, leads to tensions between the newcomers and the majority group, the Poles. In three of the studies (1a, 1b, 1d), the minority groups were refugees; in the fourth, the minority group was of one of six ethnic groups (Czech, German, Roma, Ukrainian, Vietnamese, and Russian). After reading the story, the participants were asked to list up to six types of action that the majority members might take against the minority group in the described situation. The goal was to gather as many potential behaviors as possible. Depending on the study, the questionnaires took between ten and twenty minutes to complete. In studies 1a, 1b and 1d, experimental manipulation was introduced into the description of the refugee group; this is not relevant to the present analysis, however.

2.3Results

A concept mapping technique (Jackson and Trochim 2002) was used. Prior to further analysis, data from open-ended questions from all four studies were aggregated by collapsing the semantically matching responses. The original 3,387 entries were organized into a list of 179 independent concepts (types of behavior) used as the unit of analysis. Next, a group of competent judges (N = 22: 5 males, 17 females; Mage = 23.77, SDage = 9.81) were asked to sort the 179 concepts into self-defined categories. The judges were native Polish speakers, independent of the researchers and of the original samples, and were not experts in the field of intergroup behavior. No prior definitions of categories or limit to the number of categories were provided. The only two constraints were 1) the absence of a residual category and 2) inclusion of all the statements in the list (Jackson and Trochim 2002). The number of categories created by the sorters varied from 4 to 33 (M = 7.29, SD = 11.5).

Each judge's categorization was recorded in a 0,1co-occurrence matrix (if statements were in one category = 1; if in two different = 0). Next, matrices from all judges were summed. This matrix was used as entry data for a multidimensional scaling (MDS) technique to represent distances between statements. Based on the scree test, a three-dimensional solution was accepted. Next, using Ward's (1963) hierarchical clustering method, six clusters were established. The final cluster membership for each statement was calculated based on a k-means algorithm using cluster centers from hierarchical clustering to stabilize algorithm performance (Milligan 1980, Steinley and Brusco 2007).

The first cluster was unequivocal, with 28 (M = 0.52, SD = 0.89) types of violent physical behaviors. The second cluster was more diverse than the first, consisting of behavior understood broadly as psychological violence, such as "bullying" as well as specific descriptions of verbal violence, such as "insults" Overall, it contained 38 units of analysis (M = 0.41, SD = 0.93). The third cluster consisted of 28 (M = 0.46, SD = 0.78) types of action meant to discriminate or exclude the minority group. The fourth cluster consisted of descriptions of positive reactions, which included 35 (M = 0.79, SD = 1.32) types of behavior. The fifth cluster was slightly less clear, including descriptions that can be defined as not taking any action in the conflict, but also including words that can be understood as exclusion; therefore, we labeled it as passive/exclusion. There were 24 types of behavior in this cluster (M = 0.39, SD = 0.74). Lastly, the sixth cluster included 21 (M = 0.35, SD = 0.68) descriptions of actions that can be defined broadly as violence mediated by the struc- tures of the state. These types of behavior require active participation but do not assume an actual confrontation. Notably, behaviors from this cluster were listed least frequently by participants.

2.4Discussion

The analysis revealed six dimensions of collective behavior within intergroup conflict: physical violence, verbal/psychological violence, discrimination/exclusion, positive reactions, passiveness/exclusion, and indirect violence. Out of the six clusters, physical violence and positive reactions were the most consistent and had the highest frequency. This might be related to the fact that those behaviors have the strongest influence on the conflict dynamics. Physical violence has the most escalatory power and positive reactions have de-escalatory power: therefore, those two types can be seen as the most prototypical strategies.

3Study 2

The purpose of the second study was to construct a multidimensional scale of acceptance of intergroup collective violence. Given that intergroup collective violence is an issue extremely sensitive to social desirability, the scale was constructed using a similar approach to that used in studies 1a-d. A short story described an intergroup conflict between the majority group-the Poles-and a minority group of new residents in a city in Poland. In the story, it was explicitly stated that the portrayed situation led to open hostility between the members of both groups. A list of items followed that described the potential behaviors that the majority group could undertake in this situation. The participants were asked to rate to what extent each of the actions was justified, using a scale ranging from 1 (fully unjustified) to 7 (fully justified).

The six clusters found in Study 1 served as a framework for item generation. Three researchers-experts in social psychology and intergroup behavior-independently generated statements describing a wide range of behaviors that can appear in the context of intergroup conflict. The initial pool of statements consisted of 291 items. They were further semantically aggregated so that the final pool included only descriptions of unique intergroup behavior. The final pool used in the study contained 96 items: 84 items described violent behaviors and 12 described positive actions. As this was an exploratory study, no specific hypotheses were formulated. The main goal was to see whether the multidimensional structure of intergroup collective violence would be replicated using a different analytical approach to that in Study 1.

3.1 Participants and Procedure

An internet survey was conducted in a representative sample of 1,000 Polish citizens (464 males, 536 females; Mage = 43.41, SDage = 15.79) recruited by an online polling agency. Of these, 182 respondents were excluded from the sample because they finished the questionnaire within 60 seconds and there was no variance in their answers. The final sample consisted of N = 818 participants (372 males, 446 females; Mage = 44.68, SDage = 15.96). First, a series of demographic questions was asked. The participants were then presented with the short story described above and asked to imagine how the Polish inhabitants of the city might act toward the newcomers. The minority group was either Roma or Ukrainians, chosen as they are both perceived by members of Polish society as posing a threat. Items were presented in a randomly rotated order. Lastly, the participants were asked a series of other questions related to intergroup relations that are not analyzed in this report.

3.2 Results

An exploratory factor analysis was conducted in parallel for both outgroups (Roma and Ukrainians). The strategy was to obtain an unambiguous structure that was similar for both groups. The initial analysis was conducted in ML factoring with oblique rotations; the criteria for extracting the number of factors were based on parallel analysis and minimum average partial correlation criteria (Velicer, Eaton, and Fava 2000, Costello and Osborne 2005). The analysis yielded seven factors. Items that had high cross-loadings or low loadings on all factors were removed from both datasets. Finally, we reduced the number of items to three in each factor.

The final questionnaire contained seven sub-scales (Table 3). Reliabilities of the sub-scales were at least satisfactory-disregard turned out to be the only subscale with relatively poor reliability. Relatively high correlations between all of the first four sub-scales suggest that they have a common underlying factor. Almost all of the correlations between the sub-scales describing violent behavior and positive reactions were negative and moderate, except for the assimilation sub-scale, where the correlation was positive but small.

Additionally, the between-group differences were tested using the 2 (group) x 7 (types of violence) mixed ANOVA model (Figure 1).

There was a strong main effect- F (3.03,2474.45) = 1165,35, p < 0.001, r|p2 = 0.59)-for the acceptance of various types of violence. Almost all types were significantly different, with the exception of verbal and indirect violence. Physical violence was found to be the least acceptable, followed by isolation, verbal and indirect violence, disregard, and forced assimilation. Positive actions were seen as the most justifiable. The level of acceptance for various types of violence differed significantly depending on the ethnicity of the minority group (Figure 1): violence against Roma was generally seen as more justified.

3.3Discussion

Seven sub-scales of support for violence were extracted. The first two-physical and verbal violence- represent active, confrontational types of violence that require direct interaction. The distinction between the two is coherent with studies on aggression and violence on an individual level. For example, the literature on bullying shows that despite the fact that verbal and physical peer violence tend to co-occur, their nature is different. Each type is shaped by different structural factors such as class size, gender, or age (Björkqvist 1994, Scheithauer et al. 2006).

The next two-isolation and indirect violence-can be considered active but non-confrontational types of violence. The difference between these two can be seen in the context of their utility. An isolation strategy stifles conflict by excluding the outgroup, while indirect violence is set on solving issues with the minority, but by employing a system to solve it. The first of these strategies is part of both the BIAS map model (Cuddy, Fiske, and Glick 2007) and the sociofunctional threat theory (Cottrell and Neuberg 2005). In both models, this kind of behavior is related to contempt or disgust. Indirect forms of violence and aggression on an individual level are observed in peer bullying. Aside from utility, this form allows the aggressor to remain unidentified, increasing the effect/danger ratio (Björkqvist 1994). On the collective level, for example, a direct and indirect distinction in the context of civil wars is proposed and explained by ideological factors, namely the level of political parity between factions (Balcells 2011).

The fifth and sixth sub-scales can be considered as specific, non-active, and non-confrontational types of violence. They are both characterized by rather hostile intentions toward the members of the outgroup. Though the behaviors represented by forced assimilation can be considered as consisting of some type of conflict resolution effort (which is also suggested by low, positive correlation with the positive actions subscale), those efforts only take into consideration the needs of the majority group, forcing the members of the minority to adapt to their norms.

The level of acceptance of various forms of intergroup collective violence differs depending on the ethnicity of the target group. Possible reasons for this include differences in the status of these groups in Polish society or in the way they are perceived by Poles. While the Roma are stereotypically seen as cold and incompetent, the Ukrainians are perceived as incompetent but rather warm (Bilewicz et al. 2009). The negative stereotype of the Roma is one reason they are often placed outside the realm of social norms: behav- iors that would be unacceptable if they were presented toward any other group are seen as perfectly justifiable toward the Roma people (Kende, Hadarics, and Lášticová 2017). The obtained results have important implications for the further analysis. They suggest that the specific characteristics of the target group play a role in explaining the behavioral intentions toward them. Those characteristics can also interact differently with potential predictors of intergroup collective violence, both contextual, such as the perceived threat, or individual, such as ideology. For instance, people with authoritarian tendencies can be against physical, non-normative violence toward the Ukrainians-a group that is protected by the social norms-but at the same time accept the same type of violence toward the Roma people. For that reason, the analysis of predictors of intergroup collective violence should be conducted separately for various ethnic groups.

4Study 3

The main purpose of study 3 was to investigate the relationship between the perception of intergroup threat and the acceptance of intergroup collective violence.

Using as our theoretical framework the integrated threat theory, which describes two basic components of perceived threat as a base of prejudice (Stephan and Stephan 2000), and the dual process model, which identifies parallel mechanisms of group dominance (SDO) and control over a dangerous world (RWA) as a source of prejudice (Duckitt 2006), we wanted to investigate whether acceptance of different forms of intergroup collective violence is dependent on perceptions of threat from the outgroup and perceptions of social reality. As this was a somewhat exploratory study, we had only initial intuition, and instead of forming specific hypotheses, we aimed to answer two research questions:

1) How is perceived intergroup threat related to the acceptance of various forms of intergroup collective violence?

2) How do RWA and SDO relate to the acceptance of various forms of intergroup collective violence?

4.1 Participants and Procedure

A representative sample of 1,019 Polish citizens (487 males, 532 females; Mage = 47.54, SDage = 17.83) took part in a nationwide survey, conducted using a computer-assisted personal interviewing method. The participants were chosen at random from the entire population of adult Poles using a citizen identification number. The survey was part of a larger study that included several sociopsychological measures (attitudes toward various minority groups) that are not further discussed here. The scale measuring acceptance of intergroup collective violence was the major dependent variable; it appeared at the end of the survey, after the independent variables of interest. Overall, the survey contained 289 questions and took around 60 minutes to complete.

4.1.1 Measures

Acceptance of intergroup collective violence was measured using the scale constructed in study 2, the only difference being that the minority group consisted of either Ukrainians or Jews.

Right-wing authoritarianism was measured using a short version of the RWA scale (Beierlein et al. 2014). The scale consisted of statements such as "Social rules should be strictly enforced." Participants rated their agreement on a five-point scale, from 1 meant "I definitely disagree" to 5 "I definitely agree"

Likewise, for social dominance orientation, because of the study constraints, we used a shortened version of the SDO scale (Pratto et al. 1994), with four statements coded on a five-point scale, where 1 meant "I definitely dis agree" and 5 "I definitely agree". Items included "Superior groups should dominate inferior groups"

Perception of intergroup threats was measured using agreement with single items. Adjusting questions to specific minorities, we decided to ask two questions about realistic threats-those concerning safety and resources. Participants were asked to indicate to what extent they agreed with each statement on a sevenpoint scale, where 1 meant "definitely disagree" and 7 "definitely agree"

Symbolic threat: "Do the values of the [group] pose a threat to the values of the Poles?"

Safety threat: "Does the presence of the [group] pose a threat to the safety and health of the Poles?"

Realistic threat: "Does the presence of the [group] reduce the chances of Poles finding a job?"

Demographic questions (age, gender, and place of residence) and political views (1 for definitely rightwing views to 7 for definitely left-wing) were treated as controls. Descriptive statistics of all measurements are reported in Table 4, and correlations are reported in Table 6.

4.2Results

First, a confirmatory factor analysis was conducted to see whether the structure of the scale found in study 2 remained the same. Next, in order to test relations between political ideology and threat perceptions, a series of regression analyses were conducted.

4.2.1 Structure of the Scale

A confirmatory factor analysis was conducted for both sub-groups: Ukrainians (x2 (168) = 642.77, p < .001, CFI = 0.97, TLI = 0.96, RMSEA = 0.07 CI [0.07, 0.08]) and Jews (x2 (168) = 556.61, p < 0.001, CFI = 0.97, TLI = 0.96, RMSEA = 0.07 CI [0.06, 0.07]). The fit indexes showed that the model was adequate. Next, in order to see whether the structure of the data was similar regardless of the minority group, measurement invariance was assessed using the semTools package in R. The analysis showed that the factor loadings, the intercepts, and the means are equal for the two groups-that is, scalar invariance was achieved. This model had the lowest AIC and BIC, which suggests that the groups can be compared by their scores on all seven of the sub-scales.

4.2.2 Impact of Intergroup Threat and Political Ideology

First, zero-order correlations were calculated for all the variables in both samples (see Table 5).

The results show similar patterns of correlations between violence scales and between the predictors (threats, SDO, and RWA) for both outgroups. Interestingly, the patterns of correlations between predictors and violence scales differ between the groups. This is notable in the case of ideologies and political views and their relation to passive forms of violence.

To assess the impact of the perception of the various types of threat on the acceptance of different forms of violence, a series of regression analyses were conducted in which six sub-scales of violence (excluding positive actions) were treated as dependent variables. In the first step, the demographic variables were entered as control variables; the second step included political views and three types of intergroup threat. The models were calculated for the two sub-samples separately.

SDO seems to be a stable predictor of acceptance of all active and direct forms of violence. Only a small difference between the target groups is noted concerning the explanatory power of SDO in the acceptance of verbal and indirect violence; the impact of SDO seems to be greater in relation to Jews as target group.

RWA has no relation with the acceptance of most of the active types of violence, except indirectly toward Ukrainians. However, it is the most important predictor for both passive types of action. Furthermore, it explains a substantially larger portion of variance for the acceptance of violence against Ukrainians over Jews.

Symbolic threat is an important predictor of acceptance of all direct violent acts, but only against Jews. It also predict forced assimilation of Jews and lower acceptance of forced assimilation of Ukrainians.

Safety threat is the second most important predictor of acceptance of physical violence, regardless of the minority group. It predicts acceptance of active acts of violence-verbal, isolation, and indirect-but only against the Ukrainian group. Interestingly, perceived safety threats also substantially differentiate the level of acceptance of forced assimilation for both groups. Those who perceive Jews as threatening accept forced assimilation less, while those who perceive Ukrainians as a safety threat accept forced assimilation more.

Finally, a realistic economic threat seems to be related to acceptance of verbal and indirect violence toward both minorities and, more specifically, to accepting acts of isolation of Jews and forced assimilation of Ukrainians.

4.3Discussion

The results confirm that the structure of acceptance of intergroup collective violent behavior is multidimensional. SDO is related to the acceptance of all direct and confrontational forms of violence that aim to gain superiority over the outgroup. RWA is mainly related to passive violence, specifically forcing the outgroup to adjust to the rules and culture of the ingroup. These results are in line with the dual process model (Duckitt 2001).

The relation between threats and the acceptance of specific types of violence is relatively coherent. Physical violence is dependent on basic existential fear, related to threats to a group's own safety. Verbal violence, still confrontational but much less severe, is related to a realistic economic threat. This seems to be the "appropriate" response to a threat to the group's prosperity rather than to its existence. This threat also fuels acceptance of indirect violence-acts of open pressures on authorities to deal with the problem using violence.

Interestingly, part of the relationship between threats and acceptance of violence seems to be dependent on the group context. This also seems to be coherent with our predictions, as described in section 3.4; however, our data do not allow for a more detailed answer about the origins of these effects. We can only speculate that other factors related to historical and contemporary relations, not included in our studies, could help explain the findings. The Jewish population in Poland was a big minority in the past and is now almost nonexistent, while the Ukrainian minority and migrants are the biggest outgroup in Poland, and still growing (Davies 2005). A historical perspective also shows that Jews were a minority that occupied a different social niche from Ukrainians and played a different cultural role in the collective imagination (Brataniec 2016, Cala 1992).

5General Discussion

In a series of studies, we developed and initially validated a scale for the acceptance of intergroup collective violence. Using various methods we showed that the perception and acceptance of intergroup collective violence are multidimensional. It is worth pointing out that we observed a substantial overlap in the structure of a) how participants perceive aggressive collective behavior and b) the dimensions of acceptance of such behavior by ingroup members. Although the number of dimensions of violence found differs slightly between Study 1 and Study 2, the meanings of the behavior included in the clusters and factorial solutions overlap substantially.

The six dimensions of intergroup collective violence identified in our studies seem to fit to well-grounded psychological theories. We can clearly see the correspondence with various conceptualizations of individual level aggression, such as the concepts of direct-indirect, physical-verbal, and active-passive hostility (Buss and Perry 1992). Although intergroup collective violence has not been analyzed as a multidimensional phenomenon, our results are coherent with theories that suggest that differential cognitive appraisals of outgroups lead to different emotions and behavioral intentions (Cuddy, Fiske, and Glick 2007, Cottrell and Neuberg 2005).

The results suggest that acceptance of specific types of intergroup violence can be seen as part of the broader process shaping intergroup relations. Previous analysis of collective violence has underlined a variety of structural factors influencing the process (Staub 1989, Harff 2003). Our results indicate that those structural factors influence perceptions of ingroup and outgroups (stereotypes and threats), and-using the social identity approach (Tajfel and Turner 1979)explain their role in shaping intergroup violence .

To our knowledge, this is the first study to investigate systematically the varieties of intergroup collective violence. By developing a scale to measure acceptance of collective acts of intergroup violence, we believe that we can start to uncover the mechanisms that lead various intergroup conflicts to different outcomes. Although not free from limitations, it seems that the presented scale can be utilized for studying the complex phenomenon of intergroup collective violence.

5.1Limitations

As mentioned in the introduction, social desirability is the main problem with studies on violence. Direct questions of acceptance of acts of violence are similar to direct questions on perpetrated violence and are subject to a strong social norm. Therefore, part of the variance of answers-refraining from providing a true answer-is explained not by the acceptance of violence, but by compliance to social norms. Taking this into account, one needs to treat the results with caution. However, the type of analysis that we conducted, focusing on the relationship between acceptance of violence and other factors rather than on the diagnosis, is relatively robust. Nevertheless, social desirability should be controlled for in future studies.

A second problem is that we would like to treat this scale and the data obtained as a proxy measure for collective violence and, as we mentioned in the introduction, social acceptance and norms are very important factors in shaping intergroup relations, including collective violence. However, between the wide acceptance of collective violence and actual violence there are processes that need to be taken into account, especially since some works underline the role of identity in the norms-attitudes-behaviors relationship (Terry and Hogg 1996).

5.2Further Research

Our taxonomy, the scale measuring acceptance of collective violence, and its initial validation-showing relations with social perceptions and ideology-seem promising for future research. Firstly, it is worth exploring further whether the context of the social situation, intergroup dynamics, and the social identity process shape the dynamic of acceptance of different collective intergroup acts of violence as a means of solving conflicts, gaining resources, protecting the ingroup's values, etc. Secondly, the scale can be used to test and develop further two theoretical models of collective violence: Ervin Staub's (1989) model of genocidal violence and Peter Glick's (2002) ideological model of scapegoating. Thirdly, the scale and the concept of multifaceted intergroup collective violence should be used to develop a more comprehensive model of collective violence, including all the structural and psychological factors that allow the prediction not only of an outburst of violence, but also its type.

Footnote

1 Preparation of this manuscript was supported by Grant 2014/13/B/HS6/04077 from the National Science Centre.

References

References

Aderet, Ofer. 2018. "Get Out of Germany, Jewish Pigs": Jewish Restaurant in Germany Attacked by neo-Nazis. Haaretz, 8 September. https://www.haaretz.com/worldnews/europe/jewish-restaurant-in-german-town-chemnitz-attacked-by-neo-nazis-1.6464348 (accessed 1 April 2019).

Adorno, Theodor. W., Else Frenkel-Brunswik, Daniel J. Levinson, and Nevitt R. Sanford. 1950. The Authoritarian Personality. Oxford, UK: Harpers.

Alexander, Michele G., Marilynn B. Brewer, and Robert W. Livingston. Putting Stereotype Content in Context: Image Theory and Interethnic Stereotypes. Personality and Social Psychology Bulletin 31 (6): 781-94. doi: 10.1177/0146167204271550

Altemeyer, Bob. 1981. Right-wing Authoritarianism: Winnipeg: University of Manitoba Press.

Balcells, Laia. 2011. Continuation of Politics by Two Means: Direct and Indirect Violence in Civil War. Journal of Conflict Resolution 55 (3): 397-422. doi: 10.1177/0022002711400865

Beierlein, Constanze, Frank Asbrock, Mathias Kauff, and Peter Schmidt. 2014. Die Kurzskala Autoritarismus (KSA-3): Ein ökonomisches Messinstrument zur Erfassung dreier Subdimensionen autoritärer Einstellungen, GESIS-Working Paper. Mannheim: GESIS - Leibniz-Institut für Sozialwissenschaften.

Bilewicz, Michal, Marcin Bukowski, Aleksandra Cichocka, Mikolaj Winiewski, and Adrian Wójcik. 2009. Uprzedzenia Etniczne w Polsce: Raport z Ogólnopolskiego Sondażu Polish Prejudice Survey 2009. Warsaw: Centre for Research on Prejudice.

Bilewicz, Michał, Dominika Bulska, Maria Babińska, Agnieszka Haska, and Mikołaj Winiewski. 2018. Marzec w lutym? Studium stosunku Polaków do Żydów i historii Holokaustu w kontekście debaty wokół ustawy o IPN. Nauka, no. 2, 7-41.

Bilewicz, Michal, Wiktor Soral, Marta Marchlewska, and Mikołaj Winiewski. 2017. When Authoritarians Confront Prejudice: Differential Effects of SDO and RWA on Support for Hate-Speech Prohibition. Political Psychology 38 (1): 87-99. doi: 10.1111/pops. 12313

Björkqvist, Kaj. 1994. Sex Differences in Physical, Verbal, and Indirect Aggression: A Review of Recent Research. Sex Roles 30 (3): 177-88. doi: 10.1007/bf01420988

Brataniec, Katarzyna. 2016. Narody i Ich Stereotypy a Wizerunek Ukraińców w Świadomości Zbiorowej Polaków. Państwo i Społeczeństwo 16 (1):59-74.

Breeden, Aurelien. 2019. Rumors of Abduction Flare Up in France, and Roma Become Scapegoats. New York Times, 29 March, A9. https://www.nytimes.com/2019/03/28/ world/europe/roma-kidnap-rumors-france.html

Buss, Arnold H., and Ann Durkee. 1957. An Inventory for Assessing Different Kinds of Hostility. Journal of Consulting Psychology 21 (4): 343-49. doi: 10.1037/h0046900

Buss, Arnold H., and Mark Perry. 1992. The Aggression Questionnaire. Journal of Personality and Social Psychology 63 (3): 452-59.

Cala, Alina. 1992. Wizerunek Żyda w polskiej kulturze ludowej. Warsaw: Wydawnictwo Uniwersytetu Warszawskiego.

Costello, Anna B., and Jason W. Osborne. 2005. Best Practices in Exploratory Factor Analysis: Four Recommendations for Getting the Most from Your Analysis. Practical Assessment, Research and Evaluation 10 (7): 1-9.

Cottrell, Catherine A., and Steven L. Neuberg. 2005. Different Emotional Reactions to Different Groups: A Sociofunctional Threat-Based Approach to "Prejudice". Journal of Personality and Social Psychology. 88 (5): 770-89. doi: 10.1037/0022-3514.88.5.770

Courtwright, David T. 1996. Violent Land. Cambridge, MA: Harvard University Press.

Cuddy, Amy J. C., Susan T. Fiske, and Peter Glick. 2007. The BIAS Map: Behaviors from Intergroup Affect and Stereotypes. Journal of Personality and Social Psychology 92 (4): 631-48. doi: 10.1037/0022-3514.92.4.631

Davies, Norman. 2005. God's Playground: A History of Poland. Vol. 2, 1795 to the Present. New York: Oxford University Press.

Dobash, Russell P., R. Emerson Dobash, Margo Wilson, and Martin Daly. 1992. The Myth of Sexual Symmetry in Marital Violence. Social Problems 39 (1): 71-91. doi: 10.2307/3096914

Duckitt, John. 2001. A Dual-Process Cognitive-Motivational Theory of Ideology and Prejudice. Advances in Experimental Social Psychology 33:41-113.

Duckitt, John. 2006. Differential Effects of Right Wing Authoritarianism and Social Dominance Orientation on Outgroup Attitudes and Their Mediation by Threat from and Competitiveness to Outgroups. Personality and Social Psychology Bulletin 32 (5): 684-96. doi: 10.1177/ 0146167205284282

Fiske, Susan T., Amy J. C. Cuddy, Peter Glick, and Jun Xu. 2002. A Model of (Often Mixed) Stereotype Content: Competence and Warmth Respectively Follow from Perceived Status and Competition. Journal of Personality and Social Psychology 82 (6): 878-902. doi: 10.1037/00223514.82.6.878

Glick, Peter. 2002. Sacrificial Lambs Dressed in Wolves' Clothing: Envious Prejudice, Ideology, and the Scapegoating of Jews. In Understanding Genocide: The Social Psychology of the Holocaust, eds. Leonard S. Newman and Ralph Erber, 113-42. New York: Oxford University Press.

Harff, Barbara. 2003. No Lessons Learned from the Holocaust? Assessing Risks of Genocide and Political Mass Murder since 1955. American Political Science Review 97 (1): 57-73.

Horvitz, Leslie Alan, and Christopher Catherwood. 2014. Encyclopedia of War Crimes and Genocide. New York: Facts On File.

Jackson, Kristin M., and William M. K. Trochim. 2002. Concept Mapping as an Alternative Approach for the Analysis of Open-Ended Survey Responses. Organizational Research Methods 5 (4): 307-36. doi: 10.1177/109442802237114

Johnson, Jeff W. 2000. A Heuristic Method for Estimating the Relative Weight of Predictor Variables in Multiple Regression. Multivariate Behavioral Research 35 (1): 1-19. doi: 10.1207/s15327906mbr3501_1

Johnson, Jeff W. 2004. Factors Affecting Relative Weights: The Influence of Sampling and Measurement Error. Organizational Research Methods 7 (3): 283-99. doi: 10.1177/1094428104266018

Johnson, Michael P. 2006. Conflict and Control: Gender Symmetry and Asymmetry in Domestic Violence. Violence Against Women 12 (11): 1003-1018. doi: 10.1177/1077801206293328

Kende, Anna, Márton Hadarics, and Barbara Lášticová. 2017. Anti-Roma Attitudes as Expressions of Dominant Social Norms in Eastern Europe. International Journal of Intercultural Relations 60:12-27. doi: 10.1016/ j.ij intrel.2017.06.002

Klier, John Doyle, and Shlomo Lambroza. 2004. Pogroms: Anti-Jewish Violence in Modern Russian History. New York: Cambridge University Press.

Kruglanski, Arie W., and Wolfgang Stroebe. 2011. Handbook of the History of Social Psychology. New York: Routledge.

Kruh, Ivan P., Paul J. Frick, and Carl B. Clements. 2005. Historical and Personality Correlates to the Violence Patterns of Juveniles Tried as Adults. Criminal Justice and Behavior 32 (1): 69-96. doi: 10.1177/0093854804270629

Lambroza, Shlomo. 1981. The Pogrom Movement in Tsarist Russia, 1903-06. Ann Arbor: University Microfilms International.

Mackie, Diane M., and Eliot R. Smith. 2002. From Prejudice to Intergroup Emotions: Differentiated Reactions to Social Groups. New York: Psychology Press.

Maitner, Angela T., Eliot R. Smith, and Diane M. Mackie. 2016. Intergroup Emotions Theory: Prejudice and Differentiated Emotional Reactions toward Outgroups. In The Cambridge Handbook of the Psychology of Prejudice, eds. Chris G. Sibley and Fiona Kate Barlow, 111-30. Cambridge: Cambridge University Press.

Milligan, Glenn W. 1980. An Examination of the Effect of Six Types of Error Perturbation on Fifteen Clustering Algorithms. Psychometrika 45 (3): 325-42. doi: 10.1007/ bf02293907

Pilbeam, Bruce. 2011. Eurabian Nightmares: American Conservative Discourses and the Islamisation of Europe. Journal of Transatlantic Studies 9 (2): 151-71. doi: 10.1080/14794012.2011.568166.

Pratto, Felicia, Jim Sidanius, Lisa M. Stallworth, and Bertram F. Malle. 1994. Social Dominance Orientation: A Personality Variable Predicting Social and Political Attitudes. Journal of Personality and Social Psychology 67 (4):741-63. doi: 10.1037/0022-3514.67.4.741

Richardson, Deborah S., and Laura R. Green. 2006. Direct and Indirect Aggression: Relationships as Social Context. Journal of Applied Social Psychology 36 (10): 2492-2508. doi: 10.1111 /j.0021-9029.2006.00114.x

Ritter, Dominik, and Mike Eslea. 2005. Hot Sauce, Toy Guns, and Graffiti: A Critical Account of Current Laboratory Aggression Paradigms. Aggressive Behavior 31 (5): 407-19. doi: 10.1002/ab.20066

Saunders, Daniel G. 1991. Procedures for Adjusting Self-Reports of Violence for Social Desirability Bias. Journal of Interpersonal Violence 6 (3):336-44. doi: 10.1177/088626091006003006

Scheithauer, Herbert, Tobias Hayer, Franz Petermann, and Gert Jugert. 2006. Physical, Verbal, and Relational Forms of Bullying Among German Students: Age Trends, Gender Differences, and Correlates. Aggressive Behavior 32 (3): 261-75. doi: 10.1002/ab.20128

Selby, Michael J. 1984. Assessment of Violence Potential Using Measures of Anger, Hostility and Social Desirability. Journal of Personality Assessment 48 (5): 531-44. doi: 10.1207/s15327752jpa4805_13

Staub, Ervin. 1989. The Roots of Evil: The Origins of Genocide and Other Group Violence. New York: Cambridge University Press.

Staub, Ervin. 2002. The Psychology of Bystanders, Perpetrators, and Heroic Helpers. In Understanding Genocide: The Social Psychology of the Holocaust, eds. Leonard S. Newman and Ralph Erber: 315-41, New York: Oxford University Press.

Steinke, Pamela. 1991. Using Situational Factors to Predict Types of Prison Violence. Journal of Offender Rehabilitation 17 (1-2): 119-32. doi: 10.1300/J076v17n01_09

Steinley, Douglas, and Michael J. Brusco. 2007. Initializing K-means Batch Clustering: A Critical Evaluation of Several Techniques. Journal of Classification 24 (1): 99-121. doi: 10.1007/s00357-007-0003-0

Stephan, Walter G., and Cookie White Stephan. 2000. An Integrated Threat Theory of Prejudice. In Reducing Prejudice and Discrimination., ed. Stuart Oskamp, 23-45. Mahwah, NJ: Lawrence Erlbaum.

Stephan, Walter G., Oskar Ybarra, and Kimberly Rios Morrison. 2009. Intergroup Threat Theory. In Handbook of Prejudice, Stereotyping, and Discrimination, ed. Todd D. Nelson, 43-59. Mahwah, NJ: Lawrence Erlbaum.

Straus, Murray A., and Kristi L. Gozjolko. 2014. "Intimate Terrorism" and Gender Differences in Injury of Dating Partners by Male and Female University Students. Journal of Family Violence 29 (1): 51-65. doi: 10.1007/ s10896-013-9560-7

Straus, Scott. 2007. Second-Generation Comparative Research on Genocide. World Politics 59 (3): 476-501. doi: 10.1017/S004388710002089X

Tajfel, Henri, and John C. Turner. 1979. An Integrative Theory of Intergroup Conflict. Social Psychology of Intergroup Relations 2:33-47.

Tedeschi, James T., and Brian M. Quigley. 1996. Limitations of Laboratory Paradigms for Studying Aggression. Aggression and Violent Behavior 1 (2): 163-77. doi: 10.1016/1359-1789(95)00014-3

Terry, Deborah J., and Michael A. Hogg. 1996. Group Norms and the Attitude-Behavior Relationship: A Role for Group Identification. Personality and Social Psychology Bulletin 22 (8): 776-93. doi: 10.1177/0146167296228002

Thomsen, Lotte, Eva G. T. Green, and Jim Sidanius. 2008. We Will Hunt Them Down: How Social Dominance Orientation and Right-Wing Authoritarianism Fuel Ethnic Persecution of Immigrants in Fundamentally Different Ways. Journal of Experimental Social Psychology 44 (6): 1455-64. doi: 10.1016/j.jesp.2008.06.011

Tilly, Charles. 2003. The Politics of Collective Violence. New York: Cambridge University Press.

Tolnay, Stewart E., E. M. Beck, and James L. Massey. 1989. Black Lynchings: The Power Threat Hypothesis Revisited. Social Forces 67 (3): 605-23. doi: 10.1093/sf/67.3.605

Velicer, Wayne F., Cheryl A. Eaton, and Joseph L. Fava. 2000. Construct Explication through Factor or Component Analysis: A Review and Evaluation of Alternative Procedures for Determining the Number of Factors or Components. In Problems and Solutions in Human Assessment: Honoring Douglas N. Jackson at Seventy, eds. Richard D. Goffin and Edward Helmes, 41-71. Boston, MA: Springer US.

Walker, Shaun. 2018. Refugees Crossing from Bosnia "'Beaten and Robbed by Croatian Police". Guardian, 15 August. https://www.theguardian.com/world/2018/aug/ 15/refugees-crossing-from-bosnia-beaten-and-robbed-bycroatian-police (accessed 1 April 2019)

Waller, James E. 2007. Becoming Evil: How Ordinary People Commit Genocide and Mass Killing, 2nd ed. New York: Oxford University Press.

Ward, Joe H. 1963. Hierarchical Grouping to Optimize an Objective Function. Journal of the American Statistical Association 58 (301): 236-44. doi: 10.1080/01621459.1963.10500845

Winiewski, Mikołaj. 2016. Analiza Przemocy Kolektywnej z Perspektywy Psychologii Społecznej: Próba Integracji. In Przemoc antyżydowska i konteksty akcji pogromowych na ziemiach polskich wXX wieku, eds. Konrad Zieliński and Kamil Kijaek, 47-64. Lublin, Poland: UMCS.