Threat and reactions to violated expectations in groups: Adding control to the equation

ABSTRACT Group members frequently face violations of group-based expectations by in- and outgroup members’ behaviour. Responses to such violations include confronting the violators and escaping from the group. The current article presents a novel, integrative model explaining how and why violations of group-based expectations elicit these reactions. We argue that group members experience violations of expectations as threats to their social selves, which motivates them to react at all. Crucially adding to extant theorising, we argue that how group members react depends on their perceived control, a largely overlooked factor thus far. Herein, we synthesise our model’s empirical basis, which spans studies on different violations (violations of ingroup norms, norms for cooperation, and expectations based on political ideology) and different group contexts (e.g., attitude-based lab groups, learning groups, Facebook groups). Moreover, we discuss boundary conditions for the model’s applicability and its potential for integration with classic research on social identity management.

leadership by voicing their disapproval or rallying to impeach Johnson (e.g., Elgot & Stewart, 2022) or consider escape by quitting the party themselves.Both are typical examples of reactions to violations of group-based expectations (e.g., Frings & Abrams, 2010;Marques, Abrams, Páez, et al., 2001;Sani, 2008).But which factors motivate group members to engage in confrontation and escape when facing violations of group-based expectations?
Prior research addressing this question has been disjointed and has focused on different explanations.Research on the emergence of confrontational reactions has primarily focused on threat as an underlying motivation (e.g., Abrams et al., 2000;Marques, Abrams, Páez, et al., 2001), while work on escape reactions has focused on changed perceptions of the group's identity (e.g., Sani, 2008).This article presents a novel model addressing what motivates confrontation and escape that goes beyond previous research in two important ways.First, it includes both confrontation and escape responses, thus bridging the gap between two separate areas of research.Second and more importantly, it is the first to explicitly incorporate group members' perceived control as a factor influencing their reactions to violations of group-based expectations.Thereby, it enriches our understanding of the motivational factors underlying reactions to such violations.
In what follows, violations of group-based expectations always refer to behaviours that negatively depart from individuals' expectations of which behaviour is typically shown or ought to be shown in a certain group context (e.g., violations of identity-defining group norms or norms ensuring group success).First, we will discuss why such violations may be threatening, concluding that merely focusing on threat is not sufficient to explain group members' reactions to violations of group-based expectations.We will argue that it is instead paramount to include perceived control as a motivational factor as well.After reviewing research on potential sources of perceived control in group contexts, we will describe our model.Subsequently, we will review and synthesise the research program the model builds upon.The final section of this article is then devoted to discussing the proposed model's applicability to other situations and a possible agenda for future research.

Group-based expectancy violations and threat
Social and motivational psychologists study a wide array of phenomena they term threats.These include mortality salience, reverse-coloured playing cards, thinking about being laid off, goal conflicts, and personal uncertainty (for a review, see Jonas et al., 2014).In synthesising research on these diverse phenomena, Jonas and colleagues have proposed that "all threats involve the experience of a discrepancy" (Jonas et al., 2014, p. 221).Following this definition, violations of group-based expectations should be threats, as they involve a discrepancy between the observed actual and the expected behaviour.
The idea that violations of group-based expectations are threats is not novel but well grounded in prior theorising.For instance, Festinger (1950) theorises that group-norm violations should be threats because they cause uncertainty.They do so by questioning the validity of opinions not rooted in physical reality that can only be validated through (ingroup) consensus, a lack of which may impede the attainment of group goals.The subjective group dynamics approach (e.g., Abrams et al., 2000) suggests that groupnorm violations threaten a group's distinctiveness and the positivity of its identity, both of which partly rest on group norms.Consequently, groupnorm violations might reduce the group's ability to provide its members with a stable, positive social identity (see also Marques, Abrams, & Serôdio, 2001).Social psychological research on group schisms makes a similar claim (for a review, see Sani, 2008).This work proposes that (attempted) changes to a group's norms can lead some members to experience identity subversion, which is the impression that the group's identity has fundamentally changed or no longer exists (Sani & Reicher, 1998, 1999).In other words, they perceive a discrepancy between their expectations (how the group ought to be) and the actual state of affairs (how it is).But also violations of more general societal norms by in-group members have been claimed to be threatening: They might tarnish other group members' image (due to them being perceived as similar to the deviate; Eidelman & Biernat, 2003) or of the group as a whole (Chekroun & Nugier, 2011).
In summary, there seems to be a solid basis for assuming that violations of group-based expectations -of which (group-) norm violations are one example -are threatening.However, a closer look reveals that while the work reviewed so far has argued in favour of threat as a mechanism, it has not provided direct evidence that observers actually experience violations of group-based expectations as threatening.Instead, it uses the term "threat" to refer to a (supposedly objective) situational characteristic.For instance, Chekroun and Nugier (2011) extrapolate their key variable, "threat to social image", from items assessing whether respondents thought about the image others may have of their group when observing norm-violating behaviour.Eidelman and Biernat (2003), in turn, do not assess "fear of being seen as similar" in any form.
The same applies to research on the subjective group dynamics approach, which shows that individuals who violate their in-group's norms are evaluated less favourably than individuals who adhere to the group's norms (e.g., Abrams et al., 2000;Marques, Abrams, & Serôdio, 2001;Pinto et al., 2010).Very indirectly supporting the idea that this is due to threat, researchers have found that such confrontation responses are stronger following violations that should be especially threatening (i.e., violations of norms important to the individual or the group's identity; Marques, 1990;Marques et al., 1988; violations in contexts where a group's positive identity is insecure, Marques, Abrams, & Serôdio, 2001).More direct evidence for the role of threat stems from the work by Frings et al. (2012).These authors showed that facing an ingroup norm-violator (descriptively) induced a physiological threat state among those with low coping resources.Nonetheless, it seems justified to conclude that evidence of group members truly experiencing violations of group-based expectations as threatening is lacking.This is especially noteworthy given that threat is proposed to underlie various forms of confrontational behaviour in response to violations of group-based expectations (sometimes called social control responses, 1Frings & Pinto, 2018).These include communicating disapproval to those violating expectations (e.g., Chekroun & Nugier, 2011;Frings & Abrams, 2010), nominating them for unappealing tasks (e.g., Schachter, 1951), evaluating them negatively (e.g., Abrams et al., 2000;Eidelman & Biernat, 2003;Levine & Ruback, 1980;Marques & Yzerbyt, 1988), and even excluding them from the group (Eidelman et al., 2006).In short, individuals may try persuading and punishing deviates while allowing them to remain group members, but may also exclude them from the group (cf.Frings & Pinto, 2018).
Interestingly, those violating their ingroup members' expectations by violating ingroup norms face remarkably similar reactions as outgroup critics (e.g., Hornsey et al., 2004).Outgroup criticism represents a different violation of group-based expectations.In such cases, the group's treatment and evaluation are discrepant from group members' own, more positive expectations.What Branscombe et al. (1999) call "value threats" may be one example of such a discrepancy.However, despite explicitly using the label "threat", also this work has long fallen short of assessing whether group members actually experience threat when their group-based expectations are violated.Newer studies have slightly remedied this issue by looking at negative affect caused by situations involving identity threat (e.g., Costarelli, 2009;de Vreeze & Matschke, 2019;Schmader et al., 2015).However, the measures employed in this work are often too broad to allow drawing conclusions on experienced threat (for an exception, see van der Toorn et al., 2015), an issue we have remedied in our own work discussed below.
Taken together, prior research repeatedly made the point that violations of group-based expectations are situational threats.Based on the definition of threat as involving a discrepancy, we concur with this notion.However, from a social-motivational point of view, limiting one's study of the motivational factors that shape reactions to violations of group-based expectations to perceived and/or experienced threat appears overly simplistic.This perspective suggests that when individuals face a self-relevant situation, they do not only consider the situation's possible implications for themselves.Rather, they also take the resources at their disposal for dealing with the situation into account (e.g., Blascovich & Tomaka, 1996).It seems well justified to assume that situations involving violations of group-based expectations are self-relevant, given that these violations have real or imagined (negative) consequences for the individual.Consequently, a social-motivational perspective on violations of group-based expectations suggests an even more important gap in prior research than the lack of direct evidence that group members experience threat: a disregard for the resources group members have at their disposal for dealing with the threatening situation (for an exception, see Frings et al., 2012).This is especially true for perceived control, a variable ubiquitous in (social) psychological research.

Sources of control in group contexts
Perceived control is one of the key resources individuals consider when facing a demanding, dangerous, or uncertain situation (Blascovich & Tomaka, 1996).Herein, we define perceived control as a person's perceived ability to influence their environment and future developments (cf.Burger, 1989).Most relevant to our current discussion is a person's perceived control in light of the violation of their group-based expectations: their perceived ability to influence the behaviour of the person who violates their expectations and, by extension, future developments within their group.But what feeds into group members' perceived control?Past research on group processes points to several (interrelated) factors that might play a role in fostering group members' perceived control (for an overview, see Figure 1).

Social identification as a source of perceived control
When individuals identify with a social group, they define themselves less in terms of their personal, idiosyncratic characteristics and more in terms of their group membership (Turner et al., 1987).Consequently, a group member's perceived control is no longer solely determined by their personal ability but also by their group's ability to influence its environment and pursue its goals successfully (i.e., by group efficacy, Mummendey et al., 1999).
Group-based control theory even argues that gaining a sense of collective control is what motivates some individuals to identify with groups (for a review, see Fritsche, 2022).Several studies support this notion by showing that individuals who lack personal control identify more strongly with groups, especially agentic ones (e.g., Hirsch et al., 2021;Stollberg et al., 2017).On the flip side, researchers have also found that individuals identify less with groups that appear ineffective in controlling criminal behaviour by in-group members, which might be attributed to such groups not providing group members with a sense of collective control (Pinto et al., 2016).
Though similar on a surface level, the sense of control fostered by identification with a group differs from the sense of certainty that group membership also provides.The latter is the focus of uncertainty identity theory (for a recent review, see Hogg, 2021) and refers to a group providing its members with an understanding of "who I am".The former, on the other hand, refers to a group member's perceptions of "what I can accomplish".Group membership not only directly informs this perception.It also grants group members access to control-related resources, one reason why social identification is claimed to have health benefits (for reviews of relevant research, see e.g., Cruwys et al., 2014;Muldoon et al., 2019).A key resource in this regard and an additional source of control is social support (for a discussion of the relation between identification and social support, see McKimmie et al., 2020).

Social (opinion) support as a source of perceived control
According to recent discussions on the benefits of social support, it can enhance perceived control by shaping receivers' understanding of their environment and fostering their perceived ability to manage the situation at hand (Zee et al., 2020). Festinger's (1950) work points to validation as a potential process underlying this control-enhancing function of social support.According to his work, individuals rely on others to validate their opinions -especially if these opinions are not grounded in physical reality (for a similar argument, see Postmes, Haslam, et al., 2005;Postmes, Spears, et al., 2005).Hence, receiving opinion support from others assures group members that their representation of the situation is consensual and valid.Lacking opinion support, however, places group members in a minority position, associated with low control (Guinote et al., 2006).Therefore, it appears reasonable to assume that receiving validation from others through social support is a vital ingredient to perceived control, as it should be linked to the expectation that group members will support each other in defending their shared understanding of what is right.Research on collective action provides initial evidence for perceived opinion support shaping reactions to violations of (group-related) expectations.It shows that when individuals assume that other ingroup members share their negative opinion on a new policy, their intentions to act against it are more pronounced (van Zomeren et al., 2004).However, not all group members might be able to expect the same amount of (opinion) support from others.This possibility points to a third source of perceived control previously studied by (social) psychologists -social power (or lack thereof).

Social power as a source of control
Social power implies asymmetric control over the outcomes of others or a group as a whole (Fiske, 1993).Within a group, social power is often tied to whether one occupies the role of a group leader.Thus, also the study of leadership and followership as a group phenomenon has direct implications for the study of perceived control in group contexts.
In their analysis of leadership as a group process, Hogg and colleagues (e.g., Hogg et al., 2012;Hogg, 2001) strongly emphasise leaders' role in shaping the values, norms, and goals their followers pursue.In other words, leaders can act as "entrepreneurs of identity" (Reicher & Hopkins, 2003) who shape their followers' social realities.This implies that leaders have and are granted more control over the group's future than regular group members.In line with this idea, several studies suggest that leaders (compared to regular group members) are, under certain circumstances, granted leeway to violate group members' expectations (for a review, see Abrams et al., 2018).From the perspective of a regular group member, this implies that perceived control over a fellow group members' behaviour should be higher than over a group leader's behaviour.
In summary, prior research points to three factors that should feed into a group member's perceived control and that, thus, need to be considered when seeking to understand how perceived control might shape a target group member's responses to violations of group-based expectations.One is the difference in social power between the observer and the violator.The second one is the amount of social opinion support and, thus, validation an observer receives from other group members.Finally, there are good reasons to assume that also an observer's identification with their group is linked to the extent of control this person perceives.We have looked at all three sources of control, which makes us confident in our conclusion that control needs to be added to a model of reactions to violations of group-based expectations.

Adding control to the equation
Our work was driven by prior research finding vastly different reactions to seemingly similar violations of group-based expectations.How can it be that these violations elicit both confrontational reactions, such as the derogation and exclusion of ingroup deviates (e.g., Eidelman et al., 2006;Marques, Abrams, & Serôdio, 2001;Schachter, 1951), and escape reactions such as leaving one's group (e.g., Sani, 2008)?We propose that perceived control is crucial to answering this question.
Prior research implies that violations of group-based expectations have a strong effect on confrontational reactions (Tata, 1996).Broadly considered, confrontational reactions communicate disapproval to a deviant individual -a person who has violated (group-based) expectations -for instance, by direct persuasion attempts or exclusion.This can serve several functions, including getting a deviate to change their behaviour (e.g., Brauer & Chekroun, 2005), reinforcing group consensus (e.g., Marques, Abrams, & Serôdio, 2001), as well as (re-)socialising and punishing deviant individuals (Pinto et al., 2010).This implies that a deviant's behaviour is seen as problematic, which aligns with the idea that perceived and experienced threat should foster confrontation.Confrontation, in turn, represents an attempt at resolving the threat.In addition, it implies that group members need to think that their reactions will have an impact on the deviant person and the group as a whole, suggesting that high perceived control might make confrontation more likely.Evidence that high control may foster confrontation stems from research on groupbased emotions.This work shows that when group members have the impression that public opinion supports (vs opposes) their ingroup's stance on a controversial issue (i.e., when perceived control is high), they experience anger towards an outgroup advocating the opposite stance.Moreover, they express intentions to confront members of this outgroup (Mackie et al., 2000).Our first key proposition builds on this work.We propose that Confrontation is a response to violations of group-based expectations fostered by experienced threat to one's social self and at least a certain amount of perceived control (Proposition 1). 2n the other hand, escape responses seem to reflect a different constellation of motivational factors.Social psychological research on group schisms (for a review, see Sani, 2008) maintains that group members should show escape when they have the impression that the behaviour they observe undermines their group's identity and essence.This perception correlates strongly with a composite score of dejection and agitation-related emotions (Sani, 2005), hinting at it reflecting appraisals of the situation as highly threatening.However, perceived or experienced threat has not played a role in explaining escape responses in prior research.The same holds true for perceived control, although there are at least two good reasons to assume that low perceived control should foster escape.First, perceiving that a group's essence has fundamentally changed implies having low control over the (group's) current and future situation.This is also mirrored in findings showing that perceived identity subversion is linked to lower identification with a group (e.g., Sani & Pugliese, 2008;Sani, 2005).Moreover, not seeing oneself as "in command" of a situation is related to fear and anxiety, two emotions linked to tendencies to move away from the situation (Frijda et al., 1989).Escaping from a group clearly reflects "moving away".Second, Sani (2008) argues that feeling marginalised and discriminated against in a group will increase the chances of members leaving their group.He attributes this effect to a perceived lack of voice, but given our review of sources of control above, this effect may also be attributable to low perceived control due to a lack of social (opinion) support.Therefore, we propose the following: Escape is a response to violations of group-based expectations fostered by experienced threat to one's social self and low perceived control (Proposition 2).
Taken together, prior research on different reactions to violations of groupbased expectations seems to agree that such violations are threatening for group members.However, it falls short of showing that group members actually feel threatened.Moreover, it largely neglects the role of perceived control in shaping group members' responses.We addressed these issues in our own work, which we will now review.Figure 2 depicts the model we derived from our research endeavours and which has implicitly guided our studies' design.

Are violations of group-based expectations threatening?
Our two key propositions assign a central role to threat as a factor motivating confrontation and escape in response to violations of group-based expectations.In our research program, we have used multiple approaches that, in combination, allow us to conclude whether threat actually plays such a motivating role.First, we have looked at two factors that should make a violation of group-based expectations more threatening: the (objective) extremity of the violation and the centrality of the violated norm.Second, we assessed perceived identity subversion (i.e., the perception that a behaviour undermines the group's essence, Sani & Reicher, 1998, 1999).Given the importance of a group's identity for defining group members' social selves (e. g., Turner et al., 1994), perceiving high identity subversion should reflect high perceived threat.Finally, we have also taken an approach inspired by research on the regulation of individual-level threats to show that violations of group-based expectations are experienced as threatening.Across these approaches, we used multiple types of violations tied to a group's identity, which was defined by a common goal, ostensibly shared attitudes, or (in cases where we relied on respondents' actual group membership) shared values.

Indirect evidence for the role of threat: Manipulating extremity and centrality
If experienced threat actually plays a role in shaping group members' reactions to violations of group-based expectations, one would expect reactions to increase in harshness with increasing violation extremity and increasing centrality of the violated expectations (for related evidence, see Frings et al., 2010;Marques, 1990;Mudd, 1968).Provided that extraneous circumstances are equal, objectively more extreme violations (i.e., larger discrepancies) should have a greater likelihood of being identified as potential threats.Moreover, group members should be more concerned about expectations that are central to group identity or group goal attainment.Hence, both high centrality and high extremity should make the violation subjectively more threatening, which calls for harsher reactions.This increased harshness in reactions is precisely what we found in a within-participants experiment (N = 66), manipulating both violation extremity and norm centrality (Ditrich & Sassenberg, 2016, Study 1) via vignettes set in a learning group context.For this study, we used six norms identified in a pilot study as being moderate to high in centrality for learning groups.For each of these norms, we designed two vignettes, one describing a rather mild and one describing an extreme violation.This allowed us to look at the effects of centrality and extremity independently.For instance, the vignette describing a mild violation of a norm relatively low in centrality ("decisions are only made after everyone has given input") read: [. ..] two group members are often very hesitant and do not participate in the discussion.Another group member knows as well as you do that these persons were not responsible for following up on the relevant lectures.Therefore, he/ she often closes these discussions even without all group members having had their say.
The vignette describing an extreme violation of the central norm "group members share all information with one another" read: [knowing that the professor bases his exams on questions from prior exams] . . .you decide as a group to collect all prior exams and work out the correct solutions together.[. ..]After your last meeting, it turns out that one member of your group has two more exams whose questions are different from the ones you worked on together.However, he/she has not passed these questions on to the rest of the group.
Our multi-level analyses revealed main effects of violation extremity and norm centrality on confrontation (measured as the willingness to continue working with the person described in the vignette) and escape (measured as the likelihood of preparing for the next exam with the same group).These main effects were qualified by significant Extremity x Centrality interactions.Violation extremity mainly affected confrontation and escape responses when the violated norm was of comparatively high or medium centrality but less so when it was of low centrality.Figure 3 displays the effect of norm centrality in the two extremity conditions.Taken together, these findings imply that situations which should subjectively be experienced as highly threatening evoked the strongest responses (for findings on extraneous circumstances that could make violations appear more threatening, see Marques, Abrams, & Serôdio, 2001).This constitutes initial evidence that threat plays a role in shaping group members' reactions to violations of group-based expectations.

Direct evidence for the role of perceived threat
A key variable in our studies was perceived identity subversion.It reflects group members' perception that a behaviour undermines their group's essence (Sani & Reicher, 1998, 1999) and may, thus, reflect perceived threat to the group's identity.This makes it well suited to provide further evidence of the role of threat in shaping group members' responses to violations of group-based expectations.In the study just described, we found that violation extremity and norm centrality also interacted in predicting identity subversion (operationalised with the item "This behaviour undermines our group's function").Extreme violations were judged more identity subverting than mild violations when the violated expectations were of high or medium centrality but not when they were of comparatively low centrality.Importantly, moderated mediation analyses revealed that perceived identity subversion mediated the effects of the Extremity x Centrality interaction on confrontation and escape intentions (see Table 1).Supporting our former conclusion, this suggests that extreme violations of central norms indeed led to harsher responses because they were perceived as particularly threatening.
In a follow-up study using attitude-based lab groups (N = 50), we focused on violations of expectations that were central to the group (Ditrich & Sassenberg, 2016, Study 2).We found similar effects of extremity on identity subversion and confrontation intentions.To ensure that we could hold centrality constant, we used a bogus group assignment procedure.We told participants that our study investigated group decision-making and that we would, thus, assign them to a group based on their responses to a questionnaire on honestly sharing information in cooperating groups.After a short delay, we informed them that they had been assigned to a group whose members had very high rates of agreement on the factors "honesty in interactions" (89%) and "fully sharing information" (83%).Thus, sharing information honestly was a highly central group norm for all participants.
Next, participants received their group task -solving a riddle (adapted from Steinel et al., 2010) -and 16 clues relevant to the task to deal with as they saw fit -share them honestly, withhold them from the others, or distort them before sharing them.Relevant to our extremity manipulation, some pieces of information were marked as more important for finding the solution than others.After making their own decisions, we showed participants how the other group members had decided.In the high extremity condition, one of them had withheld six important pieces of information and distorted two, while in the low extremity condition, this person had withheld six less important pieces of information and distorted two.Thus, in the high extremity condition, this person's behaviour represented a greater threat to group goal achievement than in the low extremity condition.Our measure of identity subversion (this time more elaborate; e.g., "The behaviour of Person C [the deviant target] undermines our group's identity") confirms this suspicion: Identity subversion again was significantly higher following extreme (vs mild) violations (see Table 2).
In line with the previous study, we also again found a main effect of extremity on confrontation (i.e., the extent to which participants wanted "to see one/some of the others leave this group").Identity subversion mediated this effect, B = 1.37,SE Boot = 0.40, CI 95% [0.71, 2.31], once more suggesting that perceived threat fosters confrontation.However, in this study, we did not find an effect of extremity on escape (i.e., participants wanting "to leave the group and not continue working with the others").Thus, the implications of this study are twofold.First, it further strengthens the claim that violations of group-based expectations represent threats, and that perceived threat might drive responses to such violations, as would be in line with our propositions.Second, it also implies that additional aggravating factorssuch as low perceived control -might be necessary for escape responses to emerge, as is in line with our second proposition.We will expand on this issue in the section on the influence of control below.Thus, these two studies already provide some evidence for the role of threat in the emergence of negative reactions to violations of group-based expectations.However, they also suffer from two limitations: We investigated reactions in the context of imaginary groups, and we focused on confrontation and escape intentions rather than on the respective behaviour.
A third study remedied both issues (Ditrich et al., 2017, Study 1) while yielding further evidence that threat might drive reactions to violations of group-based expectations.We analysed data from 126 students instructed to recall and describe an instance in which they belonged to a group and another group member violated their group-based expectations.Afterwards, they evaluated the behaviour, reported how they had behaved, and, after some exploratory measures, how severe they had found the behaviour they had recalled.
In line with the previously described study, we found a main effect of severity (z-stand.)on reported identity subversion (B = 0.57, SE = 0.17, CI 95% [0.25, 0.90]), assessed with items such as "I thought that this person's behaviour would fundamentally change our group".Severity also had a main effect on reported confrontational behaviours (e.g.; "I tried to exclude this person from our group."or "I would have liked to see this person leave our group"; B = 0.42, SE = 0.20, CI 95% [0.02, 0.82]).In short, participants reported having perceived more identity subversion and having engaged in stronger confrontation reactions when the situation had been more threatening.Moreover, severe violations also elicited higher escape responses, but only when the violation was accepted by others.We will return to this point when discussing the impact of perceived control.
Together, these three studies provide converging evidence that violations of group-based expectations are perceived as threatening group members' social selves and that perceived threat drives confrontational reactions (and possibly also escape), supporting our propositions.Given that this was especially the case for extreme violations of central expectations, we focused our subsequent research efforts on investigating the consequences of such violations.
One such effort provided further evidence for the relevance of threat, but was situated in a very different context: an opinion-based (faked) Facebook group (Ditrich & Sassenberg, 2017).We manipulated whether participants' group-based expectations were violated (yes vs no) and, if they were violated, whether the behaviour was accepted by other group members (yes vs no). 3  One-hundred sixty-six participants could join a Facebook group whose members shared their attitudes on introducing a student rail pass.Once they had selected a group, participants could follow its discussion in a timelapse, with new content added every 5 seconds.Among this content were posts and comments by the same focal group member (represented by a name and profile picture).This person followed the group's rules and supported the dominant (i.e., the participant's) opinion in the no-violation condition.In the violation conditions, the person violated participants' group-based expectations by not following the group's rules and consistently opposing the dominant opinion.
In the subsequent questionnaire, we assessed identity subversion (e.g., "This person's behaviour fundamentally changes what this group stands for.")and confrontation intentions ("If I were the group administrator, I would block this person."and "If I were the group administrator, I would send this person a warning.")with self-developed scales attuned to the study context.Moreover, we asked participants to evaluate other members regarding certain characteristics (e.g., considerate, helpful, bad, selfish; see also Eidelman et al., 2006) as a measure of derogation.This time, however, we assessed these variables with regard to the target group member and two other members who had not violated participants' expectations.
Perhaps unsurprisingly, the focal group member's behaviour elicited significantly more identity subversion (i.e., was perceived as more threatening) than the two other group members' behaviour in the violation conditions than in the no violation condition.Moreover, participants also reported stronger confrontation reactions (exclusion and derogation) towards the focal group member when this person had violated group-based expectations (see Table 3).Of note, as in the studies discussed above, identity subversion mediated the effect of the presence (vs absence) of a violation on exclusionary responses, B = 0.21, SE Boot = 0.05, CI 95% [0.13, 0.32].
Thus, even in the rather artificial context of faked Facebook groups, violations of group-based expectations appear to be perceived as threatening, which motivates confrontational reactions.This finding strengthens the basis for our claim that violations of group-based expectations are threats and that 3 Acceptance of the behaviour did not have an effect in the present study.
these threats motivate group members' reactions, especially in the case of confrontational reactions (see Proposition 1).However, due to the nature of the groups used in these experiments, it is unclear whether the violations threatened participants' social selves.
Thus far, we have discussed studies looking at opinion-based groups or groups primarily tied together by a common goal.Now, we will turn to two studies that used participants' existing social identity as students of their university, giving violations of group-based expectations the potential to threaten group members' actual social selves.These studies (Ditrich et al., 2017, Study 2;Ditrich et al., 2019, Study 1) used the same basic paradigm, which we will call the "Facebook paradigm" for the remainder of this article.After making participants' identity as a member of their university salient and assessing university identification, we showed them a screenshot ostensibly taken from their university's Facebook page.This screenshot depicted a (faked) post by a senator of participants' liberal university.In the violation condition, the senator made xenophobic statements and called on readers to take part in a demonstration of a nationalist group.In the no-violation condition, the senator advocated for tolerance and called on readers to participate in a peaceful demonstration (content varied slightly between studies).
After reading the post, participants reported on perceived identity subversion (e.g., "This senator's behaviour undermines the values the university stands for").We used two measures to operationalise confrontational responses, asking participants what they would have done had they actually seen the post on their university's Facebook page: participants' intentions to communicate disapproval to the senator (e.g., "I would contradict the senator in a comment") and their endorsement of excluding the senator from the page (e.g., "How likely will you ask the group administrator to exclude this person from the group?").Escape intentions were assessed with the item "How likely will you leave the Facebook group?".
In the first study (N = 499; Ditrich et al., 2017, Study 2), we found significant main effects of our violation manipulation on identity subversion, both forms of confrontational responses, and escape responses (see Table 4).
The second study largely replicated this result pattern (N = 147; Ditrich et al., 2019, Study 1), but our violation manipulation only enhanced escape when respondents did not receive opinion support.We will return to this finding when looking at the direct evidence for the role of control.
Taken together, the studies reviewed so far provide consistent evidence that violations of group-based expectations are perceived as subverting the group's identity.Given that a group's identity is closely tied to its members' social selves (e.g., Tajfel & Turner, 1979;Turner et al., 1994), this can be seen as an indicator of such violations being perceived as threatening.Thus, the reviewed findings provide evidence in line with our first proposition.Moreover, they also provide evidence for our second proposition.Naturally, this evidence is to be regarded as somewhat tentative, given that perceiving identity subversion is not a direct indicator of group members experiencing threat.We have, however, conducted three studies that more unambiguously speak to the role of experienced threat in the context of violations of group-based expectations.

Direct evidence for the role of experienced threat
To study experienced threat, we used an approach inspired by research on the regulation of individual-level threats.After all, if violations of groupbased expectations are experienced as threatening, they should evoke the same regulatory responses as individual-level threats.These responses have been linked to three regulatory systems -the Behavioural Approach System (BAS), the Behavioural Inhibition System (BIS), and the Fight-Flight-Freeze System (FFFS; e.g., Corr et al., 2013;Gray & McNaughton, 2003).If violations of group-based expectations are experienced as threatening, they should activate the same systems as individual-level threats, namely BIS and FFFS.
The FFFS responds to close, immediate dangers and threats (McNaughton & Corr, 2004).Its activation fosters reactions to avoid threats and reach safety through flight, freeze, or fight responses.The latter combine defensive and confrontational, approach-driven elements (Harmon-Jones et al., 2013).This double-faced nature of fight responses is also expressed in their close association with anger, which has been claimed to be a more common response to FFFS-activation than overtly attacking the source of the threat (e.g., Corr et al., 2013).Thus, while not classically treated as such in the emotion literature, anger can be seen as an indicator of FFFS-activation and, by extension, of individuals experiencing threat.
The BIS, on the other hand, responds to potential dangers (e.g., Gray & McNaughton, 2003;McNaughton & Corr, 2004).Its activation leads to high vigilance, high arousal, and the inhibition of ongoing behaviour (Corr et al., 2013;Gray & McNaughton, 2003).On an affective level, BIS-activation is associated with anxiety, inhibition, and worry (Corr & Cooper, 2016).Thus, if violations of group-based expectations are experienced as threatening, they should evoke FFFS-activation, specifically fight tendencies (indicated by anger) and BIS-activation (indicated by anxiety).This is precisely what we found in three studies using very different violations of group-based expectations and different group settings.In our discussion of these studies, we will focus specifically on confrontational responses, as those are associated with anger (e.g., Frijda et al., 1989;Mackie et al., 2000;Sassenberg & Hansen, 2007) and might be used to overcome the aversive state of BIS activation (Jonas et al., 2014).
In the first of these studies (Ditrich et al., 2022, Study 1), we asked participants to read a text about an altercation during a lecture.This was either sparked by a professor asking a female student to remove her headscarf (violation condition) or by the professor asking the student to put away her cell phone (no-violation condition). 4Participants (N = 188) were instructed to respond to all items as though they had experienced the professor's behaviour themselves.We assessed activation of the threat regulatory systems via participants' emotions (FFFS-associated fight: outraged, argumentative, confrontational; BIS-activation: inhibited, unsettled, uncertain).Moreover, we once more operationalised confrontational responses as intentions to communicate disapproval to the professor (e.g., "If I had directly experienced Professor W.'s behaviour, I would talk to her about her arguments after the lecture.")and exclusion intentions (i.e., intentions to rid the professor of her post; e.g., "If I had directly experienced Professor W.'s behaviour, I would try to get disciplinary action taken against her.").However, we cannot be certain whether our participants regarded the professor as a member of their ingroup.Therefore, we returned to our Facebook paradigm and, thus, a group that provided participants with an actual social identity in a second study.
In this study (Ditrich et al., 2022, Study 2) we again showed participants (N = 336) a xenophobic (violation condition) or a tolerance-endorsing (no violation condition) post. 5We again assessed activation in the FFFS and BIS via (slightly extended) emotion self-reports.After some additional measures, we assessed confrontational responses as before (e.g., "How likely is it that you will contradict this post in a comment?", "How likely is it that you will ask the administrator to delete the post?").
In a final step towards providing direct evidence that violations of groupbased expectations are threats to group members' social selves, we turned to outgroup criticism as a violation of group-based expectations (Ditrich et al., 2022, Study 3. In doing so, we capitalised on a specific socio-political situation: many Germans study in Austria.This has led to Austrian universities debating whether to limit the number of international students they accept.Therefore, in our study (N = 156), we presented one of two speeches ostensibly given by a member of the Austrian government to our German participants studying in Austria.
In one condition, the politician violated participants' expectations (of being treated with respect) by criticising international students.He implied that they were to blame for a lack of places available to Austrian students.Moreover, the speech contained derogatory terms mainly used to describe German students at Austrian universities (i.e., "NC-refugees") and the politician advocated the introduction of tuition fees for international students.In the no-violation condition, the politician's speech focused on a lack of practical education at university, and he described measures he would support to remedy this issue.After participants had read the speech, we assessed the activation of the FFFS and the BIS as in the previous two studies.
Confrontation intentions likewise were again operationalised via communication (e.g., "I would really like to give this politician a piece of my mind.")and exclusion intentions (e.g., "If I were asked, I would immediately sign a petition demanding the politician's dismissal from office.").
In combination, the three studies strongly support the proposition that violations of group-based expectations are experienced as threatening.In all three studies, we found greater activation of FFFS-associated fight-tendencies and of the BIS in the violation condition compared to the no-violation condition (see Table 5).Thus, violations of group-based expectations elicited similar threat-regulatory responses as individual-level threats (for a summary, see Jonas et al., 2014).In line with this research, we also found no consistent effect on activation of the BAS (indicated by positive emotions like feeling strong, determined, or full of energy), which only responds to threats after a delay, if at all (Jonas et al., 2014).In addition, we also found that FFFSassociated fight-tendencies mediated the effect facing a violation of groupbased expectations had on communication and exclusion intentions (see Figure 4).BIS-activation, however, did not consistently mediate the effects of facing a violation of group-based expectations on group members' reactions.In summary, while there were slight inconsistencies between the studies, their overall results align with our first proposition that violations of groupbased expectations lead to confrontational reactions because they are threats (for a similar argument, see e.g., Abrams et al., 2000;Marques, Abrams, Páez, et al., 2001).Moreover, they extend prior findings on the role of threat in shaping confrontational reactions (e.g., Fousiani et al., 2019;Frings et al., 2012) by measuring perceived identity subversion (as a possible indicator of perceived threat) and by using a fine-grained measure of experienced threat.Thereby, they, for the first time, show that violations of group-based expectations are experienced as threatening.
We find highly similar result patterns across different types of groups (learning groups, opinion-based lab groups, Facebook groups, lecture, university members, German students in Austria) and group constellations (ambiguous, ingroup, intergroup).Additionally, our studies span different forms of violations of group-based expectations (violations of ingroup norms, outgroup critique).This suggests both a broad generalisability of our findings and that the processes underlying the emergence of group members' reactions are universal, even if the reactions might differ in intensity contingent on the type of norm being violated (Brambilla et al., 2013).Our research, thus, provides substantial empirical evidence that (a) violations of group-based expectations are threatening to group members' social selves and (b) experienced threat is a key motivational factor underlying reactions to group-based expectancy violations.This has prompted us to propose this pathway in our model, as visualised in the solid white arrows (see Figure 2).However, our work also suggests that threat might not be the only factor shaping group members' reactions to violations of group-based expectations.Instead, they imply that also perceived control needs to be considered when trying to understand how group members react to such violations.Social psychologists have already investigated several potential sources of perceived control (over the violator's behaviour and the group's future; e.g., Muldoon et al., 2019;Tajfel & Turner, 1979;van Zomeren et al., 2008;Wright, 2003).Moreover, control is an important resource for coping with threats (e.g., Blascovich & Tomaka, 1996).Nonetheless, it has received scant attention in prior work on reactions to violations of groupbased expectations, as have resources in general (for an exception, see Frings et al., 2012).Our own findings suggest that control is a key resource group members can draw upon when deciding how to react.Thus, we argue that its omission from previous theorising has significantly limited our understanding of when and how group members react to violations of group-based expectations.Consequently, we are the first to explicitly include perceived control in their model of reactions to violations of group-based expectations, a decision we deem justified based on the findings reviewed in the following sections.

Does perceived control influence reactions?
Our second proposition implies that perceived control plays a role particularly in the emergence of escape responses.Group contexts offer multiple sources of perceived control and, thus, multiple angles from which to approach this proposition.First, one can look at the impact of social power, as leaders' behaviour might appear less easy to control than fellow group members' behaviour.Second, one can look at group members' relation to their group, which is associated with a sense of group-based control (e.g., Fritsche, 2022;Greenaway et al., 2015).Finally, the extent of social (opinion) support group members receive from others likely also feeds into their control perceptions (e.g., van Zomeren et al., 2004;Zee et al., 2020).Given that we have evidence speaking to all three sources of control, we deem ourselves well-equipped to draw conclusions on the role of control in shaping group members' reactions to violations of group-based expectations.

Indirect evidence for the role of control I: Responses to norm-violators with different power
The first piece of evidence that perceived control may influence group members' reactions to violations of group-based expectations stems from comparisons between various studies we conducted.The key difference between these studies regarding perceived control is the role of the person who violates group members' expectations.In one of our first studies with attitude-based lab groups already discussed above (Ditrich & Sassenberg, 2016, Study 2), the violation consisted of a fellow group member withholding or distorting information relevant to group success.Whether the person's behaviour constituted an extreme (i.e., withholding or distorting important information) or mild violation (i.e., withholding or distorting less important information) did not affect escape intentions, measured as participants' (un) willingness to continue working with their group in further tasks.At first sight, this implies that even extreme violations do not drive escape.
However, a second experiment with attitude-based lab groups (Ditrich & Sassenberg, 2016, Study 3) questions this conclusion.In this study (N = 124), the person who violated participants' expectations was a group leader ostensibly highly prototypical for the group.After we had assigned them to a group, participants received their group goal: collectively unscrambling enough words to be eligible for a raffle.To make the group experience more vivid, participants received several messages from their group leader (at the beginning of the task, in between working on two sets of scrambled words, and after the task), who addressed the group's members via the nicknames they had chosen for themselves.The leader's final message contained our violation manipulation.In the no-violation condition, the leader (who had a male-sounding nickname) stated that he would distribute the tickets equally among the group's members.In the violation condition, he stated that he would distribute the tickets based on the equity principle, thereby violating the equality norm around which the group had been formed.Participants could reply to this message.Subsequently, they indicated to what extent the leaders' behaviour subverted the group's identity (e.g., "The group speaker's behaviour undermines our group's true nature").Then, participants received messages ostensibly written by the remaining two group members (see section on opinion support for details).Finally, we assessed actual confrontation and escape behaviour.Participants were led to believe that the following part of the study could be completed alone or in a group.They were asked to indicate how much they would like to work with the leader (represented by his nickname) for the upcoming task to assess confrontation.To assess escape, we combined participants' willingness to work alone with their willingness to continue working with the two remaining group members so that high ratings would reflect a strong escape response.In line with our prior findings, confrontational responses were more pronounced in the violation conditions than in the no-violation condition (see Table 6), mediated via greater identity subversion (B = 0.43, SE = 0.12, CI 95% [0.21, 0.69]).At first sight, this finding contradicts prior research showing that (under certain circumstances) deviant leaders are treated more leniently than deviant group members (for a review, see Abrams et al., 2018).Notably, this does not mean that "leaders are completely free to deviate, they merely meet less resistance than do deviant non-leaders" (Abrams et al., 2018, p. 37).Thus, using group leaders who violate group members' expectations represents a particularly strict test of our proposition that (threatening) violations of group-based expectations are met with confrontation.
More importantly, however, participants also showed stronger escape responses when the leader had violated their expectations, especially when other group members accepted the leader's behaviour (i.e., when opinion support and consequently perceived control was low; for a discussion, see below).We attribute the significant increase in these tendencies following violations by a group leader -but not following violations by a group member in the prior study -to the two types of violators differing in social power.Leaders have more power than regular members (e.g., Hogg et al., 2012).Consequently, group members should perceive less control over the group's future when a leader violates their expectations than when a fellow group member does the same.Nonetheless, we concede that comparing the findings from these two studies alone is only very indirect evidence that perceived control plays a role in shaping group members' reactions to violations of group-based expectations.To make more definite claims in this regard, it would be necessary to directly compare the consequences of violations committed by group leaders vs regular members within one study.However, our research program also provides evidence for perceived control playing a role in shaping group members' reactions from a different angle.

Indirect evidence for the role of control II: Violations and identification
Specifically, we have looked at how violations of group-based expectations affect group members' relationships with their group.Violations of groupbased expectations have been argued to have the potential to fundamentally change a group's identity (Abrams et al., 2000;Marques, Abrams, & Serôdio, 2001).If a group actually does change (or is perceived to have changed), attempts at controlling the group's future have failed.This can signal low control and reduces the group's ability to provide its members with a valued social identity.Thus, (having had) low control may express itself through reduced identification with a group.
In fact, a closer inspection of the items we used to assess identity subversion reveals some ambiguity in this regard.For instance, an item like "The University of X that emerges as a consequence of the senator's behaviour contradicts its history, values, and principles."can reflect potential changes that might still be controlled.However, it may also reflect the perception that actual changes have already taken place and, thus, can no longer be controlled.This means that reports of high identity subversion could reflect a high sense of threat and low perceived control at the same time.This potential ambiguity makes drawing firm conclusions on perceived control that can inform model building difficult.This is easier based on indicators more unambiguously reflecting group members' relationship to their group, like identification with this group.
We have addressed identification with a group in two studies (Ditrich et al., 2017, Study 2;Ditrich et al., 2019, Study 1) using our Facebook paradigm.To account for baseline differences in identification, we measured this variable both before and after presenting participants with a violation (vs no violation) of their group-based expectation that members of their university would have liberal, egalitarian attitudes.At both points in time, we used items from established scales (e.g., Leach et al., 2008).While the items used before the manipulation were rather general, the items used after the manipulation directly targeted participants' momentary identification with their university (e.g., "Being a member of the University of X is important to me at the moment").
In both studies (N = 499 and N = 147), we found that facing a violation reduced participants' identification with their university (controlling for premanipulation identification as a covariate; see Table 4).Mediation analyses with identity subversion and post-manipulation identification as parallel mediators further revealed that both were significantly linked to escape responses (see Table 7 for indirect effects and Table 8 for a qualification of the violation main effect on escape in the second study).In other words, the less group members identified with their university after a violation, the more likely they were to express escape intentions.
We consider the reduced identification with a group we found in these studies as a hint that group members perceived low control when thinking about the violation.Prior research suggests that individuals who experience low personal control may increase their identification with groups that provide them with a sense of collective control (e.g., Fritsche, 2022).However, this also means that lowered identification with a group might indicate that it no longer provides individuals with a sense of collective control, let alone a positive sense of self.Thus, in combination with the previous findings on the consequences of violations by group leaders vs members, the present findings further strengthen our second proposition that escape is a response to threatening violations of group-based expectations that becomes more likely when perceived control is low.This also underscores the necessity to include perceived control in models of group members' reactions to violations of group-based expectations.At the same time, it clearly points out that control-related moderators of the violationreaction relationship need to be investigated.We did so by focusing on opinion support.

Direct evidence for the role of control: The impact of opinion support
Opinion support constitutes a source of perceived control in group contexts, meaning that studies investigating its consequences allow us to approach perceived control from another, more direct angle.We have conducted research on the consequences of both lacking and receiving opinion support.In both cases, this was achieved by manipulating others' reactions to a target person's behaviour.Specifically, participants who received opinion support faced other members who (re-)affirmed group-based expectations.This should validate their perceptions of the group, which should strengthen perceived control.At the same time, it should also reduce the impact the person violating participants' expectations is perceived to have on the group by signalling that other members might help defend the status quo.Participants lacking opinion support, on the other hand, faced other members who accepted a target person's expectancy-violating behaviour.This placed our participants in a minority position within their group, which is associated with low perceived control (Guinote et al., 2006).According to our second proposition that escape responses should be especially likely when perceived control is low, receiving opinion support should, thus, weaken escape responses, whereas lacking opinion support should foster them.
In a first set of studies shedding light on the validity of this assumption, we did not measure perceived control but rather the extent of control ascribed to the person violating expectations.These two variables should be inversely related: The more control a violator has, the less control observers perceive  and vice versa.Thus, receiving opinion support should at the same time strengthen observers' perceived control and weaken the violator.Consequently, escape responses should be less pronounced when receiving opinion support.In line with this train of thought, three studies indeed showed that participants reported lower escape responses when receiving opinion support from others.Within these studies, we manipulated opinion support by showing (high support) or not showing (low support) other group members' replies to a post that violated (vs did not violate) group members' group-based expectations.In the violation conditions, these replies reaffirmed the norm violated by the group leader (e.g., "I don't believe it!That's not what we are like here at the Uni of X!").In the no-violation conditions, the replies endorsed the leader's behaviour (e.g., "I am very happy about this!This is a way for us to show how we are really like here at the Uni of X!"), which led to a full 2 (violation: yes vs no) by 2 (opinion support: yes vs no) design.
In the first study (N = 147; Ditrich et al., 2019, Study 1), we investigated whether opinion support reduces escape responses and helps protect the group's identity.We only found the former.In the violation condition, receiving opinion support reduced escape intentions (see Table 8).However, opinion support did not seem to protect the group's identity: For both identity subversion and post-manipulation identification, we only found the violation main effect described above.Hence, opinion support does not seem to affect the evaluation of the leader's behaviour (as reflected in perceived identity subversion).Moreover, it does not help maintain a close relationship with the group.In line with our argument that opinion support should affect perceived control, two further studies instead revealed that opinion support influences a violator's perceived control over a group.
In these studies, we focused on perceived leader effectiveness.This variable features prominently in research on leader performance and behaviour (e.g., Giessner & van Knippenberg, 2008;Giessner et al., 2009), reflecting the perception that a leader can effectively influence their followers.Thus, assessing perceived leader effectiveness tells us how much control over a group's identity and future the leader is perceived to have -which should be little if they face open disapproval of their behaviour due to observers receiving opinion support.
The first study addressing effectiveness (N = 159; Ditrich et al., 2019, Study 2) showed that this might indeed be the case.Using the same paradigm as the previous study, we found that opinion support (i.e., reaffirmation of the violated norm) interacted with our violation manipulation in affecting perceived leader effectiveness, assessed with items like "This senator probably is very influential" or "This senator can enforce changes.":Leaders were perceived as less effective when they violated group-based expectations, especially when others opposed their behaviour (i.e., when participants received high opinion support).Replicating our previous findings, opinion support also buffered the effect of expectancy-violating behaviour on escape intentions.Escape intentions increased when a group leader violated groupbased expectations but less so when others opposed the violation by reaffirming the group's norms (see Table 8).Hence, when others opposed a leader's behaviour, the leader was ascribed less control over the group, and group members were less inclined to escape from their groups.This supports our interpretation of opinion support as a source of control.Moreover, it again points to perceived control being a relevant factor in shaping group members' reactions to violations of group-based expectations.
The second study on effectiveness (N = 477; Ditrich et al., 2019, Study 3) replicated these findings using a different group context and a more elaborate measure of escape responses.Thereby, it puts our conclusions on the relevance of perceived control for understanding reactions to violations of group-based expectations on an even firmer basis.In this study, we moved away from participants' existing social identities and asked them to imagine wanting to join a sports team.They could choose between two teams whose mottos made a group-norm salient: being highly competitive or being laidback.Subsequently, participants were asked to imagine that they had now been members of their chosen team for a while and were shown a screenshot from their team's WhatsApp chat.This screenshot contained our violation manipulation (within the team captain's message) and our manipulation of opinion support (within others' replies to the team captain).Depending on the team they had chosen to join, the content of the violation manipulation differed. 6For instance, participants who had chosen the laid-back team and were in the violation condition received a message from their captain asking for more effort and announcing that next week's after-training pub visit would be replaced with an extra training session.In the no-violation condition, the team captain stated that the team should continue to have fun and proposed going to the pub after that day's practice.The opinion support manipulation was identical across both teams: Participants either saw others' reactions to the team captain's message or did not see any responses.
The two experimental factors again interacted in predicting perceived leader effectiveness (assessed as before).Perceived leader effectiveness suffered when the leader violated group-based expectations.This effect was exacerbated by other group members opposing the leader's behaviour.Escape intentions were likewise affected by a Violation x Support interaction: Participants more strongly considered leaving their team (e.g., "I would try to back out of this team.","I don't think I would concern myself with this team any longer.")when the leader had violated their expectations, but others' affirmations of the team's norm mitigated this effect (see Table 8).Thus, once again, receiving opinion support influenced group members' reactions to a violation of group-based expectations.Strengthening our conclusion that these effects of opinion support are due to its link to control, we found reverse effects on escape intentions when respondents lacked opinion support.
In the guided recall study discussed at the beginning of our research synthesis (Ditrich et al., 2017, Study 1), simple slopes analyses revealed that violation extremity was only linked to escape responses when opinion support was low (i.e., when other members had accepted the violator's behaviour).When opinion support was high (i.e., when others had not accepted the violator's behaviour), violation severity was unrelated to escape responses.Thus, lacking support and, consequently, control enhanced the effect of violations and drove individuals away from their group.
This was also the case in the experimental study we conducted with opinion-based lab groups that we have already discussed as an example of the effects of violations committed by group leaders (Ditrich & Sassenberg, 2016, Study 3).Going beyond the guided recall study, this study provides direct, unambiguous evidence that low opinion support in the light of behaviour violating group-based expectations reduces perceived control.To manipulate opinion support, we used messages ostensibly written in reply to the leader's decision on how to distribute the lottery tickets.When he decided to distribute the tickets equally -as was in line with the group's norm and, thus, participants' group-based expectations -the two remaining members agreed with the decision.Importantly, when the leader had decided to instead distribute the tickets based on the equity principle, their messages communicated either approval (low opinion support) or disapproval (high opinion support) of the leader's decision.As in the guided recall study, others' approval of the leader's violation exacerbated its effect on escape responses (see Table 6).Crucially, the effect of other's approval (vs disapproval) of a leader's expectancy-violating behaviour was mediated by perceived control (B = 0.11, SE = 0.07, CI 95% [0.01, 0.30]), which we assessed with items such as "I have the impression that I have an impact on this group".Specifically, participants perceived less control when others had accepted the leader's violation of their group's norm (i.e., when opinion support was low) than when they had opposed his behaviour.Perceived control, in turn, was marginally related to escape responses.Thus, the less control participants experienced, the more likely they were to favour escaping from the group and working alone on further tasks.Hence, this study directly suggests that perceived control shapes group members' reactions to violations of group-based expectations: When perceived control was diminished, escape responses became more likely.
In combination, the studies reviewed in the last three sections consistently suggest that threat is not the only factor determining when and how group members react to violations of group-based expectations.Reactions differed depending on whether violations were committed by a regular group member or a group leader.Moreover, violations fostered escape responses via reducing identification with the group.Finally, we found that opinion support reduced escape responses, whereas a lack of support enhanced them.All these aspects differentially affecting escape responses have one thing in common: they are sources of perceived control.Thereby, our findings support our second proposition.Notably, we did not only approach the concept of perceived control from multiple angles.We also found evidence for its effects across different types of groups, ranging from imagined groups over opinion-based lab groups to real groups that mattered to our participants and their social identities.This went hand in hand with us demonstrating the relevance of perceived control for shaping reactions to violations of very different group-based expectations (e.g., adherence to an equality norm or the expression of liberal attitudes).Finally, our findings suggest that perceived control plays a role in shaping response intentions as well as actual behaviour.Therefore, the addition of perceived control to our model of reactions to violations of group-based expectations rests on a solid empirical basis.

Outlook: Extending the model in future research
Overall, our model suggests that two factors are relevant to understanding and explaining group members' reactions to violations of their group-based expectations: experienced threat and perceived control.In devising the model, we drew from prior theorising (e.g., Festinger, 1950;Marques, Abrams, & Serôdio, 2001;Sani & Pugliese, 2008) and a substantial body of empirical findings.By explaining both confrontation and escape responses, our model integrates two, to date, largely separated areas of research.Moreover, it gains explanatory power above prior models by casting perceived control as a central determinant of group members' reactions to violations of group-based expectations.Nonetheless, we see several lines along which the model can be extended and its boundaries tested to obtain an even more encompassing understanding of group members' reactions to violations of group-based expectations.These concern the predictors in our model as well as aspects of the social system in which the violation occurs.

A closer look at the predictors
Throughout this article, we have proposed that violations (and the resulting experience of threat) in combination with perceived control determine group members' reactions.Concerning control, this conclusion rests on studies in which we manipulated or measured sources and correlates of perceived control.While we have evidence that our manipulation of opinion support affected perceived control, future studies should include further manipulation checks or direct manipulations of control to put our conclusion to an even stricter test.Including such checks and manipulations might also prove fruitful in determining which role control plays in shaping confrontational responses.Currently, our data suggest that perceived control plays a role for escape responses (supporting Proposition 2) but could be less important for confrontational responses (partly questioning Proposition 1).Initial evidence that high perceived control might foster confrontational responses (as implied by Proposition 1) stems from prior work by Frings et al. (2010).It suggests that having normative support in the form of knowing that other members will contact those who violate expectations (i.e., high perceived control) promotes intentions to engage in confrontation.Our findings on the motivational systems involved in the emergence of confrontational responses point in a similar direction.They showed that FFFS-associated fight-tendencies, but not BIS activation mediated the effect of violations of group-based expectations on confrontational reactions (Ditrich et al., 2022).Fight-tendencies, in turn, are associated with anger and approach motivation (Harmon-Jones et al., 2013).Both are linked to a sense of agency and, thus, perceived control.Naturally, this is only indirect evidence that awaits further investigation.
Regarding violations, we see two potential avenues towards extending the model.One concerns the question of what happens when group members perceive themselves to violate a group's norms.Prior research suggests that not living up to one's own group-based expectations is associated with negative affect and lower well-being (Sassenberg et al., 2011), which may be a first hint that threat also plays a role here.A related set of studies suggests that, when group norms are hard to live up to (i.e., when perceived control should be rather low), this can likewise lead to negative affect and lower well-being, especially among those only weakly identified with a group (Scholl et al., 2019).Thus, experienced threat and perceived control may also influence responses to own violations of group-based expectations.
The second avenue towards extending the model focuses on differentiating the type of expectation being violated.In our own work, we used a broad array of expectations, some more closely tied to group goal attainment and others more closely tied to group values.Notably, all studies provided converging evidence that threat and perceived control play a role in shaping group members' reactions.Nonetheless, some of our findings also suggest that group members might not care equally about all violations of their expectations.In fact, they appear to care primarily about violations of expectations that are central to their group's identity.Prior research suggests that issues tied to a group's morality should be particularly likely to form the basis of such highly central expectations (e.g., Brambilla et al., 2013).In intra-group settings, we would not expect qualitatively different reactions but only more intense ones to violations of morality-related expectations than other central expectations.In intergroup settings, however, moral violations have been argued to have the potential to drive collective action (e.g., Agostini & van Zomeren, 2021), a reaction so far unaccounted for by our model.Thus, comparing violations of moral and non-moral expectations appears to be a highly fruitful approach towards extending our model.On the one hand, it can help us better understand intragroup processes.On the other, it can help reconcile our model with models of collective action.This appears to be a worthwhile integrative endeavour that can help us understand not only how and when groups change but also how and when they attempt to change the social world around them.It does, however, also require considering contextual factors.

A closer look at the social system
A first critical contextual factor that might shape responses to violations of group-based expectations is the type of social system (intra-vs intergroup) in which the violation occurs.Currently, our model is mainly based on research focusing on intra-group situations, that is, situations in which the person violating expectations belonged to the same group as the observer.In two studies, we showed that similar motivational processes shape responses when the person violating expectations is an outgroup member or the categorisation of this person is somewhat ambiguous.That is, there is initial evidence that the model also applies in intergroup settings.What we have not looked at so far, however, are situations in which groups are nested, that is, when a violation occurs in a (superordinate) group that consists of several subgroups (e.g., factions in parties, regions within countries).Research on group schisms implies that such situations may be the ideal breeding ground for the formation of breakaway groups (Sani, 2008).However, they might also motivate subgroup members to engage in collective action on behalf of their own group, a reaction that our model does not yet account for.Thus, studying nested groups not only allows testing the boundaries of our model but also allows integrating and explaining additional reactions.
We have already taken a first step in this direction with a correlational study in the context of Brexit (Ditrich et al., 2021).In July 2017, 202 Scots reported on how much the Brexit referendum had violated their expectations (e.g., "In general, I think that the EU referendum was carried out without the principles of the UK taken into consideration.").Moreover, they reported on escape (e.g., "I would prefer to live in a country that is part of the EU.") and collective action intentions (here: promoting Scottish independence; e.g., by volunteering time to this cause).In line with our prior findings, identity subversion mediated the relation between perceived expectancy violations and escape intentions (β = 0.37, SE = 0.07, CI 95% [0.24, 0.49]).However, the link between expectancy violations and collective action was mediated by disidentification from UK-citizens (β = 0.19, SE = 0.04, CI 95% [0.12, 0.28]).That is, those who perceived the referendum as having violated their expectations did not like belonging to the group of UK citizens (e.g., "I feel alienated from the citizens of the United Kingdom.").The less participants liked being a UK citizen, the more inclined they were to engage in collective action on behalf of their subgroup.Hence, our model is a good starting point for explaining individuals' reactions to violations of their group-based expectations.However, its explanatory power might be further enhanced by including additional reactions and their precursors.
Further knowledge could also be gained by more closely investigating characteristics of the social system in which violations of group-related expectations occur.Our model thus far builds mainly on studies of groupnorm violations.Such norms are closely tied to a group's identity (e.g., Turner et al., 1994) and, as such, strongly define the social system in which group members act.Consequently, in intragroup settings, violations of group norms can fundamentally change the social system (for a similar argument, see, e.g., Marques, Abrams, Páez, et al., 2001).This likely contributes to making such violations especially threatening.However, in reality, groups are not independent of other groups, meaning that the social system in which group members act is much broader.
Social identity theory points to three characteristics of such social systems that shape group members' behaviour in light of instances that negatively affect their group or its status: the legitimacy and stability of the situation and the permeability of group boundaries (Tajfel & Turner, 1979).While the first aspect seems closely related to judging the situation as violating expectations, the second aspect is tied to control (Wright, 2003).Thus, our model already accounts for these two characteristics.However, its current version neglects the influence of permeability.Throughout our studies, we have investigated groups whose boundaries were permeable -escape was possible.In most cases, this could even be accomplished at a relatively low cost (e.g., in case of opinion-based lab groups or the university's Facebook group).A notable exception here is our study on reactions to the Brexit referendum as, in this case, escape would be related to considerable costs.Interestingly, high costs do not seem to deter escape intentions following violations of group-based expectations.
What might deter such intentions, however, is the second facet of permeability: the availability of alternative groups one can join.As we did not investigate where group members intend to go after escaping from their current group, we can only speculate on how the permeability of group boundaries and the availability of alternative groups may be included in our model.We think that both factors could act as moderators in the model that primarily affect escape responses, such that when permeability is high and alternative groups are available, escape should become more likely.Hence, further research on the impact of socio-structural factors on group members' reactions to violations of their group-based expectations is necessary and will contribute to integrating research on inter-and intra-group processes further.
A final aspect that deserves further investigation and may need to be factored into our model is the broader social systems' response to violations of group-based expectations.So far, our work has only investigated other group members' responses.We found these to strengthen or reduce escape tendencies, depending on whether others approve or oppose a violation.However, it appears likely that reactions by the broader social system will also influence group members' reactions.This can include reactions by outgroups, which may foster embarrassment (Chekroun & Nugier, 2011), but also by non-partisan authorities.Suppose such authorities condemn a behaviour perceived as violating group-based expectations.This gives group members a reason to strive to prevent the violator from tarnishing the group's and their own image.The former can be accomplished through confrontational responses, especially exclusion (e.g., Eidelman et al., 2006;Hutchison et al., 2008).The latter may be accomplished through escape (e.g., Eidelman & Biernat, 2003).
But now suppose that authorities legitimise a behaviour perceived as violating group-based expectations.This may give group members pause, as it fundamentally changes the basis on which they originally made the judgement that the behaviour violates expectations.That is, approval by authorities changes the "ought"-part of the is-ought comparison that has led to the conclusion that a behaviour is a violation.Given that judging a behaviour as a violation justifies confronting the violator, such approvals, thus, should reduce confrontational responses.Provided that the authorities are actually non-partisan and belong to a superordinate group, their approval may even reduce escape responses.After all, there is no alternative group (imagined or real) one can turn to.Just imagine what would have happened if the United Kingdom's parliament had voted on impeaching Boris Johnson over "Partygate" and he would not have been impeached.This might not have made people happy, but it may also have signalled that it is okay for some people not to abide by the rules they have set for others.Naturally, this would have set a problematic precedent, but it points to the relevance of the exact expectations group members have when trying to understand when and how they react to behaviour that could violate their expectations.Thus, we might not only need to consider that violations of group-based expectations can be threatening and that perceived control matters for understanding group members' reactions.We might also need to understand what precisely group members expect of others in the first place.

Summary and Conclusion
Individuals who face violations of their group-based expectations often react with confrontation or escape.Herein, we have proposed a novel, integrative model shedding light on the social-motivational processes underlying these reactions.In line with prior research (e.g., Abrams et al., 2000;Marques, Abrams, & Serôdio, 2001;Sani, 2008), this model casts violations of group-based expectations as threats to individuals' social selves.Specifically, the model implies that observing a violation of their group-based expectations makes individuals feel threatened.Notably, our model is the first in which this proposition rests on direct empirical evidence.Going even further beyond prior work, our model is the first to also explicitly consider individuals' resources as a factor shaping their responses.It does so by assigning a central role to perceived control.In short, we suggest that violations of group-based expectations are threats to individuals' social selves and are experienced as such, paving the way for confrontation or escape responses.The extent of control individuals perceive over the violator's behaviour or a group's future then determines which of the two reactions they choose.In case they perceive low control, they should opt for escape.On the other hand, if they perceive at least some control, they should opt for confrontation.
We have summarised a substantial body of evidence from our own and related prior work supporting this model.Speaking to the broad applicability of the model, this evidence spans different types of groups (e.g., opinionbased lab groups, Facebook groups, personally relevant ingroups) and different violations (e.g., not contributing to group success, dishonesty, making xenophobic statements).Moreover, it addresses multiple sources of perceived control (i.e., ingroup identification, social support, and social power).This breadth of operationalisations further attests to the model's integrative potential, which can be further enhanced by following up on the several avenues for future research we have outlined above.Therefore, we hope that this article and the model it presents move us closer to understanding the factors and processes that influence group members' reactions to situations involving discrepancies between their experiences and expectations.In addition, we hope that it will inspire further research aimed at a greater integration in our field.

Figure 1 .
Figure 1.Sources of perceived control in group contexts.

Figure 2 .
Figure 2. Threat-and-control model of reactions to group-based expectancy violations.
et al. (2017), S2 are based on a SEM including the three possible outcomes at the same time; results forDitrich et al. (2019), S1 are based on separate mediation models, controlling for pre-manipulation identification as a covariate.The results for escape inDitrich et al. (2019), S1 are reported taking a moderation of the direct effect of violations on escape by opinion support into account.

Table 1 .
Conditional indirect effect of violation extremity via identity subversion at three levels of centrality.

Table 3 .
Descriptive statistics of ratings for focal group member by condition,Ditrich et  al. (2017).

Table 4 .
Descriptive statistics and inferential statistics for the violation (main) effect.We controlled for pre-manipulation identification in our analyses for post-manipulation identification.For this variable, we report estimated means and standard errors.Data fromDitrich et al. (2017).Time to go! Leaving the group in response to norm-deviations.Journal of Experimental Social Psychology, 73, 259-267.https://doi.org/10.1016/j.jesp.2017.07.005 and Ditrich et al.

Table 7 .
Indirect effects (IE) via the parallel mediators on confrontation and escape responses.