(2017). Endorsing narratives under threat: maintaining perceived collective continuity through the protective power of ingroup narratives in Northern Ireland and Cyprus. Journal of Social and (2),

In historically divided societies, narratives play an important role in understanding the maintenance and perception of identities and group status over time. Moving beyond the consequences of identity continuity perceptions, in this paper we were interested in the antecedents of perceived collective continuity, specifically the role of narratives and threat perceptions. We predicted that endorsement of ingroup narratives would be associated with higher perceived continuity through stronger perceptions of group threat. To test this hypothesis, we recruited participants from both majority and minority groups in Northern Ireland (N = 268) and in Cyprus (N = 413) to complete an online survey to examine their endorsement of ingroup and outgroup narratives, their perceptions of identity continuity and feelings of threat. We tested these predictions with path analyses. In line with our hypothesis, results demonstrate that, for both majority and minority groups, if group members feel threatened they are more likely to endorse their ingroup historical narrative as this helps them to maintain a sense of continuity of their ingroup identity. Findings are discussed in relation to the importance of considering narratives in intergroup relations.

that it can be satisfied by identifying with temporally enduring groups, such as the national community (Smeekes & Verkuyten, 2014a). Furthermore, research has found that people tend to oppose social developments and outgroups that undermine ingroup continuity (Jetten & Hutchison, 2011;Jetten & Wohl, 2012).
While these studies indicate that perceived continuity is an important feature of group life, we know little about the ways in which perceived continuity of the ingroup can be bolstered or maintained. It has been suggested that historical ingroup narratives serve to maintain a sense of collective continuity (Liu & Sibley, 2009;Paez & Liu, 2011), but this claim has so far not been empirically examined. Historical narratives are individuals' mentally represented stories of their own group (ingroup narrative) or relevant outgroups (outgroup narrative). They are typically essentialist stories about the history of the intergroup conflict and the role of the ingroup and outgroup in this trajectory. These historical narratives are part of collective memory as they are social representations (or shared knowledge about the past) that are collectively constructed through interpersonal and institutional communications (Paez & Liu, 2011).
Groups in conflict often endorse competing historical narratives that reflect the shared experience of the group and that serve to justify the group's status and position in society by explaining how it came to be that way (Liu & Sibley, 2009). Groups may especially hold on to their historical ingroup narrative when they have the feeling that their group identity is at stake, as these narratives help them to maintain a sense of continuity of their group identity. The main aim of the current paper is to examine whether group members who feel threatened are more likely to embrace their historical ingroup narrative as a means to maintain a sense of ingroup continuity. We examine these relations in two historically divided contexts, namely Northern Ireland and Cyprus. Both contexts have a long history of intergroup conflict between the majority and minority population and interventionist policies by outside "mother countries". In Northern Ireland, this concerns the conflict between Protestants and Catholics and in Cyprus, the conflict between Greek Cypriots and Turkish Cypriots.

Perceived Collective Continuity
Collective history is central in developing and shaping group identity. History provides people with an understanding of who they are, where they are coming from and what the foundations and contents of their group identity are (Liu & Hilton, 2005). Since collective history gives people a sense of common fate and identity, individuals find comfort in the belief that their group has endurance over time. A growing body of research has shown that a sense of collective continuity provides existential security (Sani et al., 2007;Sani, Bowe, & Herrera, 2008;Sani, Herrera, & Bowe, 2009) and forms an important reason for why people feel attached to their national group membership (Jetten & Hutchison, 2011;Smeekes & Verkuyten, 2014a). The reason is that even though individuals are aware that they themselves will ultimately die one day, collective continuity implies that the part of the self that is defined by group membership has transgenerational, temporal endurance (Jetten & Hutchison, 2011;Reicher, 2008).
As a continuous group history provides an important building block for a safe and secure social identity, group members are strongly affected when the continuity of their group is threatened (Jetten & Hutchison, 2011;Jetten & Wohl, 2012). Research on identity motivation argues that continuity is a basic psychological need that people will intensively seek to satisfy when it is undermined (Breakwell, 1986;Vignoles, 2011). Studies show that people increase their perceptions of ingroup continuity when they are presented with threats to their personal or group identity (Sani et al., 2009;Smeekes & Verkuyten, 2013). While this body of research indicates that perceived collective continuity can be triggered by feelings of identity threat, less is known about the factors that can help groups to maintain continuity when their identity is threatened.

Historical Narratives and Maintaining Ingroup Continuity
We propose that stronger endorsement of historical ingroup narratives could be one explanation for why threats to the group lead to greater perceptions of ingroup continuity. Following work on identity motivation (Breakwell, 1986;Vignoles, 2011), it is likely that group members seek ways of maintaining or regaining a sense of group continuity when their group identity is undermined. Endorsement of historical ingroup narratives could be one way in which perceived continuity of the ingroup is bolstered in the context of threats to the ingroup. Various studies have shown that people tend to engage in ingroup defensive strategies when their group is threatened (Smeekes & Verkuyten, 2013;Wohl, Branscombe, & Reysen, 2010). The reason is that such strategies buffer the undesired feeling of identity threat. We suggest that the endorsement of historical ingroup narratives can be seen as an ingroup defensive strategy, or coping mechanism, that helps group members to maintain a sense of ingroup continuity.
Scholars propose that one of the most important functions of historical narratives is to provide a sense of collective continuity (Paez & Liu, 2011). The reason is that historical ingroup narratives mostly give an essentialist account of the group's origins, common history and relation to the land (Kelman, 2001). Essentialism in psychology refers to people's tendency to understand their groups (and the elements that define them) in terms of immutable and fixed characteristics (Haslam, 1998). Social psychologists have shown that essentialist understandings of ingroups more strongly satisfy basic psychological needs related to group membership than an understanding of groups being more mutable (Yzerbyt, Castano, Leyens, & Paladino, 2000).
In situations of intergroup conflict, essentialist historical ingroup narratives are typically polemical, in other words different versions of the history of the intergroup conflict exist between different groups. Both groups attempt to create a historical self-narrative that serves to maintain a sense of continuity of the ingroup (e.g., Liu, Wilson, McClure, & Higgins, 1999;Psaltis, 2012Psaltis, , 2016Raudsepp & Wagner, 2012;Sen & Wagner, 2005) and hereby provides existential security. These historical narratives help people retain the core properties of their group over time in the midst of change, and are also used to justify political agendas (Liu & Sibley, 2009).
Historical narratives are in large part unique to a culture or group. While the contents of these historical narratives may depend on the specific cultural or national context, the function of these narratives may be similar for groups.
We predict that regardless of what the specific contents of these historical ingroup narratives are they will function as a buffering mechanism for maintaining continuity in the context of threat. We also predict that this works the same way for majority and minority groups.
i We focused the present research on two historically divided contexts where narratives and history continue to dominate intergroup relations; Northern Ireland and Cyprus.
Although Northern Ireland is now in a post-agreement phase, the constitutional state of the island remains a contentious issue. It is estimated that during the 30 years of The Troubles, 3,600 hundred people were killed, and a further 30,000+ injured (Fitzduff & O'Hagan, 2009).
The narrative surrounding the causes of the conflict differs depending on the group with which one identifies. According to Whyte (1991), the traditional Nationalist interpretation focuses on two key concepts: 1) the people of Ireland form one nation and 2) the fault for keeping Ireland divided lies with Britain. Similarly, the traditional Unionist interpretation can be defined in terms of two concepts which differ from the traditional Nationalist interpretation. That is, 1) there are two distinct peoples in Ireland, Unionists and Nationalists (or Protestants and Catholics) and 2) the core of the problem is the refusal of Nationalists to recognise this fact, and to allow Unionists the same right of self-determination as they claim for themselves.
Who constitutes a minority or majority in Northern Ireland is often debated. Some researchers have argued that both communities constitute a minority (Catholics in Northern Ireland, Protestants on the island of Ireland) and as a result have negative social identities (Trew & Benson, 1996). Whereas others, such as Cairns (1982), suggest that both communities represent a majority (Protestants are a majority in Northern Ireland and Catholics a majority on the island of Ireland) as both generally view their identity favourably. In the present research, we view Protestants as a majority and Catholics as the traditionally disadvantaged minority group. While these groups are nowadays numerically comparable in size (both around 45%) ii , Catholics can be considered as the minority due to their historical disadvantage during the conflict (as noted above) and their continued disadvantage. Evidence suggests that in today's Northern Ireland, Catholics are more likely to be unemployed and in poor health, compared to Protestants (Nolan, 2014).
Whilst Northern Ireland has undoubtedly moved forward since the height of the conflict, having signed a peace agreement in 1998 which was implemented in 1999, it remains a troubled society. The conflict has had disastrous, and difficult to quantify, consequences on many aspects of Northern Irish society. Beyond the economic costs, this includes the incidence of conflict related crime and violence, as well as the conflict's ongoing impact on mental health, societal segregation and intergroup relations.

Cyprus
The conflict in Cyprus goes back to the 1950s when Cyprus was still part of the British Empire. Greek Cypriots (82% of the population) began to seek a union with Greece, which was opposed by the Turkish Cypriot minority (18%) who embarked on a struggle for partition of Cyprus between Greece and Turkey. In 1960 Cyprus gained its independence and a consociational partnership between Greek Cypriots and Turkish Cypriots was established along with the Republic of Cyprus, a unitary state that joined the United Nations (UN). The divergent national aspirations of the two communities and conflict over power sharing at the elite level led to 'The troubles ' of 1963-1964 in the form of violent inter-communal clashes and increased segregation as a reduction of cohabitation in more than 100 mixed villages . A coup in 1974, aimed at the union of Cyprus with Greece, engineered by the Greek military junta, prompted a military invasion/intervention by Turkey that led to fatalities, mass executions and major displacements of the population and the division of the island into two ethnically homogeneous areas. This eventually resulted in the establishment of a breakaway state by the Turkish Cypriot leadership in the North, which is recognized only by Turkey (see Kitromilides, 1977;Papadakis, 2005).
According to Papadakis (2008), the central nationalistic historical narrative in the Greek Cypriot community as represented in history textbooks is one that begins with the arrival of Greeks (14th century BC) in Cyprus that leads to its Hellenization, where the moral centre are Greeks (of Cyprus) and the major enemy are Turks. The plot concerns a struggle for survival by Cypriot Hellenism against foreign conquerors and the tragic end is the "Barbaric Turkish Invasion" and occupation of 37% of Cyprus.
The corresponding Turkish Cypriot narrative is one that begins with the arrival of Turks in Cyprus ( Given this complexity of both internal and external dynamics, who represents a majority and who represents a minority in Cyprus issue is often also contested. Both the 1960 constitution of the Republic of Cyprus and the constitutional arrangements discussed under the UN auspices today for an establishment of a Federation refer to the Turkish Cypriot community as co-constitutive and not a minority. Our use of the term minority and majority in this paper is thus premised only on the numerical inequality between the two communities.

Method Participants and Design
Participants were recruited through word of mouth, leaflets and social media and were asked to complete an online survey, which was hosted by 'Survey Monkey'.

Measures
In Cyprus, the questionnaires were translated and back-translated from English to both Greek and Turkish. Scale reliabilities are presented separately for each group.

Outgroup Threat
Outgroup threat was assessed by using items that have been used in previous studies to assess symbolic and realistic threat (Psaltis, 2012;Stephan & Stephan, 2000). For both measures, participants were asked to respond to a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Realistic threat was measured by 4 items as follows: 'The more power the other community gain in this country, the more difficult it is for my community', 'Allowing the other community to decide on political issues means that my community has less to say in how this country is run', 'I worry that members of the other community will claim more and more from us in the future', and 'More good jobs for my community mean fewer good jobs for the other community'. For symbolic threat, participants responded to 3 items as follows: 'Protestants (Greek Cypriots) and Catholics (Turkish Cypriots) in Northern Ireland (Cyprus) have very different values', 'The other community sometimes does things that my community would never do', 'The other community is beginning to project their identity in a way that I find threatening'. Although previous studies have distinguished between these two types of threat (e.g., Psaltis, 2012) they are often highly correlated and there are several studies that have treated out-group threat as a single construct (e.g., Scheepers, Gijsberts, & Coenders, 2002). As the distinction between these two types of threat is not central to our study we will also treat them as a single construct (see below for an empirical test of the distinction). The combined scale showed good reliability for all the different groups (Protestants, α = .90, Catholics α = .84, Greek Cypriots, α = .89, Turkish Cypriots, α = .83).

Historical Minority and Majority Group Narratives
In each context, narratives were measured using a series of items, which related to the dominant discourse for either the majority or minority group.
Northern Ireland. The items used to represent majority (Protestant) and minority (Catholic) narratives in Northern Ireland were based upon John Whyte's (1991)  α = .79 for Catholics). We removed the first item as this resulted in higher reliability of the scale for both Protestants (α = .63) and Catholics (α = .80).
Cyprus. The items used to represent majority (Greek Cypriot) and minority (Turkish Cypriot) narratives in Cyprus were based upon the analysis of Papadakis (2008) and Psaltis (2016). Participants were asked to respond to a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The minority narrative also comprised three items as follows: "In 1974 Turkey intervened in Cyprus to protect Turkish Cypriots", "TMT was established as the result of Turkish Cypriot efforts to protect themselves", and "The Cyprus problem is the result of interventionist policies by Greece in Cyprus" (α = .50 for Greek Cypriots, α = .65 for Turkish Cypriots).

Perceived Collective Continuity
We measured perceived collective continuity (PCC) by using the scale developed by Sani et al. (2007)

Confirmatory Factor Analyses
We performed confirmatory factor analyses (CFA) on the full sample (including all the groups) using AMOS 22.0 software to examine whether symbolic and realistic threat form a single factor. The CFA revealed that the two types of threat were not empirically distinguishable in this sample. Although the two correlated-factors model We therefore adopted a one-factor solution for outgroup threat.
We also performed another CFA to examine whether perceived collective continuity formed a single factor. Previous studies have also found evidence for a two-factor structure for perceived collective continuity consisting of historical and cultural collective continuity (see Sani et al., 2007;Smeekes & Verkuyten, 2014a). We therefore compared a two-factor model seperating these two dimensions to a one factor model in which all items were taken together. Endorsing Narratives Under Threat 288 The CFA revealed that the two types of continuity were not empirically distinguishable in the full sample.

Measurement Model
We proceeded to test several models to assess measurement equivalence across the different groups. Measurement equivalence refers to a property of a measurement instrument (such as a scale variable measured with multiple items in a questionnaire) and means that the instrument measures the same concept across various sub-groups (Davidov, Meuleman, Cieciuch, Schmidt, & Billiet, 2014). We tested a measurement model separately for the two latent variables of outgroup threat and perceived collective continuity, to examine whether the proposed constructs had measurement equivalence across the four groups. The reason to assess this separately was because the sample size for each group was relatively small.
For both variables we took the following steps in the analyses. We first tested an unconstrained model (Model 1), in which all parameters were estimated freely for the four groups. We subsequently tested for metric invariance (Model 2), by constraining the factor loadings to be equal across the four groups. When metric invariance is observed it means that people from the different groups understand the items in the same way. In Model 3, we tested scalar invariance by constraining both the factor loadings and intercepts to be equal across groups. When scalar invariance is observed it means that the measurement of the latent variable is the same across groups. The model comparisons can be found in Table 1. As can be seen in Table 1, none of the constrained models could be rejected on the basis of the fit indices. These analyses support the scalar measurement invariance of the measurement model with the two latent factors for PCC and threat across the four groups.

Path Analysis
We subsequently conducted path analysis in AMOS 22.0 to test our predictions regarding the effects of threat on the endorsement of minority and majority historical narratives and perceived collective continuity. viii In this model, we used manifest variables for all constructs as the sample size per group was too small to estimate a multiple group model with latent variables. We created mean scores for both minority narrative (i.e., the mean score of the Catholic narrative items in Northern Ireland and the mean score of the Turkish Cypriot narrative in Cyprus) and majority narrative (i.e., the mean score of the Protestant narrative items in Northern Ireland and the mean score of the Greek Cypriot narrative in Cyprus). The path model is presented in Figures 1 (Northern Ireland) and 2 (Cyprus). In this model, we controlled for age and gender by adding paths between these variables and the mediator and the dependent variables and by correlating them with outgroup threat. Means, standard deviations, and bivariate correlations between all continuous manifest variables for the full sample are presented in Table 2.
We also report these results for each group seperately in Appendix B. The standardized paths and explained variance (R 2 ) for each of the four groups are shown in Figures 1 and 2.  Note. Influence of perceived outgroup threat on perceived collective continuity, via endorsement of majority and minority historical narratives. Path-coefficients are standardized estimates and the path coefficient in brackets reflects the standardized total effect. ***p < .001. **p < .01. *p < .05.
As can be seen in the figures, a similar pattern of results emerges for the majority and minority groups in the different countries. If group members feel threatened they are more likely to endorse their own ingroup historical narrative, and this ingroup narrative is positively related to feelings of collective continuity. For all groups, threat is a significant positive predictor of perceived collective continuity. Except for Greek Cypriots, the main effect of threat on perceived collective continuity was no longer significant when the two mediators (i.e., majority and minority historical narratives) are added to the model.
For majority and minority groups, the path model explains a moderate amount of variance in perceived collective continuity in Northern Ireland and a more substantive amount in Cyprus. For both minority groups, outgroup threat explains a substantive amount of variance in the endorsement of the minority narrative but not in the endorsement of the majority narrative. For both minority groups, the path from outgroup threat to the majority narrative is not significant. This suggests that, for minority groups, perceiving outgroup threat is a good predictor for the endorsement of their own narrative, but not so much for rejecting the narrative of the majority. For the majority groups there is quite a big difference in the explained variance: Where perceived outgroup threat explains a substantive amount of variance in endorsement of the majority and minority narrative among Protestants, only a small amount of variance in these variables is explained by outgroup threat among Greek Cypriots. We come back to this finding in the discussion.

Indirect Effects
We subsequently tested whether the endorsement of historical narratives significantly mediated the effect of threat on PCC for each of the four groups, using bootstrapping procedures in AMOS. We generated 1000 random bootstrap samples with replacement from the dataset and tested the model with these samples.

Discussion
Continuity is an important feature of identity (Sani et al., 2007) and evidence supports that the perception of identity continuity has consequences for intergroup relations (Smeekes & Verkuyten, 2015). Little is known, however, about the ways in which perceived continuity can be bolstered or maintained. The aim of this study, therefore, was to examine whether group members who feel threatened are more likely to embrace their historical ingroup narrative as a means to maintain a sense of ingroup continuity. We examined these relations in two historically divided societies, Northern Ireland and Cyprus. We predicted that regardless of the specific contents of these historical ingroup narratives, they would function as a buffering mechanism for maintaining continuity in the context of threat. We also predicted that this works the same way for majority and minority groups.
In line with our predictions, we found that when group members feel threatened they are more likely to endorse ingroup historical narratives and that these narratives are positively related to feelings of collective continuity. This was observed for minority and majority group members in both contexts and supports previous assertions that historical ingroup narratives serve to maintain a sense of collective continuity (Liu & Sibley, 2009;Paez & Liu, 2011). It also offers support for our argument that when threatened, groups may hold on to their historical ingroup narratives, as these narratives help them maintain a sense of continuity of their group identity. Importantly, indirect effects demonstrated that only ingroup narratives (not outgroup narratives) mediated the relationship between threat and perceived collective continuity for all of our groups. This highlights the importance of understanding how different groups form narratives surrounding conflict and the impact of endorsing these narratives of intergroup relations. It also suggests that a historical self-narrative provides existential security, and can help individuals to maintain a sense of continuity and core properties of identity over time (Liu & Sibley, 2009).
Based on previous literature and our findings, therefore, we argue that ingroup narratives may serve as an ingroup defensive strategy that helps group members maintain a sense of identity continuity. This is in line with previous Endorsing Narratives Under Threat 292 research showing that individuals tend to engage in ingroup defensive strategies when their group is threatened (e.g., Smeekes & Verkuyten, 2013;Wohl et al., 2010) and theoretical accounts proposing that narratives provide a sense of collective continuity (Paez & Liu, 2011).
Unlike some previous studies (see Sani et al., 2007;Scheepers et al., 2002;Smeekes & Verkuyten, 2014a) we found that both the items measuring perceived collective continuity and perceived outgroup threat formed one scale instead of two sub-scales. Previous studies on perceived outgroup threat have obtained mixed results as to whether symbolic and realistic threat can be empirically distinguished. It seems that whereas these are clearly different constructs in some contexts (e.g., Stephan & Stephan, 2000) they are grouped together in others (e.g., Scheepers et al., 2002). These different findings probably have to do with the dominant public discourses on the presence of the outgroup, which are likely to differ between national contexts. Arguably in Northern Ireland and Cyprus, perceived symbolic and realistic threat from the outgroup are not clearly separated in the public discourse.
As for perceived collective continuity, no previous work has measured this variable in Northern Ireland and Cyprus taking into consideration both majority and minority groups. It seems that in these contexts people do not clearly distinguish these forms of collective continuity. This is probably because in ethno-political conflicts as the ones described here these two forms of continuity often merge and are used interchangeably in public discourses and history teaching on national identity (Psaltis, Carretero, & Cehajic-Clancy, in press).

Findings in Context
Our findings also offer support for the importance of understanding the conditions under which ingroup narratives may be adopted and the consequences for intergroup relations in Northern Ireland and Cyprus. In both contexts for majority and minority group members, higher perceptions of threat were found to be related with increased support for ingroup narratives and, in turn, stronger feelings of collective continuity. It could be argued, therefore, that even in settings of relative peace, compared to previous years, threat perceptions play a key role in sustaining and perpetuating a cycle of intergroup relations which encourages individuals to engage in defense mechanisms that can be detrimental to intergroup relations. This is because whilst identity continuity has positive consequences for the self, it is often associated with negative attitudes towards outgroups that potentially undermine the continuity of the ingroup (Smeekes & Verkuyten, 2015). This suggests that for settings such as Northern Ireland and Cyprus, tackling threat perceptions and challenging ingroup narratives may help break down the oppositional nature of identity and assist in developing an identity with which both majority and minority group members can identify.
One way to tackle this may be through the development of intergroup contact-based interventions, which could facilitate the development of a more multi-perspective approach to representations of the past and history teaching and learning in particular. This would encourage discussions which could tackle the polarizing historical narratives that dominate society and at the same time reduce intergroup threats. Indeed, evidence from peacemaking youth programs in Northern Ireland supports the notion that addressing issues is an important step to promoting intergroup relations (McKeown & Cairns, 2012). It was also recently recognized that a new way of teaching the history of intergroup conflicts can be a very beneficial route for the cultivation of both the critical historical thinking skills of students and prejudice reduction at the same time (Psaltis, Carretero, & Cehajic-Clancy, in press). In this new way of teaching students learn how to question a historical account, to become aware of the evidentiary base upon which it rests, and to assess it in relation to contrasting accounts. In this way, it becomes clear for students that history is not a fixed story, but is subject to debate and often altered by time and hindsight.

Limitations and Future Directions
The current study has a number of limitations that could be addressed in future research. First, we used a correlational design, which means that we cannot make causal claims about the processes we tested. It is possible that the reverse path from adherence to historical narratives to threats is at work through a heightened sense of outgroup threat conveyed via narratives of ingroup victimization (cf. Psaltis, Franc, Smeekes, Ioannou, & Žeželj, in press). There is certainly a need for longitudinal studies to further explore both causal directions at the same time.
Whilst the main effect of threat on perceived collective continuity was no longer significant when majority and minority historical narratives were added to the model for Protestants, Catholics and Turkish Cypriots, the direct effect remained significant for Greek Cypriots. Also, the explained variance in endorsement of both the majority and minority narrative was low for this community. This suggests that other mechanisms, or other historical narratives, might be at work in explaining the relationship between threat and perceiving collective continuity for this group. For Greek Cypriots, a related representation of the past (that could potentially explain this relationship) was not taken into account in the current study. This historical narrative goes beyond the usual Hellenocentric narratives presented in the literature (e.g., Papadakis, 2008) and focuses on protecting the 1960 constituted Republic of Cyprus against the efforts of Turkey to dissolve the Unitary state of 1960 as a way of maintaining group continuity.
We would expect the use of scales with more items (including beliefs capturing this aspect) to yield similar findings as those reported here for Protestants, Catholics, and Turkish Cypriots (cf. Psaltis, Lytras, & Costache, 2011).
Combined with using a small number of items to measure narratives, which can reduce reliability, this additional consideration regarding historical narratives in Cyprus may also explain why the reliability of our measures of ingroup and outgroup narratives were lower for Greek Cypriots, compared to Turkish Cypriots. Future studies could examine whether these or other historical narratives could play a role in explaining the relationship between threat and perceiving ingroup continuity for Greek Cypriots. It may be particularly beneficial, for example, considering how individuals explain and describe ingroup and outgroup narratives through their own discourse. This may also facilitate the development of a more reliable measure of ingroup and outgroup narratives for each context. It would arguably also enable a deeper understanding of the nature of group status and how this interplays with narrative endorsement in given situations. Perhaps most interesting for future research, however, is thinking about how these narratives might be best challenged and changed. It may be possible, for example, to experimentally manipulate narrative endorsement and test the causality of these effects on intergroup relations. The development of a multi-perspective collective narrative that pays due respect to the historical thinking skills may provide exciting opportunities for improved relations or alternatively the promotion of a reflective and de-constructive approach to master narratives in post-conflict settings might also be the way forward (see Psaltis, Carretero, & Cehajic-Clancy, in press).
As the explained variance in the endorsement of historical in-group narratives and perceived collective continuity in our models is not always high, this suggests that there are other important factors at play that we did not take into account in the current study. Previous work has demonstrated perceived outgroup threats have stronger effects on intergroup attitudes for higher than for lower in-group identifiers (see Jetten & Wohl, 2012). This suggests that the relationships tested in our model could be dependent on the extent to which participants identify with their group and this is an interesting follow-up idea for future studies. Furthermore, recent studies show that one way in which in-group members can try to restore the continuity of their in-group in the context of group threat is by Endorsing Narratives Under Threat 294 strengthening their feelings of collective nostalgia (see Smeekes & Verkuyten, 2015). The reason is that longing for the good old days of the shared past helps people to understand which aspects of their group identity they value and want to see continued in the future. Hence, collective nostalgia could be another mechanism that explains the relationships between outgroup threat and perceptions of collective continuity, and this would also be an interesting avenue for future research.

Conclusions
A number of studies have examined the effect of identity continuity on intergroup relations, but to date little research has examined the factors which help to maintain or bolster identity continuity. In this paper, we offer support for the importance of understanding the role of threat in increasing support for ingroup narratives and thereby bolstering collective continuity in both Northern Ireland and Cyprus. We argue that it is important to better understand the role of historical narratives and their relation with continuity and intergroup relations, given that there is evidence that representations of continuity could both enhance identification with the ingroup and damage intergroup relations with various outgroups.

Notes
i) We recognise that minority/majority status can be contested. Here we use this distinction to better understand the relationship between our key variables, rather than to argue that these are accepted terms by the groups in each context.
ii) See https://namawinelake.wordpress.com/2012/12/12/what-happens-in-2016-when-northern-irelands-catholics-are-on-a-par-with-protestants/ iii) Rum is the term used to describe the Greek Orthodox inhabitants of the Ottoman empire and Greek orthodox minorities in contemporary Turkey and relates to the Greek word Romioi.
iv) Whilst the reliabilities for historical narratives is low for Greek Cypriots this is arguably a function of using few items to measure narratives. For example, when using a 6-item scale in Psaltis, Lytras, and Costache (2011) Cronbach's α for Greek Cypriots was .65 and for Turkish Cypriots .66. When using a 7-item scale in Lytras and Psaltis (2011) Cronbach's α was .67 for both communities.
v) We allowed the error term of two symbolic threat items (i.e., "Protestants and Catholics in Northern Ireland have very different values" and "The other community sometimes do things that my community would never do") and two realistic threat items (i.e., "Allowing the other community to decide on political issues means that my community have less to say in how this country is run" and "More good jobs for my community mean fewer good jobs for the other community") to correlate.
vi) These were items 7 and 12 from the perceived collective continuity scale (see Appendix A). An explanation for these lower loadings could be that these items are somewhat abstract, which could imply that participants have some difficulties in understanding and answering these items. Previous work has also observed low factor loadings for these items (see Smeekes & Verkuyten, 2014b; Study 1).
vii) Item 3 with Item 5, Item 3 with Item 2, Item 3 with Item 4, and Item 1 with Item 4 (see Appendix A).
viii) We used manifest variables as the sample size was too small to estimate a structural equation model with latent variables. This model could not comply with the standard of 10 observations per estimated parameter.
ix) We tested whether the pattern of results was affected by the deletion of the historical narrative items that had a low factor loading (see Measures section) by running all the analyses again with variables for the majority and minority narrative including these items. The pattern of results remained the same speaking to the robustness of our findings. Results are available from the authors upon request.