How does ingroup identification predict forgiveness in post­ conflict societies? The role of conflict narratives

People's religious identity is often the central identity in many ethnopolitical conflicts. These identities in conflict contexts may be associated with how people see conflict and their willingness to forgive the outgroup members for their wrongdoings in the past. Study 1 ( N = 287) tested how religious group identification in the Northern Irish context predicted forgiveness through the endorsement of dominant conflict narratives (i.e., terrorism and independence narratives) among Protestants and Catholics. We also tested how group membership may moderate these relationships. The results showed that among Protestants, higher Protestant identification predicted less forgiveness through higher endorsement of the terrorism narrative and less endorsement of the independence narrative. Among Catholics, on the other hand, higher Catholic identification predicted stronger endorsement of the independence narrative, and in turn, less forgiveness. Study 2 ( N = 526) aimed to replicate the findings of Study 1 with a larger sample and extend them by testing the role of an alternative conflict narrative (i.e., the Northern Irish identity narrative). The results were largely replicated for the independence and terrorism narratives, and the Northern Irish identity narrative was associated with higher forgiveness across both groups. We discuss the results in terms of how ingroup identities and conflict narratives can become both facilitators of and barriers to peacebuilding in post-conflict societies.


BACKGROUND
Conflict narratives can be conceptualized as stories that frame the conflict, including its definitions, causes, and potential solutions. The way people see conflicts has consequences for conflict-and peace-related outcomes because they shape people's reactions to possible solutions to the conflict (e.g., Bar-Tal, 2013;Kaufman et al., 2003). Previous research in the Turkish-Kurdish conflict context (Uluğ & Cohrs, 2016, 2017a found two opposing conflict narratives: (a) a terrorism narrative (i.e., describes the conflict as stemming mainly from the armed wing of the minority group) and (b) an independence narrative (i.e., describes the conflict as a need for independence for the minority). It has been shown that these conflict narratives not only predict support for non-violent conflict resolution and aggressive policies but also can be generalized to other conflict contexts such as the Israeli-Palestine conflict (Uluğ, Lickel, et al., 2021). However, less attention has been given to the antecedents of conflict narratives such as identifying with an ingroup identity in an ethnopolitical conflict context. We argue that people's ingroup identities such as their religious identities are particularly crucial in such contexts because it is on these identities conflicting parties form shared beliefs with respect to the conflict (Bar-Tal & Oren, 2000). In the current study, we aim to examine how conflict narratives may play an important role in the relationship between religious group identification (i.e., Catholic vs. Protestant) and willingness to forgive the other party for their wrongdoings during the conflict.
The role of conflict narratives from a social-psychological perspective has so far been examined in conflict contexts such as the Turkish-Kurdish conflict and the Israeli-Palestinian conflict (Uluğ & Cohrs, 2017a;Uluğ, Lickel, et al., 2021). However, one can argue that conflict narratives may also play a significant role in post-conflict societies, where conflicting parties still endorse some of those narratives (see, e.g., Psaltis et al., 2017) and have difficulty forgiving the perpetrator groups. As previous research has not investigated how conflict narratives that were developed in ongoing conflict contexts can be applied to post-conflict societies, it is our aim to test whether those conflict narratives are generalizable to post-conflict societies.
We addressed these research gaps in two studies conducted in Northern Ireland as a post-conflict society. We had three research objectives. Across both studies, we first aimed to examine how religious group identification may be related to endorsing certain conflict narratives and being willing to forgive the outgroup members. Second, we aimed to investigate whether conflict narratives would mediate the relationship between ingroup identification and forgiveness. Third, we sought to understand these questions from the perspective of both sides of the conflict: Protestants and Catholics.

Religious group identification in (post-)conflict societies
People's religious identities, like Catholic and Protestant identities in Northern Ireland, are often central in ethnopolitical conflicts, and as such, these identities can drive peace-related outcomes. For instance, previous studies have shown how ingroup identification plays a key role in the conflict-and peace-related outcomes in both conflict and post-conflict contexts (Cehajic et al., 2008;Hanke et al., 2013;Hewstone et al., 2006;Leonard et al., 2015;Noor, Brown, Gonzalez, et al., 2008). Accordingly, we conceptualize religious identity in the current research context as a salient ingroup identity, and we propose that a salient ingroup identity in conflict contexts is particularly important for conflict narratives for two reasons.
First, in conflict contexts, ingroup identity may help group members form shared beliefs with respect to the conflict (see, e.g., Bar-Tal & Oren, 2000). Group dynamics based on shared identities show how identities play a key role in the outbreak of a conflict (Uluğ, 2016). As soon as the group members recognize a specific situation as a conflict, they try to understand the conflict situation by trying to find answers for the sources of the conflict, responsible parties for the conflict, and characteristics of the opposing conflict party (Bar-Tal, 2000). At the end of this cognitive processing, conflicting parties form conflict beliefs, not objectively but in a biased way (Bar-Tal, 1990). This belief is shared by different groups at the societal level. When a conflict is understood by people to have religious fault lines, this understanding and the corresponding reliance on religious identity provides people more readily and easily with a perceived explanation of how the conflict developed, where it originated from, and perceptions of what the adversarial group is like.
Second, the meaning attached to identities also matters. For example, based on ingroup identities, some groups are seen positively, whereas other groups are perceived as threatening (Bilali et al., 2018). Therefore, ingroup identification could have different implications depending on the content of group identity. We argue that these group identities (depending on their content) may shape the way people understand the conflict from their own group's perspective. As a result, these identities may be associated with endorsing certain conflict narratives that legitimize one's own group's actions in conflict contexts (e.g., endorsing dominant ingroup narratives; Bilali, 2014;Uluğ & Uysal, 2021). For example, Uluğ and Uysal (2021) have shown that it is the group identities that predict endorsement of one's own group's conflict narrative positively and outgroup's conflict narrative negatively, but it is not the other way around (i.e., conflict narratives predict ingroup identification). This may be the case in religious conflicts where religious group identities provide a lens on how to view a conflict, perceive one's enemies, and one's options regarding the conflict because it is these identities and values that help conflict parties make sense of their suffering (Kadayifci-Orellana, 2009).
Beyond conflict narratives, ingroup identification can also motivate or demotivate conflicting parties to forgive one another directly. Previous studies have shown how ingroup identification plays a key role in conflict-and peace-related outcomes, including forgiveness. On the one hand, it has been shown in some of those studies that strong ingroup identification such as ethnic or religious identification is a barrier to forgiveness in various conflict contexts (Cehajic et al., 2008;Hanke et al., 2013;Hewstone et al., 2006;Leonard et al., 2015;Noor, Brown, Gonzalez, et al., 2008) whereas no significant relationship between ingroup identification and forgiveness was found in others (Hamer et al., 2017;Hewstone et al., 2006;Leonard et al., 2015;Philpot & Hornsey, 2011;Uluğ, Bilali, et al., 2021). Cehajic et al. (2008) argue that the reason behind the negative relationship between ingroup identification and forgiveness in conflict contexts may be that forgiveness as an act may be seen as disloyal by ingroup members. On the other hand, some other studies also show that ingroup identification, particularly with higher-order groups such as national identity (Baysu & Duman, 2016) or even more inclusive humanity identity (Greenaway et al., 2011;Noor et al., 2010;Wohl & Branscombe, 2005; but see Ünal et al., 2022), can be a facilitator of forgiveness.
Even though the link between ingroup identification and forgiveness has been established in previous research, and it has been shown that to the extent you identify with one of the conflicting parties as your ingroup, then you tend to forgive the other conflicting party less (see Van Tongeren et al., 2014 for the meta-analysis of intergroup forgiveness), how people view, understand and frame the conflict has never been tested as an important process variable linking the effects of ingroup identification to forgiveness. We argue that endorsement of conflict narratives can be an explanatory factor here because the degree of people's ingroup identification may be related to how they view the conflict and they form shared beliefs in conflict contexts (see, e.g., Bar-Tal & Oren, 2000). In other words, to the extent people identify with their own groups in conflict contexts, they would endorse their own group's conflict narrative. Therefore, we expect that people who highly identify with their own group, even in the context of a post-conflict society, would endorse their own group's dominant conflict narrative more, which would become a barrier to forgiving the other (i.e., outgroup members). This is especially the case in post-conflict societies where a clear-cut division between "us" and "them" still tends to exist.

Conflict narratives and forgiveness
To be able to cope with the conflict in which people live, conflicting parties develop socially shared beliefs about the conflict itself. These socially shared beliefs usually create dominant narratives that conflicting parties mostly share in that society (Bar-Tal et al., 2012). Conflict narratives can be considered one's holistic view of a conflict, and they are different from specific beliefs related to the conflict. These beliefs often become a basis for the continuation of conflict (Bilali & Mahmoud, 2017).
To approach conflict narratives, previous studies (e.g., Uluğ & Cohrs, 2017a, 2017b, 2017cÜnal et al., 2022) have used Entman's (1993) systematic frame analysis that has four domains: (a) problem definition, (b) source of the problem, (c) moral evaluation, and (d) solution to the problem. Even though specific beliefs related to each domain may be related to different conflict-and peace-related outcomes, conflict narrative is a concept that is bigger than its domains, and it is distinct from them. For example, in the Turkish-Kurdish conflict context, Uluğ and Cohrs (2017a, 2017b, 2017c showed that even if people's conflict definitions may be similar, the solutions they offer to the same conflict may differ, and these are reflected in their endorsement of different conflict narratives (e.g., democracy and rights narrative vs. democracy and Islam narrative). To be able to understand people's holistic attitudes to conflict, using the conflict narratives approach has some advantages as narratives show people's position on how they view the conflict, who/what they blame for the causes of the conflict as well as what kind of suggestions they recommend for the resolution of the conflict altogether.
Past research has shown that endorsement of the ethos of conflict has consequences for conflict resolution and peacebuilding in the Israeli-Palestine conflict context. These outcomes include but are not limited to reducing support for compromise and peaceful conflict resolution (Canetti et al., 2017), increasing positive stereotyping of ingroup members and negative stereotyping of outgroup members (Bar-Tal et al., 2009), and contributing to the continuation of the conflict and reducing the likelihood of peaceful resolution (Porat et al., 2015). Therefore, Bar-Tal and Halperin (2011) argue that the ethos of conflict can create social-psychological barriers to conflict resolution and reconciliation (see also Rafferty, 2020).
As mentioned earlier, previous research in the Turkish-Kurdish and Israeli-Palestinian conflict contexts highlighted two opposing conflict narratives: (a) a terrorism narrative as a dominant narrative mostly endorsed by the majority group members and (b) an independence narrative as a dominant narrative mostly endorsed by the minority group members. The studies conducted in these contexts have shown that among majority group members endorsing the other group's conflict narratives is related to engaging in less competitive victimhood and less support for aggressive policies against minority group members (Uluğ, Lickel, et al., 2021).
From a social-psychological perspective, conflict narratives have usually been studied in asymmetrical conflict contexts (e.g., Turkish-Kurdish conflict) where it is relatively easy to classify conflicting parties as majority vs. minority (e.g., Turks vs. Kurds). However, we do not know (1) to what extent these narratives that were developed in conflict contexts generalize to post-conflict contexts where conflict memories are still present in the minds of communities (see Psaltis et al., 2017 for a discussion) and (2) whether we see similar trends among groups that are not easily categorized as majority vs. minority (Stevenson et al., 2007). We argue that conflict narratives may still be relevant in post-conflict contexts and have consequences for forgiveness, which considerably impacts reconstructing society after conflict. We also contend that each conflicting party would endorse their own dominant conflict narrative in a relatively symmetrical conflict context.

Overview of studies
While there has been significant progress towards equality between Catholics and Protestants since the Good Friday Agreement in 1998, the hostility between these two groups still manifests in different forms such as discrimination and segregation (Duggan, 2015;see Jeong et al., 2022 for a discussion). Even though direct violence was brought under control through the peace process, this violence's structural and cultural roots still persist even today (Ferguson et al., 2014). Given that the violent clashes between Nationalist and Loyalist rioters in Belfast and Derry erupted even in April 2021, it is crucial to understand how Protestants' and Catholics' identification with their own religious groups and endorsing their own dominant conflict narratives may be related to their willingness to forgive one another in today's society. Although the relationship between ingroup identification and forgiveness has been examined in Northern Ireland (see, e.g., Leonard et al., 2015;Myers et al., 2009;Noor, Brown, & Prentice, 2008;Taylor et al., 2022;Voci et al., 2015), how conflict narratives may be related to both ingroup identification and forgiveness has not been examined. We argue that people's conflict narratives may play an important role in the relationship between ingroup identification and willingness to forgive the outgroup members.
In two studies, we investigated the role of ingroup identification on conflict narratives (i.e., terrorism and independence narratives), as well as willingness to forgive among a sample of Protestants and Catholics. We proposed that endorsement of conflict narratives serves as a key pathway that explains the association between ingroup identification and willingness to forgive (indirect effects hypothesis). We also tested how group membership (Catholic vs. Protestant) may moderate these relationships (moderation hypothesis).
Although the majority-minority distinction is less applicable in the Northern Irish context, we argue that, for Protestants, as the historically advantaged group, the terrorism narrative (i.e., seeing the Northern Irish conflict as a problem caused by Republican paramilitary groups) would be the dominant narrative. Therefore, we hypothesize that higher Protestant identification should be related to stronger endorsement of the dominant terrorism narrative for this group, and this, in turn, should be associated with less willingness to forgive Catholics for their wrongdoings.
For Catholics as the historically disadvantaged, on the other hand, the independence narrative (i.e., seeing the Northern Irish conflict as a problem caused by the division of Ireland and Northern Ireland) would be the dominant narrative. According to this narrative, the Northern Ireland problem is seen as a united Ireland problem, and the solution lies in reunification with the Republic of Ireland. Therefore, we hypothesize that higher Catholic identification should be related to stronger endorsement of the dominant independence narrative for this group, and this, in turn, should be associated with less willingness to forgive Protestants for their wrongdoings.
In addition to dominant conflict narratives, there are also alternative conflict narratives that can be defined as non-dominant conflict narratives, as they are usually not easily categorized as the narrative of either conflicting party. The content of those alternative narratives is important, though, because some of those narratives may be pro-conflict, whereas others may be pro-peace. As these narratives tend not to be endorsed by a majority of society, they tend to be less visible in public discourse and media. Still, these alternative narratives may substantially affect conflict-and peace-related outcomes. Thus, Study 2 aimed to replicate the findings of Study 1 and extend the previous findings by examining how an alternative narrative revolving around Northern Irish identity plays a role in conflicting parties' willingness to forgive one another. This narrative, which focuses on a common Northern Irish identity to keep the peace in Northern Ireland, is not the dominant narrative among either group but an alternative conflict narrative. We argue that the Northern Irish identity can function as a superordinate identity (Gaertner & Dovidio, 2000) and provide an inclusive identity category in a post-conflict society (McNicholl et al., 2019;Trew, 1998). Therefore, this alternative narrative may play a crucial role in motivating conflicting parties to forgive one another. We hypothesized that among both Protestants and Catholics, higher ingroup identification would predict less forgiveness through less endorsement of the Northern Irish narrative.

STUDY 1
Less is known about the mediating role conflict narratives can play between ingroup identities and forgiveness in post-conflict societies. Accordingly, Study 1 sought to test whether higher religious group identification would be related to stronger endorsement of the dominant ingroup narrative for each group (i.e., terrorism narrative for Protestants and independence narrative for Catholics) and, in turn, less forgiveness in Northern Ireland.

METHOD Participants
We received IRB approval for this research from Queen's University Belfast. The study was conducted between 2018 and 2019. Participants were 18 years or older and raised in Northern Ireland. Initially, 355 participants were recruited. After the participants answered the demographic questions, the following question was asked to determine their religious community background in line with how Census in Northern Ireland asks about religious community background: "Which religious community do you identify with? (If you do not strongly identify with either, please answer according to the community you were raised in)." After excluding those who did not identify as either group (n = 21), we focused on 334 participants who self-identified as either Catholic (n = 165) or Protestant (n = 169) (44% men, 56% women, 1 preferred not to say; age range: 18-82, M = 34.24, SD = 16.22; around 7% less than high school, 50% high school, 43% university or higher). Participants were recruited via convenience sampling, which involved approaching family and friends and asking these individuals to forward questionnaires to others. Participants were also recruited via local community centres. Those who wished to participate were allowed to complete the questionnaire at home, work, university, or a community centre, and thus, the researcher was often not present. No payment or incentive was involved in the recruitment process. The sample size was not determined by a sample size estimate based on power analysis. However, we still aimed to recruit at least 300 participants (n = 150 for each group).

Measures
With the exception of the demographic items mentioned above, all items used 7-point response scales (1 = strongly disagree/not at all, 7 = strongly agree/very much).

Ingroup identification
Ingroup identification was measured with four items (Hogg & Hains, 1996): "How important is it to you to be a Catholic/Protestant?" "How much do you feel a sense of belonging with other Catholics/Protestants?" "How much are other Catholics/Protestants similar to you?" and "How happy are you to be a Catholic/Protestant?" (α for Protestants = .95; α for Catholics = .92).

Conflict narratives
Conflict narratives were developed by creating short narratives to represent the core of the different viewpoints on the conflict. These narratives have been developed based on earlier Q-methodological research (Uluğ, 2016;Uluğ & Cohrs, 2016, 2017b, 2017c. There are several ways of developing survey measures from Q-methodological findings (e.g., Baker et al., 2010). Uluğ and colleagues used the approach of creating short narratives that represent the core of the different viewpoints that have been identified in the Turkish-Kurdish conflict context. After that, these narratives have been used in empirical studies not only in the Turkish-Kurdish conflict (Uluğ & Uysal, 2021) but also in the Israeli-Palestine conflict contexts (Uluğ, Lickel, et al., 2021). We adapted these narratives to the Northern Irish conflict by taking Protestants' and Catholics' dominant perspectives into account (see, e.g., Rafferty, 2020). Based on the Turkish-Kurdish conflict narratives, we first wrote similar sentences to cover various definitions of the Northern Irish conflict, causes of the conflict, and solutions offered to the conflict based on their ingroup identities. Second, we showed these narratives to lay people from both communities, and they provided minor suggestions on the content of those narratives. After that, we finalized the content of those narratives, which are presented below.
The independence narrative was measured with "In my opinion, the Northern Ireland problem is a united Ireland problem because the status of Catholics living in Northern Ireland is like a colony under the United Kingdom. Therefore, to keep the peace in the region, Northern Ireland should be united with Ireland." The terrorism narrative was also measured with: "In my opinion, the Northern Ireland problem was caused by Republican paramilitary groups. To keep the peace in the region, the UK should monitor and end any paramilitary-style attacks or sectarian crimes that are instigated by dissident Republicans. No matter what happens, Northern Ireland should not be united with the Republic of Ireland." We asked for participants' agreement with each narrative.

Intergroup forgiveness
Intergroup forgiveness was measured with four items adapted from Noor, Brown, Gonzalez, et al. (2008), with each group responding about the other group: "I do not feel resentment towards the Catholics/ Protestants for the misdeeds that they committed in the past," "I hold ill thoughts about the Catholics/ Protestants for the misdeeds that they committed in the past (reverse item)," "I am prepared to forgive the Catholics/Protestants for the misdeeds that they committed in the past" and "I am able to forgive the Catholics/Protestants for the misdeeds that they committed in the past." The scale was reliable for both groups (α for Protestants = .83; α for Catholics = .80).

Analysis strategy
Using Mplus 8.3 (Muthén & Muthén, 2017), we ran a multi-group meditation analysis for Protestants and Catholics with forgiveness as an outcome, ingroup identification as a predictor, and independence and terrorism narratives as mediators. Age, education, and gender were used as control variables. We first ran an unconstrained (saturated) model where all the parameters were estimated freely. Then we estimated a fully constrained model where the parameters (regression coefficients, covariances, and means of the predictor variables) were constrained to be equal. When the fully constrained model had a significantly worse model fit, we released the parameters that significantly made the model fit worse based on modification indices (>χ(1) = 3.84, p < .05). In other words, these parameters were significantly different across groups. For regression coefficients, this is similar to a significant interaction by group membership. Our final models were partially constrained. Missing data were handled using Full Information Maximum Likelihood estimation (FIML). FIML uses all available data without imputing missing data and is therefore unbiased and preferable to other methods (Dong & Peng, 2013). While we use "effects" to refer to directional associations, that is, regression paths, and differentiate them from the correlations, that is, bidirectional associations among the study variables, we note that our results do not indicate causal relationships, as cross-sectional indirect effects models do not test causality (see also Hayes, 2013).

RESULTS
Descriptive statistics of the study variables are presented in Table 1. As expected, Protestants endorsed the terrorism narrative significantly more than the Catholics did and Catholics endorsed the independence narrative significantly more.
Model fit comparisons can be found in Table 2. When comparing model fit across unconstrained and fully constrained models, the fully constrained model significantly worsened the model fit, as indicated by a significant chi2 and higher values of the Akaike and Bayesian information criterion [AIC, BIC]. Based on modification indices, we released the following regression coefficients to be estimated freely across groups: the effects of ingroup identification on both narratives and the effects of both narratives on forgiveness (see Tables S1 and S2 for

Protestants
Protestant identification was associated with less endorsement of the independence narrative and more of the terrorism narrative. In turn, the independence narrative was associated with more forgiveness, whereas the terrorism narrative was associated with less forgiveness. Looking at the indirect effects, negative indirect effects of ingroup identification on forgiveness were significant both via weaker endorsement of the independence narrative (B = −0.04, SE = .02, t = −2.08, p = .037) and stronger endorsement of the terrorism narrative (B = −0.09, SE = .03, t = −3.16, p = .002). As for the control variables, the higher-educated Protestants endorsed the terrorism narrative less.

Catholics
On the contrary, Catholic identification was associated with more endorsement of the independence narrative, but it was not significantly associated with the terrorism narrative (a marginal negative association). Looking at whether and how narratives predict willingness to forgive, the independence narra- tive was associated with less willingness to forgive Protestants, but the effect of the terrorism narrative was not significant. Accordingly, the negative indirect effect of ingroup identification on forgiveness was significant only via stronger endorsement of the dominant ingroup narrative (i.e., via independence narrative, B = −0.04, SE = .02, t = −1.95, p = .051; via terrorism narrative, B = 0.01, SE = .01, t = .90, p = .371). As for the control variables, the higher-educated Catholics endorsed the terrorism narrative less.

DISCUSSION
Study 1 offered support for our hypothesis that higher ingroup identification would be related to stronger endorsement of the dominant ingroup narrative for each group (i.e., terrorism narrative for Protestants and independence narrative for Catholics). As predicted, among Protestants, higher Protestant identification was related to stronger endorsement of the terrorism narrative. Among Catholics, higher Catholic identification was related to stronger endorsement of the independence narrative.
In addition, we also found that, among Protestants, Protestant identification was associated with less endorsement of the independence narrative, which is the dominant narrative of Catholics. However, Catholic identification was not significantly associated with the terrorism narrative although the association was in the predicted direction. Even though previous studies have shown the crucial role identities play in post-conflict societies (e.g., Cehajic et al., 2008;Hanke et al., 2013;Leonard et al., 2015;Noor, Brown, Gonzalez, et al., 2008), less attention has been paid to the relationship between identities and endorsement of conflict narratives (for exceptions, see Bilali, 2014;Uluğ & Uysal, 2021). As it has been argued that identities may help ingroup members form shared beliefs with respect to the conflict in conflict contexts and shape the way people understand the conflict from their group's own perspective T A B L E 2 Model fit comparisons (see, e.g., Bar-Tal & Oren, 2000), our results complement the discussions in the literature with empirical evidence that ingroup identities may indeed be related to how groups understand, frame and narrate about conflicts even in post-conflict societies. The findings of Study 1 also supported our hypothesis with respect to the role of dominant conflict narratives in forgiveness. Among Protestants, endorsement of the terrorism narrative was associated with less forgiveness whereas endorsement of the independence narrative was associated with less forgiveness among Catholics. Even though previous studies have shown that conflict narratives may be related to attitudes towards peace (Uluğ & Cohrs, 2017a), support for pro-minority policies (Uluğ & Uysal, 2021) and for non-violent conflict resolution and willingness to forgive the outgroup (Uluğ, Lickel, et al., 2021) in conflict societies, the role of conflict narratives in people's willingness to forgive the outgroup has not been studied in post-conflict societies. Our results show that dominant conflict narratives may have consequences for forgiveness as they may still be relevant in such contexts.
Last, our results supported our indirect effects hypothesis with respect to dominant conflict narratives for both groups: The indirect effects of ingroup identification on forgiveness were significant via (a) stronger endorsement of the terrorism narrative among Protestants and (b) stronger endorsement of the independence narrative among Catholics. In addition, the indirect effect of ingroup identification on forgiveness through the mediating pathway of less endorsement of the other group's narrative (i.e., independence narrative) among Protestants was significant whereas the indirect effect via the terrorism narrative was not significant among Catholics. The results show the crucial role the conflict narratives play in the relationship between ingroup identities and willingness to forgive one another even after the conflict ends in post-conflict societies. The findings highlight how ingroup identities and endorsement of one's

T A B L E 3
Results of a partially constrained multi-group mediation model for study 1 own group's conflict narratives that are related to those identities might become a barrier to forgiving the other in post-conflict societies where the boundaries between "us" and "them" still exist.

STUDY 2
In Study 2, we had three aims: (1) to replicate the findings of Study 1, (2) to test our hypotheses with a larger sample, and (3) to extend our findings with an alternative narrative that is less visible across both groups. As mentioned earlier, we proposed a Northern Irish identity narrative that emphasizes a discourse around shared group identity. Our previous study in Northern Ireland also showed that there are both Protestants and Catholics who reject dominant conflict narratives, and they tend to endorse a socially shared conflict frame that focuses on the shared Northern Irish identity as a solution to keeping the peace . One can argue that Northern Irish identity can function as a superordinate identity (Gaertner & Dovidio, 2000) and provide an inclusive identity category in a post-conflict society (McNicholl et al., 2019;Trew, 1998). As identifying as Northern Irish is related to having more tolerant attitudes towards the people of other religions (Lowe & Muldoon, 2014), we argue that endorsing a narrative that emphasizes this should be related to more willingness to forgive people of other religions. Therefore, we hypothesized that, among both Protestants and Catholics, stronger endorsement of the Northern Irish narrative would predict more forgiveness, whereas stronger ingroup identification would predict weaker endorsement of this narrative.
IDeNtItIeS, NaRRatIVeS aND FORGIVeNeSS 11 F I G U R E 1 The partially constrained multi-group mediation model in Study 1. Note: This figure presents the standardized regression coefficients (STDYX). The direct effect of ingroup identity as well as the effects of control variables were estimated and can be seen in Table 3. Dashed lines indicate non-significant effects. ***p < .001, *p < .05, b p < .10.

METHOD Participants
The study was conducted between 2019 and 2020 among participants 18 years or older and raised in Northern Ireland. Initially, 571 participants were recruited. After excluding those who did not identify as either group (n = 52), we focused on 519 participants who self-identified as either Catholic (n = 259) or Protestant (n = 260) (38% men, 61% women, 3 preferred not to say, two identified other; age range: 18-81, M = 31.97, SD = 15.16, around 4% less than high school, 52% high school, 44% university or higher). 1 Participants were recruited similarly to Study 1, except that in this study, we used both online (35%) and paper-pencil versions of the questionnaire. 2

Measures
Measures of ingroup identification (α for Protestants = .95; α for Catholics = .91) and forgiveness (α for Protestants = .86; α for Catholics = .83) were the same as Study 1. Independence and terrorism narratives were measured in the same way. In addition, we measured the Northern Irish Identity narrative, developed based on the literature (McNicholl et al., 2019;Trew, 1998) and again, discussions with lay people: "In my opinion, the Northern Ireland problem is in the past, and we should accept our past. To keep the peace in the region, we should move forward together to build a common Northern Irish identity."

Analysis strategy
The analysis strategy was the same as in Study 1, except that in addition to the independence and terrorism narratives, we also included the Northern Irish identity narrative as a mediator.

RESULTS
Descriptive statistics of the study variables are presented in Table 4. While Protestants endorsed the terrorism and Northern Irish identity narratives significantly more than the Catholics did, Catholics endorsed the independence narrative significantly more.
As seen in Table 2, when comparing the model fit across the unconstrained and fully constrained models, the fully constrained model significantly worsened the model fit (significant chi2, higher AIC and BIC values). Based on modification indices, we released the following regression coefficients to be estimated freely across groups: the effects of ingroup identification on the three narratives and forgiveness, and the effects of independence and terrorism narratives on forgiveness. Moreover, modification indices suggested that the following effects of the control variables should be freely estimated: the effects of gender on independence narrative and forgiveness, as well as the effects of age and education on the Northern Irish identity narrative (see Tables S1 and S2 for further details on model modifications). The partially constrained model had a good fit, as indicated by a non-significant chi2, RMSEA (<0.05), CFI (>0.90), and lower AIC and BIC values. Results of the partially constrained model can be seen in Table 5 and Figure 2.

Protestants
As in Study 1, Protestant identification was negatively associated with the endorsement of the independence narrative and positively with the terrorism narrative. It was not significantly associated with the Northern Irish identity narrative. Looking at which narrative predicts forgiveness, unlike Study 1, the independence narrative was not significantly associated with forgiveness despite a significant bivariate correlation between the two variables. 3 Like Study 1, a stronger endorsement of the dominant ingroup narrative-the terrorism narrative-was associated with less forgiveness. Finally, as hypothesized, the Northern Irish identity narrative was significantly associated with more forgiveness.
Looking at the indirect effects, only the negative indirect effect of ingroup identification on forgiveness via stronger endorsement of the terrorism narrative was significant, B = −0.06, SE = .03, t = −2.40, p = .017 (via independence narrative, B = −0.02, SE = .02, t = −.88, p = .378; via Northern Irish identity narrative, B = 0.00, SE = .01, t = .13, p = .895). Among the control variables, older participants endorsed all three narratives less while the higher-educated Protestants endorsed the Northern Irish identity narrative more. Younger participants and women were more willing to forgive.

Catholics
Like Study 1, Catholic identification was positively associated with the endorsement of the independence narrative, but it was not significantly associated with the terrorism narrative. Catholic identification was negatively associated with the endorsement of the Northern Irish identity narrative. Looking at which narrative predicts forgiveness, like Study 1, stronger endorsement of the independence narrative was associated with less forgiveness, while the effect of the terrorism narrative was not significant. As hypothesized, the Northern Irish identity narrative was associated with more forgiveness.
As for control variables, older participants endorsed the independence and terrorism narratives less, and women endorsed the independence narrative less. Younger and higher-educated participants were more forgiving. T A B L E 4 Correlations and descriptive statistics for variables in Study 2

DISCUSSION
As in Study 1, Study 2 offered support for our hypothesis that higher ingroup identification would be related to stronger endorsement of the dominant narrative for each group: Among Protestants, higher Protestant identification was related to stronger endorsement of the terrorism narrative whereas, among Catholics, higher Catholic identification was related to stronger endorsement of the independence narrative. Among Protestants, higher ingroup identification was again associated with less endorsement of the independence narrative, but the relationship between ingroup identification and endorsement of the terrorism narrative was not significant among Catholics. The negative bivariate correlation between ingroup identification and terrorism narrative among Catholics was not significant when taking into account the other narratives. The lack of significant relationship may be due to, in part, what this ingroup identity for Catholics entailed in terms of narratives. One can speculate that among Catholics it may not have been built based on the rejection of the terrorism narrative, but rather built on their independence, whereas, for Protestants, they may care more about rejecting the independence narrative. Protestant identification was not significantly associated with the endorsement of the Northern Irish identity narrative; however, Catholic identification was associated with less endorsement of the Northern Irish identity narrative. Similar to the findings of Study 1, among both groups, endorsement of the dominant ingroup conflict narratives (i.e., terrorism narrative for Protestants and independence narrative for Catholics) was associated with less forgiveness. The results related to the Northern Irish identity narrative supported our  hypothesis: Among both groups, stronger endorsement of the Northern Irish identity narrative predicted more willingness to forgive the outgroup members. The findings of Study 2 supported our indirect effects hypothesis as in Study 1. The indirect effects of ingroup identification on forgiveness were significant via (a) stronger endorsement of the terrorism narrative among Protestants and (b) stronger endorsement of the independence narrative among Catholics. Even though the indirect effects of ingroup identification on forgiveness through the mediating pathway of outgroups' conflict narratives were not significant across either group, the indirect effect of the Northern Irish identity narrative among Catholics was significant.

GENERAL DISCUSSION
The present research examined the role of ingroup identity and endorsement of both dominant and alternative conflict narratives on forgiveness among both Catholics and Protestants in the context of Northern Ireland. We used the conflict narratives that were developed in conflict contexts and investigated to what extent those conflict narratives are generalizable to post-conflict societies. The results of both studies showed that higher ingroup identification in an ethnopolitical post-conflict context is related to stronger endorsement of the dominant narrative of one's own group even after the conflict ends. Protestant/Catholic identity was associated with stronger endorsement of the terrorism/independence narrative, respectively. In addition to each group's own dominant narratives, the results also indicated that ingroup identification was also related to less endorsement of the other's narrative among Protestants, but F I G U R E 2 The partially constrained multi-group mediation model in Study 2. Note: This figure presents the standardized regression coefficients (STDYX). The direct effect of ingroup identity as well as the effects of control variables were estimated and can be seen in Table 5. Dashed lines indicate non-significant effects. *p < .05, **p < .01, ***p < .001. .21*** not among Catholics. Protestants endorsed the independence narrative less to the extent they identified with their ingroup. We also found that the negative correlation between ingroup identification and forgiveness is much higher among Protestants (−.53) than among Catholics (−.19). One possible explanation might be related to standard deviations and range restrictions, as standard deviations were larger among Protestants than among Catholics on both variables. In addition, while ingroup identification was relatively lower among Protestants than among Catholics, willingness to forgive among Protestants was higher than among Catholics. Our results are in line with previous studies which have shown the crucial role identities play in forgiveness, especially in post-conflict societies (e.g., Cehajic et al., 2008;Hanke et al., 2013;Hewstone et al., 2006;Leonard et al., 2015;Noor, Brown, Gonzalez, et al., 2008). However, our findings also extend these studies by showing how identities may be related to how groups understand, frame, and narrate conflicts even in post-conflict societies. For example, previous studies in post-conflict contexts have shown that conflicting parties are more likely to endorse their (ingroup) historical narrative to the extent they feel threatened because these narratives play an important role in maintaining a sense of continuity of their ingroup identity . Even though Uluğ and Uysal (2021) have shown how ingroup identities may predict endorsement of the minority group's conflict narrative (i.e., how both Turkish and Kurdish identities relate to the independence narrative), no single study has tested how ingroup identities may be related to both own and the other group's dominant narrative as well as alternative narratives, especially in post-conflict societies. The findings are in line with the observations of Bar-Tal and Oren (2000) that identities may help ingroup members see the conflict from their group's own perspective and endorse a narrative that revolves around this view.
In Study 2, we also introduced a new conflict narrative to reflect the discourse of today's political context in Northern Ireland. We focused on the Northern Irish identity and created a narrative that emphasizes this identity as a way to keep the peace in the country. The findings of Study 2 supported our hypothesis only for Catholics: Catholic identification was associated with less endorsement of the Northern Irish identity narrative; Protestant identification was not related to the endorsement of this narrative. We should also note that the correlation between the dominant Catholic narrative and the Northern Irish identity narrative only exists among Catholics. One possible explanation is that Northern Irish identity is an overarching identity, and the typical Northern Irish identity is generally associated with being a Protestant (McKeown, 2014). So, the lack of any association between the dominant Protestant narrative (terrorism) and the Northern Irish identity narrative among Protestants implies that these two narratives are not incompatible or are not mutually exclusive, whereas, among the Catholics, the dominant Catholic narrative (independence) is significantly negatively associated with the NI identity narrative, thus implying incompatibility. Our results complement these findings as we found a negative relationship between Catholic identity and endorsement of the Northern Irish identity narrative. Moreover, in terms of the levels of endorsement, Protestants endorsed this narrative significantly more than the Catholics did.
Across both studies, we also found that endorsement of the dominant ingroup conflict narratives predicted less forgiveness. In other words, both Protestants and Catholics were less willing to forgive the outgroup members for their wrongdoings to the extent they endorsed their dominant ingroup narrative (i.e., terrorism narrative for Protestants and independence narrative for Catholics). These results align with the previous studies that have shown that dominant conflict narratives may predict stronger negative attitudes towards peace and reconciliation (Uluğ & Cohrs, 2017a) and weaker support for non-violent conflict resolution and less willingness to forgive the outgroup (Uluğ, Lickel, et al., 2021). However, as previous studies have been conducted in conflict contexts, our findings extend these studies to post-conflict societies by showing that dominant conflict narratives may have consequences for forgiveness even in such contexts. For example, the conflict in Northern Ireland may have ended, but the way people understand it may still relate to whether one can forgive the perpetrator group for their misdeeds. More importantly, alternative narratives like the Northern Irish identity narrative are less prevalent and less visible in society than dominant narratives and therefore get less attention in media. Therefore, even though they have the potential to create a long-lasting peace process, they cannot contribute to peacebuilding processes enough (Uluğ & Cohrs, 2017a, 2017b, 2017c. However, it is those narratives that may pave the way for conflict resolution and peacebuilding. Our results confirm our hypothesis with respect to alternative narratives by showing that stronger endorsement of the Northern Irish narrative predicts more willingness to forgive the outgroup members across former conflicting parties. Finally, our indirect effects hypothesis was supported by the findings of both studies. The indirect effects of ingroup identification on forgiveness via dominant narratives for each group were significant. Even though higher ingroup identification predicted less forgiveness among both groups which was in line with previous studies (e.g., Odak & Čehajic-Clancy, 2021), we extended previous findings by showing that that is the case at least partially through stronger endorsement of the terrorism narrative among Protestants and stronger endorsement of the independence narrative among Catholics. In addition, indirect effects of ingroup identification on forgiveness were not significant via endorsement of outgroups' conflict narratives (except with respect to the independence narrative in Study 1). Based on these results, it is fair to argue that it is the dominant ingroup narratives that drive the effects of ingroup identification on forgiveness. However, the narratives of outgroups are less relevant for these effects to occur. When it comes to the indirect effects of ingroup identification on forgiveness through the alternative Northern Irish identity narrative, the results were significant only for Catholics, but not for Protestants. In line with McKeown (2014) observations, we reason that, compared to Protestants, identifying as Catholic might be related to rejecting the Northern Irish narrative, and therefore, we observe the indirect effects for this group, but not for Protestants.
For the first time, these results show that conflict narratives mediate the association between ingroup identity and willingness to forgive. In other words, the way people see conflicts can act as a process variable in this relationship. Although previous research has shown that the more you identify with one of the conflicting parties as your ingroup, the less you tend to forgive the other conflicting party (Cehajic et al., 2008;Hanke et al., 2013;Hewstone et al., 2006;Leonard et al., 2015;Noor, Brown, Gonzalez, et al., 2008;Van Tongeren et al., 2014), we extend these findings by showing how conflict narratives play a role in this relationship. One can say that if conflicting parties endorse a dominant conflict narrative, then the negative effects of strong ingroup identification on forgiveness can be exacerbated. However, if conflicting parties endorse an alternative conflict narrative that emphasizes a shared identity, then forgiveness can be possible.

Limitations and future directions
Our studies have a few limitations. First of all, even though we used religious group identification as an antecedent of intergroup-related outcomes, we do not know how both Catholic and Protestant participants construe their religious identity. Previous studies conducted in the context of the Alevi-Sunni conflict in Turkey showed that construing a religious identity differently-religious, political, or cultural identity-may have different outcomes for intergroup relations (Acar et al., 2022). Acar and colleagues have shown that construing Alevi identity as a religious or cultural identity is unrelated to forgiveness whereas construing Alevi identity as a political identity is related to less forgiveness among Alevis. As conflicting groups are rarely homogenous (Cohrs et al., 2015), yet tend to be categorized as if they are two homogenous groups, future research should take a more differentiated approach to examine variations within each conflicting group in Northern Ireland context (see Elcheroth & Spini, 2012;Reicher, 2004 for a discussion) and how these may affect people's behavioural outcomes.
Second, across both studies, we did not create the dominant narratives based on the Northern Irish conflict, but rather adapted those narratives based on the Turkish-Kurdish conflict to the Northern Irish conflict. In Study 2, we created a narrative to represent a more inclusive narrative that emphasizes the Northern Irish identity. Although the results show narratives that were developed in conflict contexts generalize to post-conflict contexts, these narratives were not developed in a bottom-up approach. Future research should employ a bottom-up approach using either a qualitative method such as interviews to understand different groups' conflict frames (see, e.g., Jelić et al., 2021;Moss, 2019;Neufeld & Schmitt, 2019;Twali, 2019;Uluğ et al., 2017 in different conflict contexts) or a mixed-method such as Q methodology to identify socially shared conflict narratives (see, e.g., Uluğ & Cohrs, 2016;Ünal et al., 2022).
Third, although we conducted two studies at two different time points, our studies were not longitudinal. Rather, we relied on cross-sectional data to conduct indirect effects analyses. However, it has been shown that cross-sectional data for indirect effects analysis has some shortcomings (see, e.g., Fiedler et al., 2011), leading to biased estimates even under ideal conditions (Maxwell & Cole, 2007). Moreover, in a cross-sectional study, associations between predictor and mediator/outcome variables might work both ways, as an indirect effect analysis does not prove causality. Testing for alternative models by changing the direction of associations is not statistically informative either (Lemmer & Gollwitzer, 2017;Thoemmes, 2015). In reality, it is possible that in a conflict context, while ingroup identities predict continued endorsement of dominant conflict narratives, being born and raised in such a context might have also shaped these identities. Therefore, future studies should use longitudinal research designs to be able to assess the effects of identification on peace-related outcomes.
Fourth, one can argue that the focus on the solutions of the conflict indicating group goals in these narratives (e.g., unity with Ireland) might in itself have influenced how our participants responded to forgiveness. For example, a group member might be more likely to forgive outgroup members if they believe they can achieve their conflict goals. Future studies should create different narratives that focus on these future goals as well as on the past and examine how these might influence the way participants forgive each other. While doing so, future studies may also employ different alternative narratives by creating solution strategies that are not close to one group more than the other (e.g., keeping Northern Ireland as it is vs. uniting it with the Republic of Ireland).
Last, we collected data from community samples with a cross-sectional design; however, we should note that the samples we present here are not representative and can only cautiously be interpreted to contextualize these different relationships. Compared to the population in Northern Ireland, our samples were highly educated and slightly younger. Therefore, one can ask to what extent the relationships between our study variables would hold for the general population. For example, one of the findings was lower endorsement of the terrorism narrative among more educated across both groups in Study 1. Although this finding did not reach significance in Study 2, one may argue that educated people may prefer peace over conflict (see, e.g., Hong, 2021). Moreover, higher education allows for more intergroup contact opportunities in Northern Ireland (Tausch et al., 2007), and intergroup contact, in turn, might help to endorse alternative narratives (Uluğ & Cohrs, 2017a). Although the findings of this research among educated younger people are still important for the resolution of the intergroup conflict in Northern Ireland, future studies should include both more and less educated segments of society. Even though we controlled for age, gender, and education, future studies should also consider the other relevant socio-demographic variables such as religiosity and political ideology. In post-conflict contexts where former conflicting groups do not share the same religion, there may be negative outcomes of religiosity on forgiveness and reconciliation due to its link to group centricity (Odak & Čehajic-Clancy, 2021). Therefore, the role of religiosity in contexts where there are religious sects in the same religion as well as its relation to forgiveness should be well understood.

Conclusion
This research investigates the perspectives of former conflicting groups in a post-conflict society. Our findings highlight the role of ingroup identification and endorsement of dominant conflict narratives as an obstacle to forgiveness among former conflicting parties. However, our results also highlight that endorsing an alternative narrative around a common identity may open the door to forgiveness. We hope that our studies contribute to the discussions on how to keep peace in post-conflict societies where a straightforward division between "ingroup" and "outgroup" is likely to survive, and narratives around those identities are still very alive.