Promoting social cohesion, compassion and forgiveness among Rwandan survivors and perpetrators through community resiliency model training: a longitudinal randomized comparison study

ABSTRACT

Introduction: The 1994 genocide against the Tutsi severely broke Rwandan society, leaving deep family and societal scars. Efforts to restore resilience and social cohesion hold promise but continue to face significant challenges. This study aimed to investigate whether the Community Resiliency Model (CRM) intervention was able to promote social cohesion, forgiveness, and compassion while strengthening personal resilience among survivors and released genocide perpetrators in post-genocide, Rwanda. We set out to address a methodological gap by exploring whether participants who experienced the Rwandan genocide do better in mixed groups versus homogeneous groups on key social function outcomes.

Methods: A longitudinal randomized comparison group study design was employed to evaluate the impact of CRM on social cohesion, compassion, resilience, and forgiveness in a convenience sample of three comparison groups to which participants were randomly assigned: one with genocide survivors (N = 51), one with genocide perpetrators (N = 51), and one with a combination of both (N = 50). In August 2023, participants completed a consecutive three-day Community Resiliency Model (CRM) training in community settings. Data were collected using standardized tools at baseline, immediately after the intervention, and at six-month follow-up, between August 2023 and March 2024. ANOVA was conducted to identify significant differences among outcome variables.

Results: The results showed that social cohesion (F(1) = 36.1, p < .001), compassion (F(1) = 131, p < .001), forgiveness (F(1) = 19.8, p < .001), and resilience (F(1) = 8.10, p = .005) significantly improved across all groups. Participants in the mixed group showed slightly more improvement than other groups, but the difference was not statistically significant.

Conclusion: The results are promising, suggesting that CRM could serve as an effective intervention to address the challenges of rebuilding cohesive and resilient communities by improving social functioning in societies that have experienced fragmentation due to societal traumas.

HIGHLIGHTS

• More than 30 years after the genocide in Rwanda, survivors and released perpetrators continue to struggle with trauma and broken social relationships, impacting community life.

• This study shows that the Community Resiliency Model can effectively enhance social cohesion, compassion, forgiveness, and resilience, whether implemented with only survivors, only perpetrators, or in mixed groups.

• All groups improved in social functioning over time, including the Mixed groups were both genocide survivors and perpetrators participated in the training together, promising the possibility of collective healing involving both survivors and released perpetrators, from Rwanda.

1. Introduction

1.1. Rwanda pre and during colonialism

Before colonialism, mutual support, social cohesion, compassion, and the request for forgiveness were traditional values cherished by Rwandans across the regions (Bigirumwami, 1987). These values were taught under traditional Itorero, which in modern terms can be considered an educational programme in traditional cultural practices. It was established in macro and micro societies, and individuals who deviated from these values were taught or punished in families to realign with the cultural values (Gierszewska & Sinining, 2023).

Between 1889 and 1900, Western European powers such as Belgium, France, United Kingdom, Germany, Portugal, Spain, and Italy established colonies in various places around the world, especially in Africa (Aptt, 2022; Races & Johnston, 1913; Vigouroux & Mufwene, 2008). The history shows that with the arrival of colonizers in different societies, many values and rituals that facilitated social functioning were abruptly changed (Sentama, 2022). The indigenous people and citizens started learning and assimilating the values and beliefs of the colonizers. This phenomenon, which has occurred worldwide, entails a variety of actions taken by nations for social and cultural dominance and, most importantly, economic gain, without regard for the people living in those countries (Diamond, 1999).

Through colonization, European powers intentionally transformed the social, economic, psychological, cultural, and political lives of African peoples and societies. In many cases, these changes remained in place even after the colonizers had departed (Ocheni & Nwankwo, 2012). In his book, Africa’s World War: Congo, the Rwandan Genocide, and the Making of a Continental Catastrophe, historian and sociologist Gerard Prunier discusses how, before colonization by Britain, France, and Germany, Africa was a continent with distinct socio-economic structures (Prunier, 1998) and systems of accountability. With the colonizers’ arrival, these structures were often changed in favour of rules that promoted the colonists' functioning over societal stability and usefulness. This occurred only when the colonizers brought divisiveness in their politics and made one group more favourable, while making another group from the same nation seem like enemies (Bizimana, 2022). For example, in pre-colonial Rwanda, social classes were primarily based on social roles and responsibilities within the community. These structures were generally accepted and contributed to social order and cohesion. With the arrival of Belgian colonialism, however, these classes were politicized and manipulated to serve colonial interests (Uwizeyimana, 2017).

Moreover, previous studies showed that the politicization of ethnicity, introduced and entrenched during colonization, severely weakened social cohesion and solidarity in Rwanda. This colonial legacy hindered the country’s ability to value diversity and sustain unity, giving rise to ethnic divisions and hostility that eroded traditional harmony (Bizimana, 2022; Jefremovas, 1997). Over time, these divisions, deepened by colonial influence, generated both direct and indirect consequences, fuelling political instability, repeated civil conflicts, and ultimately leading to the 1994 genocide against the Tutsi (Mamdani, 2001; Purdeková & Mwambari, 2021), in which over a million people were killed within just three months and left devastating long-term consequences (Fajth et al., 2019; Musanabaganwa et al., 2020).

1.1.1. Impact of genocide on trust and social cohesion

The 1994 genocide against the Tutsi in Rwanda had lingering consequences, not only on mental health but also eroding trust, weakening social cohesion, and diminishing compassion among Rwandans (Lordos et al., 2021). This atrocity was preceded by years of societal discrimination and fragmentation, civil wars, anger, culminating in mass violence that tore communities apart. During the genocide, the breakdown of unity was marked by extreme distrust, betrayal, and division along ethnic lines (Bigabo et al., 2025). An empirical investigation was conducted examining trust dynamics between different Rwandan survivor and perpetrator groups in the post-genocide context (Ingelaere & Verpoorten, 2020). Among 400 participants in the study, trust within the same group was relatively high, while trust between different groups was significantly lower. Qualitative interviews with survivors and perpetrators further illustrate prevalent mistrust and fear between the groups. For example, a 56-year-old female survivor expressed her misgivings about the perpetrators, stating, ‘I don't trust them. They are all the same. They could kill us again’ (Ingelaere & Verpoorten, 2020). In another interview, a 40-year-old male perpetrator, after serving his punishment, expressed his belief that the Tutsi survivors still harbour negative sentiments towards the released genocide perpetrators, saying, ‘I can't see what they carry in their hearts, but I'm certain they are not happy with us. They still hold onto ethnic ideologies and believe they are superior’ (Ingelaere & Verpoorten, 2020). These accounts are just two examples of the lack of social connection and mutual trust that continues to exist between post-genocide survivors and perpetrators. This resulted in a considerable gap in social connectedness, resilience, and community support among Rwandan families, and was observed in both the older and younger generations (Bigabo et al., 2025; Kagoyire et al., 2023).

1.1.2. Cultural strengths and post-genocide reconciliation efforts

In the aftermath of genocide, Rwanda’s Ministry of Health and the Ministry of Unity and Reconciliation and Civic Education (MINUBUMWE) decided to draw upon the inherent strengths in Rwandan culture to promote resilience, social cohesion, and well-being (Habimana & Biracyaza, 2023) by refocusing on concepts from its own culture and establishing government policies to help the community increase healing, resilience, and social cohesion in the aftermath of the genocide. For instance, Rwandans uphold cultural traditions like Gusurana (visiting each other) to advance social bonds and promote healing. Other Rwandan practices such as Gusangira (sharing), Gufashanya (helping), and Kwiyunga (reconciliation) were found to help rebuild trust, social cohesion, and resilience among community members (Otake & Tamming, 2021). In addition, cultural practices include Ibitaramo (cultural music and performance), Imikino (sports and games such as football), Kuzindukirana (visiting each other with baskets of food), government policies such as Umuganda (translated as community work or monthly community service day) have helped bring communities together, support social cohesion, and support post-genocide recovery in Rwanda (Dehrone et al., 2022; Habimana & Biracyaza, 2023).

At the national level, Rwanda established the Rwanda National Unity and Reconciliation Commission (NURC) in March 1999, whose mission was to use models and community and cultural-based initiatives to restore harmony and foster reconciliation (Sentama, 2022). The NURC has contributed much to helping Rwanda unite by coordinating platforms that aim to make the community live together. This programme is not used only in Rwanda but also in other countries that have experienced violence and civil wars, such as Burundi and Sierra Leone (Casey & Glennerster, 2016) and has been implemented in South Africa with 134 survivors and perpetrators of the apartheid system. The participants were exposed to South Africa's Truth and Reconciliation Commission (TRC), which aimed to help survivors and perpetrators learn to coexist. The findings revealed that female survivors were more forgiving of the perpetrators than male survivors. Furthermore, those who forgave also reduced their mental health problems (Bloomfield et al., 2018).

Rwanda also adapted the ‘I am Rwandan’ programme as a national policy initiative, encouraging the population to develop a shared identity as Rwandans, moving away from identifications as Tutsi, Hutu, or Twa ethnics (Collins et al., 2021). A qualitative study from 2017 explored the effects of reconciliation initiatives on genocide survivors and perpetrators in Rwanda, highlighting the potential of these programmes to unify society by reducing group distinctions and fostering social cohesion. Nevertheless, subsequent scholarly critiques have highlighted important limitations of this approach. For example, Blackie and Hitchcott (2018) caution that some practices of these initiatives may unintentionally generalize responsibility for the genocide across the entire Hutu population, including individuals who were not involved in the atrocities or children of perpetrators, thereby risking stigmatization and the reinforcement of latent social tensions (Blackie & Hitchcott, 2018).

These discussions highlight that social cohesion programmes are essential for community resilience and post-disaster recovery, while also showing the complexity of achieving it in post-conflict settings. Scholars such as Patel et al. emphasize that community resilience is a multidimensional and context-dependent construct, conceptualized as a dynamic process of adaptation, the capacity to maintain stable functioning, and consideration of social, political, environmental, and psychological qualities that enable communities to respond effectively to collective stressors (Ludin et al., 2019; Patel et al., 2017).

Building cultural and policy-driven strategies, diverse psychosocial programmes and interventions have been implemented for promoting healing and restoring community trust (Ministry of National Unity and Civic Engagement (MINUBUMWE), 2023). Among these, sociotherapy has been one of the most prominent approaches. Originating in Byumba in 2005 through collaboration between local communities, the Anglican Church of Rwanda, and Dutch practitioners, sociotherapy was built around the principle that sustainable peace requires deep social change. Implemented under the motto ‘Mvura Nkuvure’ (translated as ‘Heal me, I heal you’), the programme gradually expanded across several provinces, using six phases including safety, trust, care, respect, new life orientation, and memories to create safe group spaces where participants could share experiences, rebuild dignity, and regain a sense of belonging (Ingabire et al., 2017; Jansen et al., 2022; Richters et al., 2010). Evidence has demonstrated its effectiveness in reducing trauma symptoms, enhancing mutual support, and facilitating social cohesion (Biracyaza & Habimana, 2020; Kagabo et al., 2023). Additionally, a societal healing programme from Interpeace Rwanda [a non-governmental organization (NGO)], that includes mental health, psychosocial support, and livelihood components, has contributed to social cohesion, the rehabilitation and reintegration of genocide perpetrators, and the strengthening of community capacity for reconciliation (Lordos et al., 2021). However, while these community-based initiatives have significantly contributed to healing and reconciliation, they faced persistent challenges including limited geographical coverage, resource constraints, lack of multisystem solutions, and the inability to reach all individuals and communities in need (Lordos et al., 2021; Upreti, 2017). Survivors often continue to grapple with unresolved trauma, while released perpetrators struggle with stigma and reintegration (Bigabo et al., 2025). Addressing these dual challenges calls for interventions that concurrently promote individual healing and bolster compassion, forgiveness, and collective resilience.

To respond to these gaps, the Rwanda Resilience and Grounding Organization (RRGO) introduced the Community Resiliency Model (CRM) in 2015 (Habimana et al., 2021). CRM is a body-based, skills-oriented intervention that was adapted to address trauma and collective distress by helping individuals regulate their nervous systems and build resilience (Freeman et al., 2022; Miller-Karas, 2023). Initially implemented with genocide survivors, university students, Community Health Workers (CHW) and health professionals, CRM demonstrated promising results, including reductions in depression and anxiety as well as improvements in self-care and emotional regulation (Habimana et al., 2021). To date, CRM has reached approximately 3,400 people in Rwanda, but many affected individuals and regions have yet to benefit. Importantly, while survivors have been included in CRM programmes, perpetrators released from prison and mixed groups of both survivors and perpetrators who must coexist in daily community life have rarely been engaged in same intervention. To our knowledge, the role of CRM in affecting social functioning such as social cohesion, resilience, forgiveness, and compassion, which are essential for rebuilding trust and strengthening collective social well-being among survivors and released perpetrators, remains underexplored. Likewise, no study has yet examined whether CRM is more effective when delivered in homogeneous groups (survivors or perpetrators separately) or in mixed groups that mirror the realities of daily community life. Addressing these concerns is important because engaging homogenous groups of only either survivors or perpetrators in post-conflict contexts sometimes limit psychosocial healing and reconciliation because they do not address intergroup psychosocial needs and can lead to prejudice (Shnabel et al., 2009). For instance, survivor-only groups can reinforce trauma-focused narratives and collective victim identity, while perpetrator-only groups may restrict empathy development and acknowledgment of harm. In both cases, the lack of intergroup interaction can sustain social distance and in-group polarization. Thus, interventions that target all groups together could be encouraged in post-conflict seatings.

Building on these needs, this study aimed to evaluate the effectiveness of the CRM in enhancing social functioning, specifically social cohesion, resilience, forgiveness, and compassion among genocide survivors and released perpetrators in Rwanda. It further sought to determine whether CRM is equally effective when delivered within ‘like’ groups (survivors or perpetrators) or in mixed groups that reflect real-life community dynamics. Further, this study tested two hypotheses. First, we hypothesized that at baseline, there would be no significant differences in social functioning variables (social cohesion, compassion, resilience, and forgiveness) among the three groups. Second, we hypothesized that participants trained in the CRM would show statistically significant improvements in key social functioning variables (social cohesion, compassion, resilience, and forgiveness) six months after the intervention, and thirdly, we hypothesized that the effectiveness of the CRM training would be consistent regardless of the delivery modality, namely training groups of only survivors, versus only released perpetrators, versus mixed groups of both populations. By conducting this research, the results would inform policy makers to consider CRM as an alternative culturally adapted interventions aiming to promote social cohesion and resilience in post-conflicts settings.

2. Methods

2.1. Study area

This research was conducted in Nyamagabe District, which has an estimated total population of approximately 371,501 inhabitants in 2022. This study area is in the Southern Province of Rwanda. Nyamagabe is a rural district composed of 17 sectors, four of which Gasaka, Tare, Cyanika, and Kamegeri were randomly selected for the study. All four are rural, suburban and impoverished areas situated in the mountainous region near the Nyungwe natural forest. According to the National Institute of Statistics of Rwanda (National Institute of statistics of Rwanda N, 2023), 35% of the population in this district is married and 44% are unmarried. The population shares similar socio-cultural characteristics including using the Kinyarwanda language as their mother-tongue, geographical locations, local lifestyle and social interactions, experiencing the same histories of genocide, and experience comparable socio-economic activities (e.g. farming as primary means of livelihood). Nyamagabe is considered the poorest district in Rwanda, according to a 2025 report by the Government of Rwanda. We selected this district as our study area because it was one of the areas most affected by the genocide. It is one of the districts where recent reports have also documented the persistence of genocide ideologies, ethnic tensions, and relatively low levels of reconciliation compared to other districts in the country, making it a critical setting for interventions aimed at supporting social cohesion and psychosocial healing (Ministry of National Unity and Civic Engagement (MINUBUMWE), 2023).

2.2. Study design and participants

This study employed a randomized longitudinal comparison group design to evaluate the effectiveness of CRM trainings in improving key dimensions of social functioning, including resilience, social cohesion, compassion, and forgiveness, after 31 years since the genocide occurred in Rwanda. The study also examined if these outcomes differed depending on the modality of CRM delivery in three different comparison groups: survivors of genocide alone, perpetrators alone, and mixed groups of both survivors and perpetrators. We measured our participants at three time points: pre-intervention (baseline), immediate post-intervention (after the three-day training), and six months post-intervention (follow-up). The primary outcomes were the changes in resilience, social cohesion, forgiveness and compassion over time across all groups.

Participants were randomly divided into three groups: genocide survivors, genocide perpetrators, and a mixed group consisting of both survivors and perpetrators. Before beginning the recruitment process, consultations were held with local leaders in the Nyamagabe District, who provided the names of genocide survivors and perpetrators who might wish to participate in this research. The inclusion criteria were being either a genocide survivors or released perpetrators from Nyamagabe District, living in the selected villages, having resided in Rwanda during the genocide, and having willingness to participate in the study. The exclusion criteria were being born after the genocide or not yet released from prison due to a genocide conviction or unable to provide either oral or written informed consent. The study was conducted between August 2023 and March 2024 to address the psychosocial effects persisting approximately 31 years after the genocide. The participants were required to provide free, informed, and signed consent. However, for participants with low literacy, trained data enumerators read the informed consent form aloud in the participants Kinyarwanda (native language), ensured comprehension, and obtained verbally informed consent prior to participation.

According to the report of the National Commission for the Fight against Genocide (CNLG), a genocide survivor is a Rwandan who was present in the genocide and was specifically targeted with the intent to be killed National Commission for The Fight against Genocide (CNLG, 2020). Genocide survivors are those who lived through and survived the 1994 genocide against the Tutsis, despite being directly targeted for extermination or losing family members during the atrocities. They belong to the Tutsi ethnic group that was the main target of the genocide. In Rwandan contexts, a released perpetrator refers to a person convicted of committing crimes during the 1994 genocide against the Tutsi, who has since completed their sentence or been released back into the community through pardon or other judicial mechanisms (National Commission for The Fight against Genocide (CNLG) 2021; Sullo, 2018).

2.3. Data collection and measurements

At baseline, socio-demographic data such as age, gender, marital status, education, household size, employment status (with or without wages), religion, food security concerns (worrying about food or running out of food), perceived safety, and access to electricity at home. We used the following psychometric instruments:

2.3.1. Brief resilience scale (BRS)

The BRS is a psychometric instrument, which consists of six items that assess an individual’s resilience or their capacity to recover from stress and adversity. It employs a 6-point Likert scale ranging from 1 (Strongly Disagree) to 5 (Strongly Agree), with total scores ranging from 6 to 30, and higher scores indicating greater resilience (Ye et al., 2022). The reliability, validity, and factor structure of the scale were evaluated, and reference scores were established based on recent studies indicating high internal consistency within Rwandan communities (Nicely, 2019); The Cronbach’s alpha in the current study was excellent (α = 0.89) indicating that it is very reliable.

2.3.2. Santa Clara Brief Compassion Scale (SCBCS)

The SCBCS is a 5-item shortened version of the original 21-item Compassion Love Scale developed by Sprecher and Fehr and later evaluated by Hwang and his colleagues (Novak et al., 2021; Plante, 2022). It measures compassion-related responses to others’ suffering on a 7-point Likert scale, where 1 indicates complete non-identification with compassionate feelings and 7 indicates a high degree of compassion with total scores ranging from 7 to 49, and higher scores indicating greater compassion. Its internal reliability and validity were excellent (Cronbach’s alpha; α = 0.91). The SCBCS scale is very reliable.

2.3.3. Social cohesion scale (SCS)

The SCS is a psychometric instrument that measures the level of social connectedness and solidarity and cohesiveness of people (community) living together (Li et al., 2015). It is an 18-item tool that measures levels of social cohesion and trust and sense of belonging among participants. Responses are assessed on a 5-point scale from 1 (Strongly Disagree) to 5 (Strongly Agree), with total scores ranging from 18 to 90, and with higher scores indicating greater social cohesion. Internal reliability and validity were high with a good internal consistency (Cronbach’s alpha, α = 0.89).

2.3.4. Forgiveness reconciliation inventory (FRI)

The FRI is a psychometric instrument that was used to assess aspects of forgiveness in the context of reconciliation (Fajth et al., 2017). The inventory is comprised of 24 items that explore dynamics such as Collaborative Exploration, Role of Reconciliation, Remorse/ Change, and Interpersonal/ Intrapersonal Forgiveness. We used the short version with 8 items. Responses are assessed on a 5-point scale from 0 (Strongly Disagree) to 4 (Strongly Agree), with total scores ranging from 0 to 32, and with higher scores indicating greater forgiveness. The development of this inventory was inspired by the Forgiveness and Reconciliation Model, which draws from Jewish perspectives on forgiveness (Balkin et al., 2014). The psychometric properties of this measure were assessed and indicated excellent internal consistency (Cronbach’s alpha, α = 0.93).

Prior to data collection, all instruments were translated from English into Kinyarwanda (native language of participants), by bilingual professionals trained in CRM. The translation process included back-translation to ensure that questions were relevant, understandable, and culturally appropriate, and all items from the original scales were retained. After translation, the survey was programmed and deployed on the KoboCollect platform by an independent, trained data engineer, who then shared the questionnaire link and access with the data collectors to begin data collection. It was conducted by five clinical psychologists and social workers who had experience in data collection from previous assignment; however, they did not participate in delivering the CRM training sessions in this research. Under coordination of the research team, they collected data at baseline, immediately after the intervention, and six months post-intervention. Data enumerators completed refresher training on questionnaires as well as a session on ethical considerations. The duration of the self-administered survey ranged between 30 and 45 min, and each participant received a 2 USD for transportation facilitation no additional per diems were provided.

2.4. Description of the intervention: community resiliency model (CRM) skills

The intervention used in this study is the CRM skills training, a biologically-based approach that builds on managing the autonomic nervous system (ANS) and its responses to stressful situations (Aréchiga et al., 2024; Grabbe & Miller-Karas, 2018; Miller-Karas, 2023). It seeks to help participants better manage stress and restore mental, physical, and spiritual well-being by training them in six wellness skills after a traumatic experience (Miller-Karas, 2015). CRM training is teaching-based, where CRM trainers guide and invite participants to learn the model. The training combines both teaching using didactic and practice-based modalities, allowing participants to learn, apply the skills through guided, practical exercises, thus developing skills to apply the model for themselves and to support their family members. During the training, the trainers showed the participants, with examples, how they can use the model to help themselves and others. The intervention was adapted to the Rwandan context and translated into Kinyarwanda by local CRM teachers, psychotherapists, and social workers who are both CRM trainers and mental health professionals. Further, the CRM trainings were delivered as community-based participatory sessions. Although this intervention involved structured teaching, it encouraged active engagement of participants, and reflection which allowed local needs, lived experiences, and community priorities to shape the learning process (Grabbe et al., 2023). As an approach that aims to promote individual and community ability based on local and collective strengths, CRM has also been previously implemented among genocide survivors and shown promising impact in helping to decrease mental health- and trauma-related problems in their daily living activities (Habimana & Biracyaza, 2023). Trainers were working with RRGO, the Prison Fellowship Rwanda,¹ and as high school teachers. All sessions of CRM training were delivered in Kinyarwanda. The models’ six core skills that are explained in Table 1.

Table 1.

Community Resiliency Model (CRM) skills.

CRM skills	Meaning	Application
Skill 1. Tracking (Gukurikirana)	Involves paying attention to sensations, noticing current sensations, and identifying their connections with parts of the body.	This skill was taught before any other skills were discussed but was reiterated during the teaching of other skills. It is the centre of the model and is also used during the practice of other skills for adapting and recovering from negative sensations to strengths/positive sensations or neutral sensations.
Skill 2. Resourcing (Ubufasha)	Anything that provides joy, happiness, and a restoration of positive emotions.	Participants learn about different resources in Rwandan culture, identify them, and practice using resources that are associated with their way of living. These are reported to be relevant and easily usable in the Rwandan context.
Skill 3. Grounding (Kuba mugihe cya none)	The ability to be in the present moment and connect one’s body with the ground/reality.	Participants learn various forms and methods of grounding, benefitting from sharing their sensations and outcomes.
Skill 4. Gesturing (Amarenga)	Positive, spontaneous movements that guide and are associated with present sensations.	In this skill, participants learn different ways to use and recall gestures that surround them such as Rwandan gestures for explaining comfort, triumphs, and positive energy.
Skill 5. Help now (Fasha none)	Different strategies that help enhance the present moment, especially when stress arises.	Trainees learn ‘Help Now’ strategies, and everyone practices exercises made up of these skills. Surprisingly, during this skill training, attendees said these skills reminded them of their nuclear families’ way of living.
Skill 6. Shift and stay (Hindura erekeza)	The ability to track and shift from negative sensations to the present, or at least neutral sensations. Participants learn to use this skill to shift from stress to present sensations; if one skill does not accomplish this, they are encouraged to move to other skills.	This skill was taught after other skills, helping to memorize and practice for adapting and recovering from negative sensation to positive sensation. During the trainings, participants were encouraged to shift and use other skills like tracking, resource, grounding Help Now and gesturing skills.

CRM intervention was delivered to six groups of 25–26 participants comprised of separate groups of survivors (51 participants) and perpetrators (51 participants) and combined group of survivors and perpetrators (50 participants) residing in the same neighbourhood. The training was conducted over 3 consecutive days, for 7 h per day. The sessions were conducted in safe locations within the participants’ immediate living environments, such as schools, churches, offices, private spaces, or outdoor areas such as a grassy field or under a tree. The trainings were delivered by CRM teachers (one lead instructor and three facilitators) certified by the Trauma Resource Institute (Miller-Karas, 2023).

2.5. Statistical analysis

Data collected were analysed using the Statistical Package for the Social Sciences (SPSS) version 29 (Stehlik-Barry & Babinec, 2017). After data cleaning, descriptive statistics were conducted including frequency and percentages for categorical data, means and standard deviations (M, SD) for continuous variables. Regarding inferential analyses, all scale scores were computed following published guidelines, and internal consistency was assessed using Cronbach’s alpha. Values above 0.9 were considered excellent, and those above 0.8 indicated strong reliability. Greenhouse-Geisser correction was applied to adjust for violations of sphericity. We used one-way ANOVA to demonstrate the mean difference in the outcome variable for baseline data. We were also interested in whether the CRM intervention affected changes in means over time across the outcome variables. Therefore, we conducted repeated-measures ANOVA to examine mean differences across pre-evaluation, immediate post-evaluation, and 6-month post-evaluation in social function variables. Bonferroni post hoc tests were run to identify group differences. Lastly, the significance of variables was confirmed through ANOVA, using a p-value threshold of less than 0.05 and 95% for confidence intervals.

3. Results

3.1. Participants characteristics

Table 2 summarizes the baseline demographic characteristics of participants, categorized into three groups: survivors (n = 51), perpetrators (n = 51), and a mixed group (n = 50). There was no drop out among the participants. The overall mean age was 55.62 years (SD = 12.6), with participants ranging from 30 to 75 years. Although the majority across all groups was male, the survivor group was an exception, with females constituting 62.7%. Marital status patterns varied notably, with 54.9% of survivors being married, while 78.4% of perpetrators were widowed. Regarding education, most participants in all groups attended secondary level with 47.1% for survivors, 43.1% for perpetrators and 48% for the mixed group. Food insecurity was evident, particularly among perpetrators, of whom 60.8% reported frequently running out of food, compared to 35.3% of survivors. While most participants reported feeling safe, 7.8% of survivors indicated they never felt safe. Household sizes were typically large, with over 40% of both survivors and perpetrators residing in households with more than five members. In addition, the predominant livelihood, with over 80% of participants in each group identifying as farmers and their predominant domination is mainly Catholic and Protestant, with 56.9% of perpetrators identifying as Catholic and 51.0% of survivors as Protestant. Marked differences were observed in access to electricity whereby only 27.5% of perpetrators reported having electricity at home, compared to 52.9% of survivors.

Table 2.

Baseline demographic characteristics of CRM intervention participants by three targeted groups (survivors or perpetrators only and mixed groups of survivors or perpetrators).

Variables	Survivors (51) Number (%)	Perpetrators (51) Number (%)	Mixed Group (50) Number (%)
Gender
Female	32(62.7)	12(23.5)	18(38.0)
Male	19(37.3)	39(76.5)	32(64.0)
Age (Mean, SD)	53.16(11.79)	59.75(12.16)	53.96(13.49)
Marital Status
Widower	23(45.1)	40(78.4)	33(66.0)
Marriage	28(54.9)	10(19.6)	17(34.0)
Single	0	1(2.0)	0.
Education
Non-formal	11(21.6)	13(25.5)	16(32.0)
Primary	4(7.8)	7(13.7)	6(12.0)
Secondary	24(47.1)	22(43.1)	24(48.0)
Technical	6(11.8)	8(15.7)	2(4.0)
University	6(11.8)	1(2.0)	2(4.0)
Worried about food
Never True	3(5.9)	2(3.9)	3(6.0)
Sometimes True	37(72.5)	25(49.0)	23(46.0)
Often True	11(21.6)	24(47.1)	24(48.0)
Running out of food
Never True	3(5.9)	1(2.0)	3(6.0)
Sometimes True	30(58.8)	19(39.3)	21(42)
Often True	18(35.3)	31(60.8)	26(52.0)
Feeling safe
Never True	4(7.8)	0	2(4.0)
Sometimes True	17(33.3)	14(27.5)	21(42.0)
Often True	30(58.8)	37(72.5)	27(54.0)
Number of People in Household
Alone	2(3.9)	4(7.8)	4(8.0)
Three people	14(27.5)	11(21.6)	15(30.0)
Four people	11(21.6)	14(27.5)	12(23.0)
More than 5 people	24(47.1)	22(43.1)	19(38.0)
Employment
Unemployed	5(9.8)	2(3.9)	2(4.0)
Employed with a wage	2(3.9)	2(3.9)	5(10.0)
Farmers	41(80.4)	44(86.3)	41(82.0)
Retired	0	2(3.9)	0
Self-employed	3(5.9)	1(2.0)	2(4.0)
Religion
No religion	1(2)	0	0
Catholic	19(37.3)	29(56.9)	26(52%)
SDA	2(3.9)	2(3.9)	1(2.0)
Protestant	26(51.0)	18(35.3)	19(38.0)
Others	3(5.9)	2(3.9)	4(8.0)
Electricity in home
Yes	27(52.9)	14(27.5)	29(58.0)
No	24(47.1)	37(72.5)	21(42.0)

Note. SDA: Seventh Day Adventist; SD: Standard deviation; M: Mean.

3.2. Baseline differences in social function outcomes, comparing the groups trained in the CRM

Our findings showed how participant groups differed on variables of interest at baseline using one way ANOVA. The results showed no statistically significant differences across the groups for resilience, F(2) = 0.97, p = .37; compassion, F(2) = 2.11, p = .125; and forgiveness, F(2) = 1.15, p = .317. However, for social cohesion we found that groups presented with statistically significant differences with F(2) = 3.59, p = .030 at baseline (Table 3).

Table 3.

Baseline Comparisons of social functioning variables across the groups.

Social function variables	Source	Mean Square	DF	F	p-value	ηp2
Resilience	Three groups	0.52	2	.27	.759	0.04
Social cohesion	Three groups	446.63	2	3.59	.030*	0.49
Compassion	Three groups	63.72	2	2.11	.125	0.12
Forgiveness	Three groups	75.12	2	1.15	.317	0.16

Notes: DF: Degrees of freedom; F: F-test; ηp²: partial eta squared.

To examine differences in social cohesion across the three groups, Bonferroni post hoc test tests were conducted. Results indicated that participants in the perpetrators-only group reported significantly higher social cohesion than those in the survivors-only group (mean difference = 3.36, SE = 1.33, p = .038, 95% CI [0.13, 6.60]). No significant differences were found between the mixed group and the survivors-only group (p = .143) or between the mixed group and the perpetrators-only group (p = 1.00) at baseline.

3.3. The impact of CRM for social function variables over time

The results of repeated-measures ANOVA to evaluate changes in resilience, social cohesion, compassion, and forgiveness at pre-test, immediate post-test, and 6 months post-test. We compared these changes across the 3 groups. The results indicated that social cohesion, F(2) = 36.19, p < .001, with Partial Eta Squared (ηp²) is 0.20(20%); compassion, F(2) = 131.6, p < .001 with Partial Eta Squared (ηp²) is 0.48(48%) and forgiveness, F(2) = 19.87, p < .001, with Partial Eta Squared (ηp²) is 0.12(12%) statistically improved over time across all groups. Also, resilience was statistically significant, with F(1) = 9.10, p = .005, with Partial Eta Squared (ηp²) is 0.054 (5.4%). Bonferroni post hoc test was employed to evaluate whether there were any significant differences across groups. The results revealed that there were no significant differences among the survivor, perpetrator, and combined survivor–perpetrator groups (Table 4).

Table 4.

Repeated measures ANOVA comparing social function changes over 6 months between the three groups of participants.

Social function variables	Source	Mean Square	DF	F	p-value	ηp2
Resilience	Time	1.55	1	8.10	.005*	0.05
Social Cohesion	Time	3517.81	1	36.19	<.001*	0.20
Compassion	Time	3972.77	1	131.60	<.001*	0.48
Forgiveness	Time	401.14	1	19.87	<.001*	0.12

Notes. DF: Degrees of freedom; F: F-test; ηp2: partial eta squared, Times: Evaluation times.

3.4. Changes in social functioning dimensions over time by delivery modality

The results showed that compassion improved across all groups over time. Perpetrators began with lower mean score than survivors (21.64) but increased steadily, reaching 29.24 at six months. Survivors also demonstrated consistent gains rising from 22.34 at baseline to 27.93 immediately post intervention, with a slight increase to 28.07 at six months. The mixed group showed the largest improvement, starting from the lowest baseline score (20.82), increasing to 27.0 immediately post intervention, and reaching 29.86 at six months (Figure 1).

Figure 1.

Impact of CRM skills on compassion over time.

3.4.1. Ts

Overall, CRM training was associated with improvements in resilience across all groups, although the magnitude and stability of change varied over time (Figure 2). The figure depicts the different patterns of resilience across participant groups, following the CRM training. Genocide survivors revealed the strongest immediate response, with resilience scores increasing from 3.36 at baseline to 3.44 post-intervention, followed by a modest decline to 3.30 at six-month follow-up. Perpetrators exhibited slightly more moderate improvement, with scores rising from 3.23 at baseline to 3.61 post-intervention and decreasing to 3.29 at six months. In contrast, the mixed group showed a more stable continuing pattern, with resilience increasing from 3.10 at baseline to 3.35 post-intervention and remaining relatively sustained at 3.31 at six months.

Figure 2.

Impact of CRM skills on resilience over time.

The results also showed significant improvements in social cohesion following CRM training. Survivors’ scores increased from 58.52 at baseline to 66.80 immediately post-training, remaining stable at 66.74 at six months. Perpetrators exhibited the sharpest short-term gain, rising from 61.88 at baseline to 74.86 immediately post-training, followed by a decline to 67.41 at six months. The mixed group also showed substantial improvement, increasing from 59.00 at baseline to 74.36 immediately post-intervention, and stabilizing at 67.74 at six months (Figure 3).

Figure 3.

Impact of CRM skills on social cohesion over time.

Our findings indicated that forgiveness followed a pattern of steady improvement across all groups, though the trajectories varied. Survivors showed a sharp early increase from 20.98 at baseline to 24.23 immediately post-training, followed by a slight decline to 22.92 at six months, which remained above baseline. Perpetrators demonstrated the greatest gains, rising from 20.36 at baseline to 29.35 immediately post-training, and maintaining high levels at 28.90 after six months. The mixed group also improved, increasing from 20.36 at baseline to 27.00 immediately post-intervention, and sustaining a score of 23.10 at six months (Figure 4).

Figure 4.

Impact of CRM skills on forgiveness over time by participant group.

4. Discussion

This study offers preliminary support for the first hypothesis, which stated that at baseline, there would be no significant differences in social functioning variables specifically compassion, resilience, and forgiveness among the three groups. On this hypothesis, our baseline analyses revealed no significant differences in resilience, compassion, or forgiveness across groups. So, the findings revealed low compassion, forgiveness, and resilience among the three groups. These findings are align with previous research conducted in post-conflict settings, which has shown that both survivors, and offenders often continue to struggle with complex psychosocial problems comprising mistrust, unsafe, psychosocial wounds, unresolved trauma, stigma, and shame especially when adequate interventions have not been implemented (Habimana et al., 2023; Somasundaram & Sivayokan, 2013). The preceding studies also stated no significant differences between these groups in their willingness to co-exist and engage in joint community or livelihood activities (Lordos et al., 2021).

Regarding the component of social cohesion, our results indicated relatively low levels of social cohesion across the three groups. Post hoc analysis revealed a significant difference between the group composed solely of genocide survivors and released perpetrators, with survivors exhibiting lower levels of social cohesion at baseline. These differences could be due to deep psychosocial wounds, especially post-traumatic disorders, depression, loss of social networks, and prolonged grief experienced by survivors which could reduce trust, mutual assistance, and openness towards others (Habimana et al., 2023). This result may be explained through previous trauma research, which highlights the psychological consequences for victims, including heightened sensitivity to threat and difficulties with social trust and connection (Staub et al., 2005). As Staub (2014) notes, survivors often struggle to balance their own emotional needs with those of others in post-conflict environments, leading to reduced perceptions of social connectedness (Staub, 2014).

The second hypothesis, that the CRM would enhance social functioning dimensions regardless of intervention modality cohesion, compassion, resilience, and forgiveness among survivors and perpetrators of the Rwandan genocide regardless of the intervention modality (survivors only, perpetrators only, or mixed groups), was also supported. Participants across all groups demonstrated statistically significant improvements in social cohesion, compassion, resilience, and forgiveness from baseline to six months post-intervention. These results align with previous group-based interventions in Rwanda, where survivors and perpetrators engaging in community-based activities demonstrated increased altruism, social reintegration, and a renewed sense of belonging (Bigabo, 2025; King, 2019). Sociotherapy similarly supports empathy, social responsibility, emotional healing, and relationship restoration among genocide survivors and released perpetrators (Jansen et al., 2022; Ubels et al., 2025).

In line with the objectives of our study, we further examined whether the effects of CRM training would vary according to group composition. This third hypothesis anticipated participants completed CRM training would experience similar benefits across delivery modalities, whether the training was conducted with survivors, released genocide perpetrators, or mixed groups of both. The results confirmed this assumption, showing no significant differences in the improvement of resilience, social cohesion, forgiveness and compassion among the three groups. Importantly, this study contributes novel evidence by comparing different delivery modalities, demonstrating that CRM can effectively improve social functioning across survivors-only, perpetrators-only, and mixed groups. Improvements in resilience and social cohesion showed slight declines between immediate post-intervention and six-month follow-up assessments, which may be attributable to the timing of data collection during Rwanda’s national genocide commemoration period, a time known to amplify trauma symptoms (Kayiteshonga et al., 2023). Other longitudinal studies similarly show that intervention effects peak immediately after training and may decrease over time, highlighting the potential benefit of booster sessions (Doyle et al., 2023; Verduin et al., 2014). Nevertheless, all social improvements remained statistically significant compared to baseline, indicating sustained CRM effects.

The mechanisms underlying these improvements appear linked to CRM’s body-centred and skills-based approach. First, Tracking Skills enhances sensory awareness and emotional regulation, promoting well-being and togetherness among participants (Grabbe & Miller-Karas, 2018; Miller-Karas, 2018). Another skill that contributes to social cohesion and reshaping perceptions among participants is resource identification, which focuses on amplifying elements in life that bring joy, comfort, peace, strength, and hope for the future (Miller-Karas & Citron, 2013). These Resources skills may use the image and memories of loved ones, personal beliefs, or supportive figures (Grabbe et al., 2019; Grabbe & Miller-Karas, 2017). During training, many participants expressed surprise at discovering that survivors and perpetrators shared similar Resources. For example, combined groups often cited faith in God as a vital source of support, enabling them to navigate their challenging experiences and fostering a renewed sense of community within their neighbourhoods. According to the CRM developer, identifying resources could be helpful for participants even in cases where distress arises due to the absence of resources (Miller-Karas, 2023). For example, survivors who had lost loved family members often mentioned them as a resource, although they were killed during the genocide. Through practicing this skill, participants were encouraged to recall positive memories or words of wisdom shared by their beloved ones, which served as a source of strength. One participant described how following his late father’s advice gave him purpose and resilience. Resource identification proved potential in promoting social cohesion and forgiveness, as participants reported a renewed ability to view others, whether survivors or perpetrators, as created in the image of God rather than as adversaries.

Our intervention also includes Grounding skills, which focus on ability of participants to remain present and connected in the current moment (Baek et al., 2024; Neal, 2024). We thought this skill could enable participants to separate themselves from traumatic memories and decide to focus on that which provide feelings of happiness and calm (Miller-Karas, 2023). The final component of our intervention, Gesturing, Help Now and shift and stay skills, that were reported by many participants to provide them immediate, practical strategies for managing acute stress, allowing participants to restore balance and well-being when feeling overwhelmed (Miller-Karas, 2023). Practicing these skills helped both survivors and perpetrators find new perspectives on living together, strengthen their existing social connections, and advance their healing journeys.

The findings from this study provided promising evidence that CRM is an effective approach for fostering social functioning components mainly resilience, social cohesion, compassion, and forgiveness. In addition, at baseline, the levels of social cohesion significantly differed across the three groups. The data showed that the perpetrators-only group reported higher social cohesion compared to the survivors-only group and the mixed group. After the intervention, the data showed that all three groups experienced similar improvements in social cohesion. These results aligned with the studies indicating that CRM interventions effectively improve individual and collective wellbeing, and enhance emotional regulation and interpersonal connections among persons affected by the conflicts (Habimana et al., 2021). Additionally, among the variables measured, compassion consistently improved across all evaluation points, showing a stronger improvement compared to other variables such as social cohesion, resilience, and forgiveness. This pattern was observed across all three groups. A possible explanation for this continued improvement over time (while other variables declined somewhat) is those participants seemed to have learned and applied the CRM intervention skills, which enhanced their emotional regulation and the ability to remain present in the moment, which is foundational to cultivating compassion. Indeed, the direct application to their relationship challenges with the other parties was repeatedly discussed during trainings and became a focus, and even observationally, participants softened in their attitudes and fears toward the other party seemed to decrease. These results reflect the productivity of participants jointly practicing and applying CRM skills, as well as the effectiveness in restoring balance to the nervous system, which might in turn fosters compassion, social cohesion, forgiveness and resilience, and overall mental well-being (Grabbe et al., 2023). Furthermore, CRM has successfully been delivered in many African countries, including Kenya, Sierra Leone, Nigeria, and Angola, where it documented its cultural adaptability and strong potential for use in other post-conflict contexts (Aréchiga et al., 2023). Its ability to strengthen social functioning and rebuild trust indicates its prominence in precluding recurrence of violence and boosting social bonds among victims and convicts, within both Africa and beyond.

Taken together, the results from this research suggest that heterogeneous (mixed) groups are not inferior to homogeneous groups and may even offer added benefits for certain outcomes, particularly compassion, social cohesion, and the stability of resilience when potential risks are carefully managed during session facilitation. During all trainings, participants were treated equally and encouraged to actively engage in practicing the model. At the same time, the effectiveness of group composition appears to vary slightly by outcome, indicating that mixed and homogeneous formats are best viewed as complementary approaches rather than universally superior models. These results corroborate previous studies that showed that heterogeneous group of offenders and survivors showed higher social resilience and more favourable attitudes towards out-groups after conflicts and intercommunal violence than homogenous (Bigabo et al., 2025; Hewstone et al., 2014).

4.1. Study strengths and limitations

Our study demonstrated several key strengths. First, it is the first to implement a biologically informed CRM intervention bringing together genocide survivors and perpetrators, which promotes sustainable collective healing and strengthen social cohesion at both individual and community levels. Second, it holds significant importance in a country like ongoing post-genocide social reconstruction of Rwanda by enhancing social cohesion, forgiveness and compassion. It provides an intervention that supports a sense of unity, thereby contributing to long-term prevention of ethnic divisions and future conflicts. Third, this study is novel in evaluating the long-term impact of CRM skills, with outcomes assessed immediately post-interventions and at six-months follow-up, providing evidence of sustained effects on psychosocial and relational functioning. Finally, the study strengthens methodological rigour by employing a randomized longitudinal comparison group design, which allows for robust evaluation of effectiveness of the intervention across different groups of participants and delivery modalities.

Despite its strengths, the study has several limitations. First, the sample size was relatively small, highlighting the need for future research with larger participant groups to validate the findings. Second, while practicing communities were randomized to groups and participants randomized once they were recruited, overall participant recruitment was still utilizing a convenience sample design, leaving the study not fully randomized. Third, we did not include a control group, which limits the ability to draw causal conclusions about the intervention’s impact on the outcome variables. Fourth, participants were recruited using a list provided by local leaders, which may introduce selection bias. However, this method was also a practical way to identify genocide survivors and perpetrators, as they are often registered through local leaders for other services they receive. Fifth, the 6-month post-intervention data were collected in March, one month before Rwanda’s national commemoration period, during which trauma and stress are known to increase (Gishoma et al., 2015). This timing may have introduced biases or errors due to remembrance-related stress. Sixth, the study was conducted in a single district, predominantly rural, out of Rwanda’s 30 districts, which limit the generalizability of the findings to other regions with potentially different socio-cultural or urban contexts. Lastly, the evaluation relied exclusively on quantitative measures; the absence of qualitative data restricted our ability to capture lived experiences of participants and provide deeper insights into the mechanisms and nuanced impacts of the intervention.

5. Conclusion

The CRM intervention appears to be effective in promoting resilience, compassion, enhancing social cohesion, and promoting forgiveness among participants taught in groups of survivors and perpetrators, but also when taught in a mixed group format that included both. This model holds promise for implementation in other populations that have experienced conflict, potentially contributing to the restoration of social functioning and the rebuilding of fractured communities. Based on our results, we have the following implications. First, the implications for future research can be strengthened by replicating this model across diverse contexts such as different geographic regions, conflict types, active conflict settings and among various populations. Such replication would help determine the mechanistic underpinnings of the model and assess its long-term effects. Second, the CRM could be integrated into community-based reconciliation programmes to complement existing psychosocial, peace-building, and social healing initiatives. Future studies should expand on this work by addressing current limitations, particularly through larger sample sizes and fully randomized controlled trial (RCT) designs, while avoiding data collection during periods of heightened emotional sensitivity. Moreover, the CRM holds promise from both policy and programmatic perspectives, particularly for advancing mental health and psychosocial support, peacebuilding, community reconciliation, and post-conflict recovery efforts. The broader implication lies in the potential of CRM to contribute to breaking cycles of conflict and violence by fostering individual and collective trauma healing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Availability of data and materials

The datasets generated and analysed during this study are not publicity available but can be accessed upon a reasonable request from the corresponding author.