Children's perceptions of bullying,exclusion and defending in English primary schools: A qualitative study

Theoretical perspectives on bullying and defending

Contemporary research conceptualises bullying as a socially embedded process shaped by interactions between individual characteristics and environmental systems. The social–ecological model developed by Swearer and Espelage (2003) positions bullying within nested contexts, namely individual, peer, classroom, school, family, and community contexts, that interact to reinforce or challenge aggressive and prosocial norms. This framework indicates that defending behaviour cannot be understood solely in terms of individual morality or empathy; it also reflects the social conditions that determine when standing up for others is possible, supported, or risky. Subsequent work extends this ecological understanding by showing how peer group structures and school climate influence both the prevalence of bullying and the likelihood of bystander intervention (Huitsing et al., 2020; Waasdorp et al., 2022).

Developmental theories of moral reasoning offer further insight into how children interpret harm and fairness within these contexts. Classic accounts proposed by Piaget (1932) and Kohlberg (1984) describe a gradual shift from externally guided rule adherence toward more autonomous moral reasoning. More recent work emphasises that children's moral understanding develops through everyday relationships and emotionally meaningful interactions rather than through abstract reasoning alone (Killen & Smetana, 2022; Rutland et al., 2024). Empathy, guilt, and concern for others are now viewed as affective foundations of moral action that are shaped by cognitive maturation and social experience (Eisenberg et al., 2006; Malti et al., 2016). Yet moral awareness does not necessarily predict behaviour. Children often encounter tensions between what they judge to be right and what feels safe or socially acceptable within peer hierarchies (Thornberg et al., 2012).

Bringing ecological and moral-developmental perspectives together suggests that prosocial defending is both moral and relational. Decisions to intervene are situated within the affordances and constraints of children's social environments and are shaped by competing values such as loyalty, belonging, and fairness. Understanding how children navigate these tensions requires attention to both the moral meanings they attach to peer harm and the situational factors that shape their responses. This framing underpins the present study's focus on children's perceptions of bullying, victimisation, and prosocial action within everyday school contexts.

Defending as a moral and relational act

Defending is often treated as a simple expression of empathy. In practice, pupils weigh what is happening, who is involved, and what it might cost them to act. A social–cognitive view helps to explain this judgement work: children read situational cues, attend to peer hierarchies, and consider likely outcomes before stepping in (Thornberg & Jungert, 2014). These judgements are moral as well as strategic, drawing on ideas about fairness, justice, and belonging.

Evidence shows that defending is easier when the social environment is safe and prosocial norms are visible, and harder when harm is tied to status and group loyalty. In classrooms where high-status pupils drive the behaviour, bystanders often hold back for fear of exclusion or retaliation (Pozzoli & Gini, 2010; Salmivalli et al., 2011). Where schools make inclusion normative and adults are present and responsive, pupils report greater willingness to comfort or defend peers (Aldridge & McChesney, 2018; Thornberg et al., 2012). Programme trials that shift peer norms have reduced bullying and supported defending, underlining the role of context in enabling moral action (Huitsing et al., 2020).

Moral domain work adds further nuance. Children distinguish moral concerns from social-conventional and personal ones and use that distinction when making sense of peer harm (Smetana, 2013). Many express sympathy yet still ask whether help is deserved, especially where a peer has previously excluded others. Such reasoning brings fairness and reciprocity into view and shows how empathy is filtered through relational history (Killen & Smetana, 2022; Malti et al., 2016).

Defending also involves emotional labour. Acting for someone else can produce anxiety, guilt, or strain when loyalty to friends clashes with personal values (van der Ploeg et al., 2017). Pupils describe efforts to hold on to kindness and fairness while managing the realities of peer life. This shifts the focus away from trait-based accounts of prosociality and towards the affective and relational conditions that make action possible.

In this study, the analysis centres on pupils’ perceptions and anticipated responses rather than observed behaviour. Defending is treated as a situated moral act that depends on empathy and fairness but is calibrated to perceived safety and peer norms (Huitsing et al., 2020; Salmivalli et al., 2011).

What remains under-explored

Existing work shows that bullying is socially embedded and that defending depends on peer norms, perceived safety, and school climate (Huitsing et al., 2020; Hymel & Swearer, 2015; Waasdorp et al., 2022). Qualitative studies also indicate that pupils’ decisions are context sensitive and negotiated within everyday relationships (Forsberg et al., 2014; Thornberg, 2015). Yet three areas require further attention. First, there is limited fine-grained analysis of how primary-aged pupils reason about deservingness, reciprocity, and fairness when deciding whether to help, beyond broad references to empathy or prosocial tendencies (Killen & Smetana, 2022; Malti et al., 2016). Second, we know comparatively less about how children integrate emotional understanding with assessments of peer power and anticipated social costs in routine school situations, rather than in abstract tasks alone (Rutland et al., 2024; Thornberg & Jungert, 2014). Third, the English primary context needs further pupil-centred evidence that connects these moral judgements to the social organisation of classrooms and playgrounds, given the demonstrated importance of climate and norms for bystander behaviour (Aldridge & McChesney, 2018; Salmivalli et al., 2011). Addressing these gaps can sharpen theoretical accounts of defending as a situated moral practice and inform school-level work that targets the conditions which make prosocial action feel possible.

The present study

This study investigates how primary-aged pupils in England conceptualise bullying and social exclusion, and how they weigh emotional, social, and moral considerations when deciding whether and for whom to act. Working from a realist orientation and a child-centred design, the analysis focuses on pupils’ perceptions and anticipated responses in school settings rather than observed behaviour. The study extends prior research by tracing how fairness, reciprocity, and perceived peer power feature in pupils’ reasoning about defending within everyday interactions. Three research questions guided the study:

How do children define bullying and exclusion in context, and what emotional meanings do they attach to these experiences?

Research design

This study adopted a realist epistemological stance, treating children's accounts as meaningful representations of their lived experiences and school contexts while recognising the interpretive role of analysis (Maxwell, 2012). A realist orientation treats interview accounts as meaningful reports of perceived experience and social meaning, while acknowledging the interpretive role of analysis; reflexive thematic analysis (Braun & Clarke, 2006, 2019, 2021) was chosen because it aligns with this stance and focuses on meaning-making in participants’ accounts.

Procedures were guided by child-centred participatory principles. In line with the United Nations Convention on the Rights of the Child (Article 12) and the Lundy model, interviews were designed to provide space for expression, facilitate voice, ensure an audience, and allow influence over topic emphasis and pacing (Lundy, 2007; United Nations, 1989). These principles were operationalised through: (i) child-friendly information and assent procedures that emphasised the right to pause or decline questions; (ii) a paired interview format to enhance comfort and mutual support; and (iii) an interview sequence that moved from open prompts about everyday school life to short vignettes, reducing social risk while enabling reflection on morally ambiguous situations.

Participants

Participants were recruited through purposive sampling, with schools selected on the basis of accessibility and willingness to participate while aiming for diversity across educational contexts. Schools were identified through professional networks and public databases. Once agreement was secured, pupils from eligible year groups were identified in collaboration with school staff. This strategy sought to capture a range of social experiences without privileging any particular profile. Variation was therefore sought across school type (state-maintained and independent), school context, and year group (Years 4 and 5), recognising that differences in school ethos, peer culture, and everyday social organisation can shape children's experiences and interpretations of peer harm.

Data collection took place in five primary schools across England, comprising two state-maintained and three independent settings. This variation was intentional because differences in school ethos, peer culture and resource environments can shape children's social behaviours and interpretations of harm (Salmivalli, 2010). The final sample included 36 children (13 boys and 23 girls) aged between 8 and 10 years (M = 9.00, SD = 0.76), all in Key Stage 2, Years 4 or 5. Years 4 and 5 were selected because peer relationships become increasingly significant in middle childhood and children at this stage are more likely to reason about moral dilemmas in contextually sensitive ways (Warden & MacKinnon, 2003). Demographic details (location, sex, school-year) are summarised in Table 1. To protect anonymity and minimise burden, no additional demographic information (such as ethnicity, or special educational needs) was collected. All participants were assigned a participant number, which was used throughout transcription, analysis, and reporting to protect confidentiality. All interviews were conducted in pairs (dyads), yielding 18 dyadic interviews in total.

OPEN IN VIEWER

Table 1.

School location, participant numbers, and year group distribution.

			School year (N)
School location	Sample size (N)	Sex (male: female)	Year 4	Year 5
Sussex	8	4: 4	4	4
Middlesex	3	2: 1	0	3
West Yorkshire	8	4: 4	8	0
Birmingham	9	0: 9	0	9
Hertfordshire	8	4: 4	4	4
Overview	36	13: 23	16	20

Data collection and instruments

A semi-structured interview guide was developed to explore children's perceptions of bullying, peer victimisation and defending. The schedule comprised 25 open-ended questions, of which 11 focused specifically on how pupils conceptualised these behaviours and the relational, emotional and moral considerations shaping their responses. A semi-structured format was chosen to support depth and flexibility while maintaining consistency across core topics (Kallio et al., 2016).

To support accessibility and engagement, we used visual cue cards and a short vignette sequence as concrete prompts. Visual and task-based materials are recommended in interviews with children to scaffold talk, reduce abstraction, and create psychological safety, particularly for sensitive topics (Barter & Renold, 1999; Clark, 2005; Darbyshire et al., 2005; Fargas-Malet et al., 2010; Punch, 2002).

A fictional character, ‘Alex’, was introduced as a narrative device to enable discussion at a safe psychological distance. Alex encountered varied peer dynamics, such as being helped, ignored or harmed by others. By reflecting on Alex's experiences, pupils could articulate perceptions and anticipated responses without feeling personally exposed. The interview schedule focused on in-person, school-based peer interactions (Years 4–5). Cyberbullying was not included by design, as the study examined how pupils interpret and respond to peer dynamics they encounter directly during the school day.

Prior to beginning the study, ethical approval was granted by the University of Nottingham, School of Psychology Research Ethics Committee (approval F1405). Schools were initially approached by email and telephone. Once agreement was secured, logistics were confirmed with designated staff. Parents and carers received letters and age-appropriate information sheets describing the purpose, procedures and ethical safeguards. Written parental consent and written child assent were obtained prior to participation. Interviews were conducted between January and February 2024 in quiet school spaces during the school day. Each session followed a consistent sequence; it began with open questions about pupils’ own school experiences, then moved to short vignette scenarios featuring Alex. This ordering prioritised rapport and psychological safety before introducing morally complex situations that invited reflection and justification. All interviews were conducted in pairs, balancing individual expression with social comfort. Grouping decisions were made with classroom teachers, taking account of existing peer relationships and communication needs.

All interviews were audio-recorded using the Voice Memos application on iPhones and transcribed verbatim using Whisper AI's speech-to-text software. Transcripts were then checked manually by the research team to ensure fidelity to participants’ words. A sample of the interview schedule is provided in Supplementary Appendix A.

Researcher positionality

The research team's perspectives were recognised as integral to the production and interpretation of knowledge. The lead researcher's background in developmental and social psychology, alongside professional experience in education and child participation research, shaped the study's focus on children's meaning-making and relational school dynamics. This orientation influenced the design of the interview materials and the analytic attention to children's emotional and moral reasoning.

Reflexive practice was embedded throughout. A journal was maintained to record methodological decisions and analytic reflections, and regular peer debriefs were used to question assumptions and refine interpretations. Attention was given to the power asymmetries inherent in adult-child research and to the interpretive privilege researchers hold when representing children's voices. Care was taken to centre children's meanings as they emerged in the data rather than imposing adult-centred or normative interpretations. The approach was iterative and dialogic, not positioned as neutral or bias-free, and treated reflexivity as an ethical commitment to honour the specificity and complexity of children's accounts, consistent with participatory research traditions that uphold children's rights to express views and contribute meaningfully to research that affects them (Lundy, 2007; Pascal & Bertram, 2009).

Data analysis

Data analysis proceeded through an inductive, iterative process, beginning with repeated readings of the transcripts to develop familiarity with pupils’ language, context, and accounts of peer interaction. Coding was conducted manually, line by line, to maintain closeness to participants’ wording and preserve the integrity of children's phrasing. Both descriptive and interpretive codes were applied, for example, ‘noticing exclusion’ and ‘helping quietly’, alongside more analytical codes such as ‘weighing social risk’ and ‘conditional empathy’. Related codes were reviewed and collated into provisional themes through an iterative process that involved continuous movement between the dataset and analytic notes.

Theme development was dialogic and collaborative. Draft theme maps were reviewed repeatedly in light of the full dataset and through critical-friend discussions within the research team. Disconfirming and minority cases were retained to test theme boundaries and refine definitions. Analytic memos were used throughout to document decisions, uncertainties and the evolution of themes. For example, instances where pupils expressed clear empathy but described withholding action were examined as disconfirming cases, prompting refinement of the distinction between moral concern and enacted defence. Reflexive memos were used to record such analytic tensions and to document decisions about theme boundaries and naming as analysis progressed. Illustrative examples of disconfirming cases, memo extracts, and the analytic trail from data to themes are provided in Supplementary Appendix B. The final thematic framework reflected patterns that were coherent internally and distinct from one another, illustrated with extracts that captured the range of children's perspectives.

Trustworthiness and analytic quality

Trustworthiness was approached as an integral feature of the analytic process rather than as a set of procedural checks. Reflexivity was supported through ongoing memo writing, repeated return to the dataset, and critical-friend discussions within the research team, allowing analytic assumptions and decisions to be questioned and refined over time. In keeping with reflexive thematic analysis, inter-rater reliability statistics and code agreement metrics were not used; instead, analytic quality was evaluated through coherence, reflexivity, and the fit between data and interpretation (Braun & Clarke, 2021; Yardley, 2008).

Attention to coherence involved ensuring that themes captured patterned meaning across the dataset while remaining conceptually distinct, with theme boundaries revised in response to disconfirming and minority cases. The fit between data and interpretation was maintained through careful selection of extracts that substantiated analytic claims and demonstrated how interpretations were grounded in participants’ accounts. Transparency was supported through the provision of an analytic trail in Supplementary Appendix B, which documents the progression from data excerpts through initial codes and sub-themes to final themes, alongside reflexive memo extracts that illustrate analytic decision-making.

Theme 1: Understanding bullying dynamics

Across most dyadic interviews, children differentiated bullying from everyday conflict by emphasising repetition, intention and an imbalance of power, and they linked this to how it feels for the child on the receiving end. These were not stock definitions; pupils used them as working rules to sort everyday situations in school. P3 (Year 4, state) described bullying as ‘hurting someone's feelings or physically hurting them over and over again’, while P9 (Year 5, independent) added, ‘it is when someone is being mean on purpose, not by accident. They keep doing it’, together marking repetition and deliberate harm as the boundary with one-off fallouts. This often unfolded as a stepwise judgement: pupils first identified patterned harm, then considered how it felt for the target, before finally considering what the situation called for.

Alongside this behavioural framing, pupils foregrounded emotional impact. Asked how a bullied child might feel, P12 (Year 5, state) said ‘really scared and frightened … probably scared and nervous’, and P19 (Year 4, independent) noted, ‘I think he is feeling, like, left out’. In practice, pupils held together what happens and how often with how it lands on the child who is targeted. These criteria tended to operate as thresholds. Isolated unkindness was usually treated as conflict, whereas repetition and intent typically led pupils to name it as bullying and to see adult support or peer backing as more appropriate.

Criteria were applied with attention to context, status and audience. Several pupils recognised that who is involved and who is watching shape what happens next. P24 (Year 5, state) reflected, ‘If it is the popular kids doing the bullying, they might turn on me. I want to help, but then they will make fun of me too. Sometimes you just stay quiet, so you do not get noticed’. Others described reading the situation before naming it as bullying or stepping in. P30 (Year 5, state) said ‘I would go with a friend so I am not alone’. In practical terms, recognition did not always lead to naming or challenge. Decisions were often routed through an appraisal of who held influence, who was present, and how public the moment felt. Some pupils described acting despite risk when they felt the harm was clear and immediate, although most adjusted what they did to the audience and setting.

Analytically, accounts combine a clear definition of bullying with a situational appraisal of risk, such that recognition and interpretation are shaped by status, audience, and how public the moment feels.

Theme 2: Empathy constrained: barriers to moral action in school settings

Across the majority of dyadic interviews, pupils described feeling for children who were left out or upset and suggested ways they might help. The focus here is on pupils’ perceptions and anticipated responses in everyday and hypothetical situations, not on observed behaviour. At the same time, pupils outlined why help did not always follow. Fear of retaliation, practical or emotional pressures, and uncertainty about what to do often shaped whether intentions became action.

Empathy was clear in how pupils read situations and inferred feeling states. P14 (Year 4, state) noticed, ‘He is sitting on his own and no one is playing with him. It must feel really lonely’, and P7 (Year 5, independent) offered a simple response, ‘I would ask them if they want to come and play with me’. Others described checking in quietly, ‘I would ask if they are ok, maybe sit with them for a bit’ (P11, Year 5, state), or trying to include someone without drawing attention, ‘I would pick them for my team so they are not by themselves’ (P18, Year 4, independent).

Alongside this, pupils weighed the social cost of stepping in. When those involved were seen as popular or influential, fear of becoming a target was salient. P23 (Year 5, state) reflected, ‘If it is the popular kids doing the bullying, they might turn on me. I want to help, but then they will make fun of me too. Sometimes you just stay quiet so you do not get noticed’. P26 (Year 5, independent) explained the calculation more plainly: ‘You do not want everyone to look at you and think you are taking sides’. In practice, moral concern did not always translate into overt action. Decisions were often routed through an appraisal of who held influence, who was present and how public the moment felt.

Practical and emotional pressures also shaped choices. Some pupils described moments where immediate needs or mood took priority, ‘If it is right before lunch and I am really hungry, I might go and eat first’ (P21, Year 4, state). Others held back because they were unsure, ‘I want to help, but I do not know what to say’ (P10, Year 5, independent), or uncertain whether it was their place, ‘Sometimes you are not sure if it is a joke, or if they are friends, so you leave it’ (P29, Year 4, state). For some, involving an adult felt safer: ‘I would tell the teacher because if I say something it could make it worse’ (P33, Year 5, independent).

Pupils described small strategies for managing these tensions. Several preferred to watch first, ‘I would see if it keeps happening or if it stops’ (P4, Year 4, state), to seek support, ‘I would go with a friend so I am not alone’ (P16, Year 5, state), or to adjust how and when they helped, ‘I would talk to them at break instead of saying it in front of everyone’ (P20, Year 4, independent). A few described acting despite risk when they felt the harm was clear and immediate, although most adjusted what they did to the audience and setting.

Across accounts, empathy met constraint. Pupils recognised exclusion and distress and often wanted to help, but naming, timing and action were shaped by perceived risk, competing pressures and uncertainty about the best next step.

Theme 3: Moral dimensions of prosocial defence and reciprocity

Beyond immediate concerns about risk or uncertainty, pupils also described helping as the right thing to do, yet accounts showed that kindness and defence were not always unconditional. Decisions were filtered through ideas of fairness, deservingness and the hope that care would be returned, a pattern evident across most interview pairs. In practice, pupils moved from feeling for someone to weighing what seemed fair and what might follow.

Several pupils spoke of kindness as a guiding principle. P1 (Year 4, state) said, ‘Even if they were mean to me before, you should still be kind’, and P22 (Year 5, independent) added, ‘If someone is sad, you help them. That is just what you should do’. Others linked helping with a sense of shared good: ‘If you cheer them up, everyone feels better in class’ (P31, Year 5, state). For these pupils, including others and standing up for them felt like part of keeping things fair.

Alongside this, many reflected on how helping felt. P6 (Year 4, independent) explained, ‘When I help someone, it makes me feel happy inside’, while P27 (Year 5, state) noted, ‘If you make them smile, you feel proud after’. Pupils portrayed prosocial acts as fitting with who they wanted to be, as well as with what they thought was right.

Kindness was also evaluated. Some pupils qualified their willingness to help based on peers’ past behaviour. P15 (Year 4, independent) commented, ‘If they are always unkind to others, I would leave it. They need to learn’, while P34 (Year 5, state) reasoned, ‘It is fair to help if they have tried to be nice before, not if they always make fun of people’. Others combined both views, leaning towards inclusion while keeping an eye on fairness: ‘Even if they were not nice, I would still help a bit because it could make them change’ (P17, Year 4, state). These judgements suggest a weighing of moral obligation against relational history.

Pupils also expressed hope that kindness might spread. P36 (Year 5, independent) suggested, ‘If you are nice to them, maybe next time they will be nice to someone else’, and P5 (Year 4, state) added, ‘Helping someone makes them want to help too’. For some, helping was not only the right thing to do but a way to influence what kind of class they wanted to be part of.

Analytically, Theme 3 captures pupils’ shift from empathy to evaluation. Acts of kindness were often weighed against fairness, past behaviour and likely outcomes, so prosocial defence appeared both principled and strategic, shaped by relationships and by what pupils hoped their actions might set in motion.

Beyond behavioural recognition: emotional and relational understandings of bullying

Children in this study described bullying not simply as repeated acts of aggression but as behaviour that was deliberate, patterned and emotionally charged. Their accounts reflected their own perceptions and experiences of how harm occurs and is felt in school life. These descriptions resonated with the widely accepted definition of bullying as intentional, repeated harm within relationships marked by power imbalance (Olweus, 1993) while extending it by emphasising the emotional impact and relational consequences. Pupils’ references to feeling ‘scared’, ‘nervous’, or ‘left out’ showed that they understood bullying through an affective lens, locating harm in the experience of exclusion as much as in overt aggression. This way of speaking about bullying aligns with field syntheses that view it as a complex social process rather than an individual trait or discrete event (Hymel & Swearer, 2015).

This perspective is consistent with ecological accounts in which bullying unfolds within layered peer and classroom systems shaped by status hierarchies, audience reactions and local norms. Recent studies show that pupils’ perceptions of school climate and peer relationships influence how they recognise and interpret bullying and whether they see intervention as possible (Waasdorp et al., 2022). In the present study, children's accounts similarly highlighted awareness of who was involved, who witnessed events and how social positioning could determine what counted as bullying and whether it might be challenged.

Such insights affirm that primary-aged children are capable of empathic perspective-taking and moral evaluation. Rather than supporting earlier assumptions that younger pupils lack the social understanding to appreciate complexity, these accounts show a nuanced moral vocabulary and an early sensitivity to the relational dimensions of harm often underplayed in behaviourist models. Current summaries of bullying research also recognise that children's lived experiences of belonging, exclusion and fairness shape how bullying is sustained or resisted within schools (Hymel & Swearer, 2015).

Yet recognising harm did not always lead to action. Many pupils described hesitation about defending peers, particularly when those involved were socially dominant or when intervention felt risky. This finding aligns with classroom-level studies linking bystander behaviour to bullying frequency and with evidence that pupils’ perceptions of safety and peer norms influence whether they act (Salmivalli et al., 2011). Emotional awareness may therefore be necessary but not sufficient for prosocial response: intentions were negotiated within perceived hierarchies and concerns about belonging. These relational dilemmas form the basis for the next section, which examines how empathy was expressed yet frequently constrained by social and situational pressures.

Barriers to defending: the social cost of prosocial action

While many participants expressed clear empathy for peers who were bullied or excluded, their accounts revealed how this awareness was often met with hesitation or restraint. Children described feeling sympathy and moral concern but recognised that acting on these feelings could entail social risk. These reflections situate bullying as a social context where moral intentions are balanced against the need for safety and belonging.

Across interviews, pupils spoke about the pressures of peer hierarchies and the fear of retaliation. Several noted that bullies were sometimes ‘popular’ or ‘leaders’, which made challenging them feel unsafe. Such perceptions are consistent with research showing that social dominance and audience approval can deter bystander intervention (Salmivalli et al., 2011). Qualitative work similarly indicates that children's decisions to act depend on their interpretations of group dynamics and potential consequences (Thornberg et al., 2012). From pupils’ perspectives, defending was rarely a neutral act; it could disrupt friendships, invite ridicule, or jeopardise one's own standing within the group.

Children also identified situational and emotional factors that shaped their willingness to intervene. Some described feeling unsure about what to do or when to do it, indicating a lack of procedural clarity rather than moral disengagement. Others mentioned being tired, distracted or preoccupied, moments where immediate needs outweighed empathic impulses. Developmental research supports these accounts: self-regulatory demands, emotional arousal and stress can all reduce children's capacity to act on prosocial motivations (Eisenberg et al., 2006). Pupils’ narratives here reflected similar internal negotiations, where empathy was present but constrained by context.

Importantly, these were not excuses but thoughtful appraisals of what intervention might cost. Children's reflections showed an awareness that defending required both moral courage and emotional safety. Their accounts echo social identity perspectives in which individuals act to protect group belonging and self-esteem within valued social contexts (Tajfel & Turner, 1979). The implication is that defending cannot be understood only as a moral decision; it is also a relational one, entangled with how pupils perceive peer norms, teacher visibility and the likelihood of support.

These findings align with broader ecological evidence that positive peer norms and supportive school climates increase pupils’ willingness to defend others, in U.S. and European samples (Waasdorp et al., 2022). Participants in this study recognised exclusion and distress, but their sense of agency was often diminished by uncertainty about how others, peers or adults, would respond. Their accounts portray empathy as a necessary but insufficient condition for prosocial action. This moral negotiation between empathy and safety underpins the next theme, which examines how fairness, reciprocity and relational history shape when and why pupils choose to defend.

Moral dimensions of prosocial defence and reciprocity

Participants’ reflections on their experiences revealed that decisions to defend were shaped not only by empathy or fear, but by moral evaluations of fairness, reciprocity and relational history. Children described wanting to help but hesitating when the individual in question had previously been unkind or exclusionary. These accounts suggest that prosocial intentions were filtered through pupils’ sense of justice, as they weighed who was deserving of help and why, consistent with developmental evidence that moral evaluations in middle childhood integrate considerations of fairness, intent and social context (Killen & Smetana, 2022).

Across interviews, pupils presented kindness as both an ethical value and a relational practice. Many described helping because it was ‘the right thing to do’, while others linked kindness to maintaining or repairing relationships. Longitudinal work shows that sympathy, guilt and moral reasoning contribute to children's helping, cooperation and sharing across time, supporting the view that prosocial action is grounded in moral emotions and evaluative judgement rather than altruism alone (Malti et al., 2016). At the same time, some participants framed kindness as conditional, expressing reluctance to help peers who had previously hurt them or others. This conditional reasoning reflects a developing sense of distributive justice, in which fairness and reciprocity are weighed alongside empathy (Killen & Dahl, 2021).

Children's moral reflections also carried a forward-looking dimension. Several described hoping that helping someone might ‘make them nicer next time’ or encourage others to behave more kindly. This orientation towards relational repair echoes the logic of reciprocal altruism (Trivers, 1971) and aligns with evidence that peer-norm interventions can shift behaviour at scale. For example, a large randomised trial in the Netherlands replicated the effectiveness of the KiVa programme in reducing bullying and victimisation, consistent with the idea that visible, valued prosocial action can alter classroom dynamics (Huitsing et al., 2020).

Yet these reflections also revealed tension between moral ideals and lived experience. Some children suggested that certain peers ‘did not deserve’ help, showing how empathy can be moderated by perceived fairness and prior interaction. Others described selectively helping those who ‘really needed it’, implying a hierarchy of moral obligation based on context and deservingness. Such distinctions correspond with research indicating that by late primary school children differentiate between moral transgressions, social conventions and personal choice (Turiel, 2006). Pupils’ reasoning therefore illustrates the evaluative and context-sensitive nature of moral action in peer relationships, consistent with contemporary accounts of moral development as relational and situated rather than linear or stage-bound (Rutland et al., 2024).

Understanding these moral dimensions has practical implications for how schools foster defending and inclusion. Programmes that focus solely on increasing empathy may neglect the evaluative and reciprocal nature of children's moral reasoning. Approaches grounded in restorative dialogue, moral inquiry and peer-led discussion may better support pupils to reconcile empathy with fairness and relational accountability, complementing whole-school efforts that target peer norms and provide safe, practical scripts for action (Garandeau et al., 2023; Huitsing et al., 2020).

Implications for practice, theory, and policy

The findings of this study offer a nuanced understanding of how children perceive, interpret, and emotionally experience moral and social tensions in situations of bullying and exclusion. Rather than depicting pupils as passive bystanders, the results portray them as moral agents whose prosocial intentions are negotiated within relational hierarchies and institutional structures. These insights carry important implications for how theory conceptualises moral development, how educators design interventions, and how policy frames children's moral agency within school contexts.

From a theoretical standpoint, the study contributes to an emerging shift in developmental psychology that views moral reasoning as relational, emotionally embedded, and contextually situated (Rutland et al., 2024). Children's accounts demonstrate that empathy, fairness, and belonging are intertwined rather than sequential, supporting perspectives that move beyond stage-based models of moral development (Killen & Smetana, 2022). By framing defending as a process of negotiation rather than an automatic response, these findings extend ecological models of bullying to show how individual reasoning interacts with the broader social ecology of the classroom (Hymel & Swearer, 2015; Waasdorp et al., 2022).

For practice, the findings suggest that anti-bullying and character education interventions should move beyond promoting empathy alone. Many pupils recognised distress and wished to act but lacked confidence, procedural clarity, or emotional safety. Interventions should therefore address both what children feel and how they can act safely and effectively. Strategies such as role-play, peer-led discussion, and restorative dialogue can help pupils explore moral reasoning in real-world scenarios while modelling inclusive norms (Garandeau et al., 2023). Whole-school approaches that strengthen teacher visibility and responsiveness, reinforce collective responsibility, and provide clear behavioural scripts are also vital for enabling prosocial action (Huitsing et al., 2020).

At the policy level, these findings underline the need for relational approaches to bullying prevention that recognise children's perspectives as evidence of moral competence rather than immaturity. Behaviourist responses that rely primarily on sanctions or surveillance risk ignoring the social and emotional dynamics that sustain bullying. In contrast, policies that promote supportive climates, characterised by warmth, inclusion, and visible adult care, have been shown to reduce peer aggression and foster a culture of defending (Aldridge & McChesney, 2018; Waasdorp et al., 2022). While these findings derive from English primary schools, their relational emphasis resonates with international evidence on how peer norms and adult responsiveness shape children's sense of safety and belonging. Embedding relational ethics into national guidance and school policy would align more closely with children's lived moral experiences, positioning prosociality as a collective moral practice rather than an individual trait.

Limitations and future directions

This study offers contextually grounded insights into children's perceptions of bullying, prosociality, and moral reasoning, and its findings are best understood in relation to the settings and populations in which the research was conducted. The sample was drawn from five primary schools in England, including both state-maintained and independent settings. Rather than aiming for statistical generalisation, the study supports analytic transferability to primary school aged children in middle childhood (approximately 8-10 years), particularly those educated in formal school environments where peer relationships are organised within classrooms, adult supervision is routine, and face-to-face interaction shapes everyday social life. The findings may therefore be most relevant for researchers, educators, and practitioners working with children in comparable primary education contexts where peer hierarchies, belonging, and moral expectations are negotiated in daily school interactions. The study relied on children's self-reported perceptions rather than observed behaviours. Paired interviews and hypothetical scenarios offered an ethically sensitive means of accessing moral reasoning but may reflect aspirational or socially desirable accounts. Combining qualitative interviews with ethnographic or participatory methods could offer deeper insights into how defending is enacted in everyday school life.

Demographic information such as race and ethnicity was not collected in this study. This decision was taken to minimise participant burden and protect anonymity within small school-based samples; however, it limits the extent to which the findings can be interpreted in relation to racialised experiences of bullying, exclusion, and defending. Research indicates that children's peer relationships, perceptions of harm, and access to moral and social safety are shaped by intersecting social positions, including race and ethnicity (Hymel & Swearer, 2015; Killen & Smetana, 2022). The absence of these data means that potential differences in how children from racially minoritised backgrounds reason about bullying or experience social risk cannot be examined here. Future research would benefit from attending more explicitly to racial and ethnic diversity alongside school context, in order to better understand how moral reasoning and prosocial action are shaped by structural and relational inequalities in school settings. The sample focused on children aged 8 to 10, capturing a formative stage of moral and social development in primary school contexts. As moral reasoning and peer influence evolve through adolescence (Eisenberg et al., 2006), longitudinal work is needed to trace how empathy, fairness, and belonging interact over time.

Cyberbullying was intentionally excluded to focus on face-to-face dynamics, yet digital interactions increasingly shape children's moral and relational worlds. Future research should examine how empathy and moral judgement operate across online contexts (Hong et al., 2025), informing integrated approaches to prevention.

Finally, adult and institutional perspectives were beyond the scope of this study. Teachers’ modelling, peer norms, and school climate are likely to influence how children perceive moral safety and responsibility (Hymel & Swearer, 2015). Multi-informant and systems-level research could further clarify how relational environments enable or constrain prosocial action.