Exploring Children’s Self-Reports of Victimisation Experiences and Solitary, Prosocial, and Aggressive Behaviours

Abstract

Young children’s social behaviour plays an important role in their lives. Most research makes use of reports from adults and peers, rather than speaking to young children about their own behaviour. The current study was exploratory and looked at children’s self-reports of solitary, prosocial and aggressive behaviours, and victimisation experiences, during the formative school years. Children aged between four and seven years (n = 193) were recruited from three primary schools. Using an innovative method with stick-figure animations, they were asked to provide frequency ratings of 16 of their own behaviours. Class teachers were also asked to provide ratings for the participating children’s behaviours. Key findings showed that children offer unique insight into their behaviour and experiences. Specifically, children’s self-reports of victimisation experiences are more complex than their teachers; and teachers were found to under-report solitary behaviour in boys in comparison to self-reports. Self-reports of prosocial and aggressive behaviour showed some possible social desirability amongst 6–7-year-olds, but there are also potential situational and gender biases in reports from their teachers. These findings show the importance of including young children’s self-reports in the study of their behaviour.

1. Introduction

Young children spend a significant amount of time with peers, and their social behaviour within the peer group can play an important role in their lives and outcomes. For example, research has shown that the peer relations formed during early classroom experiences can act as precursors of school achievement [1], school adjustment [2], emotional competence [3], and mental health [4]. However, research in this area traditionally makes use of teacher-, peer-, and observer-reports of children’s behaviour and the use of self-reports is generally neglected, particularly with younger children [5]. Therefore, this study aims to explore the use of 4–7-year-old children’s self-reports of their own behaviour.

Research examining the behaviour of young children spans a range of peer-related behaviours such as solitary behaviour [6], prosocial behaviour [7], aggression [8] and victimisation experiences [9]. Such research mostly concentrates on one or two of these areas. These concepts are complex (see Section 1.2) but are closely linked to different outcomes and experiences for young children. Whilst these is some variation by form and function, general patterns are evident which show the importance of understanding these behaviours and experiences. Solitary behaviour tends to be associated with challenges such as speech and language difficulties [10], academic performance [11], and increased anxiety [12]. Similarly, displays of aggressive behaviour in young children have been linked to challenges in academic performance [13], peer rejection [14], and increased criminal behaviour later in life [15]. Children who experience victimisation have been found to experience negative feelings such as anxiety, loneliness, and depression [16]. Furthermore, children who experience victimisation have been found to show less academic motivation and engagement, and in turn, lower academic achievement [17]. In contrast, prosocial behaviour amongst young children has been linked to positive experiences such as higher levels of academic achievement [18], quality student–teacher relationships [19], and positive emotions [20]. Prosocial behaviour has also been found to protect against the potentially negative effects of victimisation and loneliness [21].

More recently, research has shown the importance of considering all four areas when studying children’s behaviour and has employed this approach in looking at friendships [22] and Theory of Mind (ToM) [23]. Therefore, the current research also adopts this approach and addresses gaps in the research by exploring children’s self-reports across the four following areas: solitary behaviour, prosocial behaviour, aggression, and victimisation experiences.

1.1. Teacher-Reports and Self-Reports

Teachers spend a substantial amount of time with children in various contexts at school and are able to make comparative assessments across multiple children [13]. As such, teacher-reports of children’s behaviour are seen as highly cost-effective and reliable. However, there are disadvantages to the use of teacher-reports, such as demographic [5] and situational [24] biases, which highlight the need to also consider children’s self-reports. In the UK, a child’s first year at school (when they are 4–5 years old) is part of the play-based Early Years’ Foundation Stage (EYFS). Once children move into their second year of formal schooling (Key Stage 1: ages 5–6 years), the focus is more on acquiring skills and knowledge [25,26]. Therefore, teachers may be less aware of children’s behaviour that takes place on the playground, or even informally within the classroom, because of their focus on more formal education. Furthermore, teachers (and others) are not able to reliably report on the internal processes or motivations that impact a child’s own behaviour [27,28,29]. In contrast, self-reports can show a unique individual view of the situation, particularly for behaviours that may be less obvious to the outside observer, occur out of the sight of others, or relate to how the actor or recipient perceives the behaviour. Children’s own perspectives may also influence any impact that the behaviour has on the child [30,31]. Thus, it is important to understand the subjective experience of children across a spectrum of behaviours, rather than relying on the reports of those around them.

Where children’s self-reports of behaviour have been collected, this has tended to be with children aged seven years and over [32]. Self-reports by young children are often neglected because of concerns around their reliability and age constraints [5]. One of these concerns relates to social desirability. However, these risks are not unique to young children, and researchers have argued that young children may not even be aware of social desirability [31]. For example, Malti et al. [33] looked at emotional attributions and moral reasoning in kindergarten (mean age 6.2 years) and elementary school children (mean age 7.6 years) and argued that older children may be more inclined to show social desirability than younger children. This suggests that the usefulness of self-reports from young children should not be overshadowed by concerns of social desirability.

Challenges around collecting self-reports from young children also relate to time and resource allocation. For instance, young children’s reading and comprehension abilities, and shorter concentration spans, mean that traditional methods such as surveys and interviews can be difficult and, as such, innovative methods are required. Methods such as the use of puppets [34], static cartoons [35], pictures [36], and animations [37] have been used to support children in providing insights into their own behaviours. Studies that have utilised self-reports (mostly looking at aggression and victimisation) have demonstrated that young children can offer insights to their own behaviour [37,38,39] when appropriate time and resources are given to collecting these.

There is also evidence to suggest that children should have the cognitive capacity to provide self-reports of their own behaviour. By four years of age, most children have a sense of self and self-awareness [40] and there are substantial gains in their Executive Function (EF) between ages 2 and 5 years [41]. As such, children have the developing abilities to understand themselves as individuals, self-regulate behaviour, and engage in abstract thinking. Young children also have an increasing ability to see things and feel things from others’ perspectives with a reduction in egocentrism [42] and increasing ToM [43,44]. They also demonstrate increased empathy where they can relate to how others may feel [45]. These developments suggest that they should be capable of reflecting on their own behaviour and may relate to the types of behaviours they report as these skills develop. This strengthens the idea that children’s self-reports should be explored.

The above discussions suggest that there is additional value in comparing reports from a younger group (ages 4–5 years) and older group (ages 6–7 years) within ages 4–7 years. This is for both practical and theoretical reasons. Firstly, these make up two distinct year groups at school, where the focus of peer-interactions are very different [25,26]. Secondly, children’s cognitive abilities reach various milestones around age five. For instance, they are able to pass ToM first-order false belief tasks [43,44] and have rapid EF growth from age 3 to 5 years [41]. These continue to develop with age, including the ability to pass second-order false belief tasks [46] and further developing, but less rapidly, EF [47]. The comparison of a younger and older group takes these ideas into consideration in the current research.

1.2. Behaviour Conceptualisations

1.2.1. Solitary Behaviour

Solitary behaviour in young children occurs when they stay on their own despite the presence of peers [48]. Researchers have identified various subtypes of solitary behaviour. For instance, ‘active isolation’ [49] or ‘peer rejection’ [50] refer to a child being isolated by their peers. In contrast, children may withdraw from their peers because they prefer solitary activities and are unsociable or are socially disinterested [49]. Coplan et al. [6] also identified ‘avoidance’, where children actively seek out situations where they can be by themselves. Another distinction is made for those children who may want to play with others but are ‘shy’ [49] or show ‘anxious-solitude’ [51], meaning that they feel fear and wariness over their desire to interact socially with their peers. Whilst these motivations are mostly internal, there is still a dearth in research collecting self-reports of solitary behaviour from younger children. The current research aims to fill this research gap. Exploration of multi-informant reports of solitary behaviour in slightly older children aged 7–10 years found some ‘mid-level’ agreement with other school-based informants (such as teachers and peers) [32]. Based on this previous research, the current study may show some agreement between children’s self-reports and teacher-reports of solitary behaviour.

Whilst research looking at adult reports of children’s solitary behaviour has found value in collapsing various solitary behaviours into a single measure [23], research has also shown different patterns in subtypes of solitary behaviour. For instance, teacher-reports have suggested that a preference for solitude, but not social anxiety, increases after early childhood [52]. It has also been shown that different forms of solitary behaviour relate to different outcomes for children [53]. Given that children aged 3 to 6 years are able to distinguish between the subtypes of solitary behaviour [54], there is value in studying these separately. It is tentatively predicted that young children will distinguish between types of solitary behaviour in their self-reports.

It is also worthwhile to consider gender. When using reports from others, researchers have found no gender differences in the frequency of solitary behaviour [11]. Therefore, it is expected that neither teacher-, nor self-reports, will show any gender differences. However, given that others have shown differences in the outcomes of solitary behaviour for boys and girls [55], a consideration of gender differences in children’s self-reports will help develop this understanding further. Others’ perceptions about solitary behaviour may also play a role in children’s self-reports. For example, shyness is generally seen as more acceptable amongst girls than boys by parents, teachers, and peers, although gender differences in acceptance reduce as children become older [11]. Therefore, the current study may show some age group differences across behaviour subtypes.

1.2.2. Prosocial Behaviour

There is a range of research that suggests that young children may be able to provide insightful self-reports of their prosocial behaviour. For instance, Jackson and Tisak [56] found that children aged 7–9 years were capable of reporting how they would behave in hypothetical situations. Further, children’s own perceptions are unique in that they appear to be less influenced by gender stereotypes of prosocial behaviour held by teachers [57], which was found to be the case in children aged 3–6 years [58]. Teachers and others may have perceptions of gender norms and stereotypes, where girls are seen as more prosocial than boys [55,59]. For instance, self-reports of prosocial behaviour by 11-year-olds were found to be higher than teacher-reports, but there was a greater discrepancy amongst boys [57]. The authors explain that this may be due to boys over-reporting, or it may relate to expectations and perceptions of other reporters. This demonstrates that there is value in exploring gender differences in children’s self-reports of their behaviour. Whilst the current study is exploratory, it is expected that self-reports of prosocial behaviour will be higher than teacher-reports, particularly for boys.

Studies looking at children’s prosociality often include socially desirable behaviours, such as sharing, helping, and comforting peers, regardless of motives [60]. Relational inclusion, where a child invites another to join them or a group, has also been included as an example of prosocial behaviour [23,61]. Research has shown that children may differentiate between subtypes of prosocial behaviour differently to others. For instance, researchers have shown high levels of consistency across subtypes of prosocial behaviour, but this relates to reports from others such teachers [62], rather than self-reports. In contrast, research has shown differentiation between subtypes amongst children, where children aged 7 years were less likely to report comforting hypothetical peers than other prosocial behaviours [56]. That said, there was less differentiation between other prosocial behaviours in 7-year-olds, compared to 9-year-olds. Therefore, the current research aims to explore whether this differentiation occurs in self-reports of 4–7-year-olds. It is tentatively predicted that there may be similarities across self-reports of different prosocial behaviours, but will be lowest for caring. Research has reported that children’s prosocial behaviour also tends to increase with age [5] and so it is possible that self-reports will be higher in the older age group. Fabes and Eisenberg’s [63] meta-analysis showed that gender differences in reports of prosocial behaviour, tend to increase with age but that this differed across subtypes of prosocial behaviour and recipients. Therefore, studying prosocial behaviours separately may help to understand different age trajectories and it is tentatively predicted that this, and gender differences, may vary across subtypes of behaviour.

1.2.3. Aggressive Behaviour and Victimisation Experiences

Unlike prosocial and solitary behaviour, studies of aggression and victimisation have utilised young children’s self-reports and demonstrated that they provide unique views from those of peers, parents, teachers, and observers [28,38,64]. Children have been found to self-report lower levels of aggression than peers and teachers [5,64]. In addition, research has shown more agreement in reports of victimisation between other informants than with self-reports [39,65]. This may be because other reporters are unaware of the internal motivations of children when displaying aggressive acts or their perceptions of their victimisation experiences [29]. Huitsing et al. [38] showed more agreement between 5 and 7-year-old peer- and self-reports of aggressor–victim relationships, than with teachers, also suggesting that teachers’ reports may be influenced by a lack of awareness of some situations. However, previous research using self-reports in the study of aggressive behaviour and victimisation experiences has generally focused on the identification of specific roles [64], or compared results across reporters on different measures [28], and there is scope to consider frequencies of behaviours on the same measures. Based on previous research, it is expected that children’s self-reports of aggressive behaviour will be lower than their teachers’. For victimisation, it is expected that teacher-reports will differ from self-reports across subtypes, but reports will not be specifically higher or lower.

Researchers use multiple distinctions when studying aggression and victimisation, including physical, verbal, indirect aggression (involvement of a third party), and relational aggression (damage to peer relationships) [66,67,68]. Direct aggression (a face-to-face encounter) has been found to be more common in early childhood than indirect aggression (such as rumour spreading) [64,65]. Studies collecting teacher- and peer-reports, also show some variation by gender. In early and middle childhood, non-relational aggression seems to be more common amongst boys, whereas relational aggression seems to occur more within girls [69,70]. Smith [71] has also argued that gender differences in relational aggression are shown amongst children aged under 7 years. When looking at gender differences in victimisation, the findings are complex. For instance, Rose et al. [69] conducted observations with younger children and showed that boys aged 2 and 3 years were more likely to be victims than girls, but that this was not the case for 4–6-year-olds, where boys and girls were equally victimised. The current study will explore whether these patterns occur within children’s self-reports, and it is tentatively predicted that there may be some gender differences in self- and teacher-reports of aggressive behaviour, but not victimisation experiences.

Research, mostly conducted with other reporters, has also looked at the trajectory of aggressive behaviour and victimisation. Overt aggressive behaviour has been found to decrease as children become older due to a developing ability to take perspectives of others and increased socialisation [5]. Others have found age differences across subtypes of aggression [64]. For victimisation, children’s self- and peer-reports suggest that children only tend to be the target of victimisation briefly [64], whereas teacher-reports have indicated that these experiences are stable for up to five months [65]. These findings further highlight the uniqueness of children’s own reports. It is tentatively predicted that reports of aggressive behaviour may differ between the two age groups but that reports of victimisation will be similar.

1.3. Research Gaps and Questions

The aim of the current study is to further understand children’s self-reports of their own behaviour with peers at school. The current study is informed by other research on peer-related behaviour and fills a gap by exploring 4–7-year-olds’ self-reports of their victimisation experiences, solitary, prosocial, and aggressive behaviours. The first research question considers how children self-report their behaviours, and how these vary by subtypes, age, and gender. The current research also aims to conduct preliminary exploration into interactions between these factors. The second research question considers how findings from self-reports compare to the same analysis of teacher-reports. The third research question looks to see how far individual young children’s self-reports of behaviour correlate with reports from their class teachers. Understanding children’s own perceptions of their behaviour will be useful for supporting outcomes relating to school achievement [1], school adjustment [2], emotional competence [3], and mental health [4].

2. Materials and Methods

2.1. Participants

The participants for this study included 193 children across two age groups (see

Table 1. Participant information.

	Younger Group (4–5 Years)	Older Group (6–7 Years)	Totals
Boys	50	46	96
Girls	52	45	97
Totals	102	91	193

Note: The mean age was 71.98 months. The Standard Deviation was 12.39.

for demographic information). The sample size was similar to other research collecting behaviour reports from children of similar ages and designs [39,54,56].

Participants for the study were recruited by approaching headteachers at schools who had previously been willing to take part in research. Three mainstream primary school headteachers, in the Southeast of England, agreed for their schools to participate. Class teachers (n = 8, all female) also provided reports about children’s behaviour for all child participants. Child participants were required to be a pupil in one of the three mainstream schools in either the younger (UK reception year, ages 4–5 years) or older age group (UK Year 2, ages 6–7 years). Teachers were required to be the class teacher of the participating children. It was not necessary to apply any exclusion criteria to the sample and the data for all 193 children who opted to participate was included.

2.2. Measures

A measure of 16 behaviours was developed based on existing scales, as in Rix et al. [23]. Teacher-reports were collected via an online survey on a five-point scale, where 1 was representative of ‘never/almost never’ and 5 was ‘always/almost always’. Research has shown that young children can provide reports on three-point scales [54], but tend to report at the extremes of five-point scales [72]. Therefore, children’s self-reports of their behaviour were collected on a three-point scale of ‘lots’, ‘sometimes’, and ‘never’.

Items were as per the measures in Rix et al. [23]: solitary behaviour—shyness, unsociable, active solitude, and contextual solitude; prosocial—sharing, caring, helping, and including; aggressive behaviour—indirect relational, direct relational, physical, and verbal aggression; victimisation (being the target of aggression)—indirect relational, direct relational, physical, and verbal victimisation. Cronbach alphas were acceptable for most teacher-reports (aggression: α = 0.78; victimisation α = 0.679; prosocial α = 0.702; solitary: α = 0.721) but were weaker for self-reports (aggression: α = 0.609; victimisation α = 0.613; prosocial α = 0.539; solitary: α = 0.438), suggesting that that children may have different views across subtypes of behaviour. Therefore, these were kept separate for analysis.

2.3. Procedure

There were several processes implemented to overcome the challenges of colleting self-reports from young children. Firstly, a team of five trained researchers collected data from children, meaning that appropriate time could be given to collecting a sufficient sample size. In addition, all researchers spent half a day with the participating class in each school within their normal classroom setting. This allowed children to build familiarity with the researchers and helped children to approach the research session with ease. Furthermore, we built on a previously successful method of engaging animations to ensure that the session was interactive. The research followed on from the work conducted by Rix [37], in which the animations were thoroughly tested through input from children in both preliminary research and piloting.

Child assessments were carried out individually in a quiet area of the school by trained researchers (n = 5, all female) during the regular school day. Each session lasted approximately 20 min. The task was explained to children with use of pictures and emphasis that there were no right or wrong answers. Stick people animations were shown in a set order, with an arrow above the stick person depicting each behaviour. To ensure that they fully understood the behaviour, children were asked to describe what they thought was happening in the animation, which was then confirmed or further explained by the researcher. The child was then asked how often they behave like this at school and provide their answer on a response pad, selecting ‘lots’ (a large circle), ‘sometimes’ (a medium sized circle), or ‘never’ (a small circle). Teachers were asked to complete the online form for the participating children in their class at a convenient time. All researchers were fully trained and observed by the lead researchers to ensure standardisation of methods with children.

2.4. Ethics

Ethical approval for the study was obtained from the University of Greenwich Ethics Committee (Reference 14.1.5.5, 10 October 2014). The headteachers in each of the three schools acted as gatekeepers and selected a class from a younger year group (ages 4–5 years) and older year group (ages 6–7 years) who were invited to take part. The school sent home letters to parents/carers, who had the choice to opt out of the research. This approach results in a more representative sample and places less burden on the school [73]. This was followed up by gaining verbal assent from children who could opt out or withdraw at any point. Parents/carers could withdraw their child’s data, up to a date, approximately one month after data collection. Children were fully debriefed after the study. Teachers provided consent for their own participation and completed the questionnaires in their own time. Privacy and confidentiality were maintained by using codes throughout data collection and assuring participants that their answers would not be shared with anyone.

2.5. Statistical Analysis Appraoch

To address the first research question, a series of mixed Analysis of Variance tests (ANOVAs) were run on the self-reports for the four main categories of behaviour (solitary, prosocial, aggression, victimisation). For each, the four subtypes of behaviour were entered as within-subject variables. Age and gender were entered as between-subject variables. An identical set of mixed ANOVAs were run on teacher-reports to address the second research question. Means were adjusted for means of other factors in the model and Estimated Marginal Means (EMMs) and Standard Errors (SEs) (to measure the statistical accuracy of the EMMs) have been reported for significant main effects and interactions.

Mixed ANOVAs (for which there is no non-parametric equivalent) were deemed appropriate for the necessary analysis across several variables. Whilst assumptions of normality were not met in all cases, the current study employed an appropriate sample size [74] and ANOVA tests are regarded as robust against non-normal data [75,76]. In addition, transforming non-normal scores closer to normality (suggested as a comparison by Bryman [77]), did not change the results of the mixed ANOVA tests. There were approximately equal numbers of children for each age group and gender as independent variables to ensure that statistical assumptions of homogeneity of variance were tolerable [78]. Where Mauchley’s test of sphericity was significant, the Greenhouse–Geisser correction was used.

To address the third research question, a series of Spearman’s Rho Correlations were run, directly comparing self-and teacher-reports for individual children. This non-parametric test was selected because of the non-normal distribution and the ability to determine monotonic relationships between variables in this exploratory research.

Analyses were conducted in SPSS (IBM SPSS Statistics for Windows, Version 27.0. Armok, NY, USA: IBM Corp.). Power analysis in SPSS and G*Power (version 3.1.9.6; Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany) showed that, even when split by age and gender, sample sizes were sufficient to detect medium effects.

Due to the exploratory nature of this research, it was not necessary to apply Bonferroni corrections [65,79,80,81]. That said, adjustments to the mixed ANOVA and correlations would mean that only results with p < 0.001 were significant; and therefore, some caution has been applied to the interpretation of results where this is not the case.

3. Results

3.1. Self-Reports

The results in this section address the first research question—how do children’s self-reports vary by behaviour subtype, age group, and gender?

3.1.1. ANOVA Results from Self-Reports

Table 2. Results of mixed Analysis of Variance (ANOVA) for self-reports.

ANOVA Results	df	F	ηp2	p
	Solitary behaviour
Behaviour subtype	1	3.78	0.02	0.011 *
Gender	3	5.01	0.03	0.026 *
Age group	3	0.02	0.00	0.887
Behaviour subtype × age group	1	0.21	0.00	0.887
Behaviour subtype × gender	3	0.17	0.00	0.363
Age group × gender	1	0.31	0.00	0.554
Behaviour subtype × age group × gender	3	0.22	0.00	0.881
	Prosocial behaviour
Behaviour subtype	2.78	7.44	0.04	<0.001 ***
Gender	1	0.13	0.00	0.724
Age group	1	11.60	0.06	<0.001 ***
Behaviour subtype × age group	2.78	7.73	0.04	<0.001 ***
Behaviour subtype × gender	2.78	0.35	0.00	0.777
Age group × gender	1	0.13	0.00	0.724
Behaviour subtype × age group × gender	2.78	2.49	0.01	0.064
	Aggressive behaviour
Behaviour subtype	2.62	5.30	0.03	0.001 ***
Gender	1	11.31	0.06	<0.001 **
Age group	1	15.75	0.08	<0.001 **
Behaviour subtype × age group	2.62	1.620	0.01	0.176
Behaviour subtype × gender	2.2	0.684	0.00	0.543
Age group × gender	1	6.310	0.03	0.013 *
Behaviour subtype × age group × gender	2.62	0.745	0.00	0.509
	Victimisation
Behaviour subtype	3	0.79	0.00	0.499
Gender	1	2.49	0.01	0.116
Age group	1	4.48	0.01	0.036 *
Behaviour type × age group	3	1.68	0.01	0.170
Behaviour type × gender	3	0.27	0.00	0.838
Age group × gender	1	0.25	0.00	0.620
Behaviour subtype × age group × gender	3	3.27	0.02	0.022 *

Note: F: ANOVA F ratio; df: degrees of freedom; ηp²: partial eta squared effect size. * p < 0.05; ** p < 0.01; *** p < 0.001

shows main effects and interactions from the mixed ANOVA conducted on self-reports. Significant results had small to medium effect sizes. With the exception of victimisation, there was a main effect of behaviour subtype for all behaviours. Gender played a role in self-reports of solitary behaviour and aggressive behaviour. Age group differences were evident in self-reports of prosocial and aggressive behaviour. There was also a main effect of age group and interaction between behaviour subtype, age group, and gender, for victimisation, but these should be treated with caution due to very small effect sizes and low significance.

Bonferroni post hoc tests were used to understand significant main effects, and plots and contrasts were used to interpret significant interactions. These are discussed in the following sections.

3.1.2. Significant Main Effects of Behaviour Subtypes on Self-Reports (Based on Table 2)

Results of Bonferroni post hoc tests (see

Table 3. Estimated Marginal Means (EMMs) and Standard Errors (SEs) for behaviour subtypes (self-reports).

	EMM	SE
	Solitary behaviour
Unsociable	1.84 *a	0.06
Contextual solitude	1.79	0.05
Shyness	1.67	0.06
Active isolation	1.62 *a	0.06
	Prosocial behaviour
Helping	2.79 ***ab	0.04
Sharing	2.73	0.04
Including	2.59 ***a	0.05
Caring	2.57 ***b	0.05
	Aggressive behaviour
Direct relational	1.35 *a	0.05
Indirect relational	1.30 **b	0.04
Verbal	1.25 *c	0.04
Physical	1.15 *ac**b	0.03
	Victimisation
Direct relational	1.77	0.05
Indirect relational	1.69	0.06
Verbal	1.79	0.06
Physical	1.78	0.06

Note: Significant differences are shown in bold, and letters (a,b,c) denote where there were significant differences within each behaviour type. * p < 0.05; ** p < 0.01; *** p < 0.001. Victimisation results were not significant but are included for comparative purposes.

) showed that self-reports for some behaviour subtypes differed from each other. For solitary, unsociable was reported significantly more frequently than active isolation but there were no other significant differences. For prosocial, helping was reported significantly more than caring and including. For aggression, physical aggression was self-reported significantly less than all other subtypes.

3.1.3. Significant Main Effects of Gender and Age Group Effects in Self-Reports (Based on Table 2)

Boys reported higher levels of solitary behaviours (EMM = 1.81, SE = 0.05) than girls (EMM = 1.65, SE = 0.05) and higher levels of aggressive behaviours (EMM = 1.35, SE = 0.04) than girls (M = 1.17, SE = 0.04).

Children in the older age group reported higher levels of prosocial behaviours (EMM = 2.76, SE = 0.04) than children in the younger age group (EMM = 2.76, SE = 0.04). Older children also reported higher levels of victimisation experiences (EMM = 1.84, SE = 0.05) than children in the younger age group (EMM = 1.68, SE = 0.05). For aggressive behaviour, children in the older age group (EMM = 1.15, SE = 0.04) reported lower levels than children in the younger age group (EMM = 1.37, SE = 0.04).

3.1.4. Significant Interactions in Self-Reports (Based on Table 2)

Prosocial behaviour self-reports of sharing followed a different pattern to self-reports of other prosocial behaviours and was lower in the older group than the younger group (see Figure 1). Contrasts showed that sharing was significantly different to all other behaviours (at least p < 0.01).

For aggressive behaviour, boys in the younger age group (EMM = 1.53, SE = 0.05) reported higher levels of overall aggression than boys in the older age group (EMM = 1.17, SE = 0.06). Boys’ ratings were also higher than their female counter-parts, although girls in the younger age group (EMM = 1.21, SE = 0.05) reported higher levels than girls in the older age group (EMM = 1.13, SE = 0.06).

For victimisation experiences, there were similar patterns across gender and age group for each subtype of victimisation, with the exception of physical, where boys’ self-reports were higher for the younger age group (p < 0.005) (see Figure 2).

3.2. Teacher-Reports

The results in this section address the second research question which considers how teacher-reports vary by subtypes of behaviour, age group, and gender; and enables comparisons with self-reports.

3.2.1. ANOVA Results from Teacher-Reports

Table 4. Results of mixed ANOVA for teacher-reports of behaviour.

	df	F	ηp2	p
	Solitary behaviour
Behaviour subtype	2.74	17.16	0.00	0.021 *
Gender	1	0.01	0.00	0.937
Age group	1	2.56	0.01	0.111
Behaviour subtype × age group	2.74	17.16	0.08	<0.001 ***
Behaviour subtype × gender	2.74	1.44	0.01	0.232
Age group × gender	1	0.43	0.00	0.511
Behaviour subtype × age group × gender	2.74	1.38	0.01	0.250
	Prosocial behaviour
Behaviour subtype	2.47	37.55	0.17	<0.001 ***
Gender	1	11.98	0.06	<0.001 ***
Age group	1	55.31	0.23	<0.001 ***
Behaviour subtype × age group	2.47	6.09	0.31	<0.001 ***
Behaviour subtype × gender	2.47	1.00	0.01	0.394
Age group × gender	1	0.40	0.00	0.528
Behaviour subtype × age group × gender	2.47	3.85	0.02	0.015 *
	Aggressive behaviour
Behaviour subtype	2.67	24.44	0.11	<0.001 ***
Gender	1	0.10	0.00	0.750
Age group	1	22.21	0.11	<0.001 ***
Behaviour subtype × age group	2.67	10.64	0.05	<0.001 ***
Behaviour subtype × gender	2.67	10.90	0.05	<0.001 ***
Age group × gender	1	1.96	0.01	0.160
Behaviour subtype × age group × gender	2.67	3.63	0.02	0.016 *
	Victimisation
Behaviour subtype	2.60	20.97	0.10	<0.001 ***
Gender	1	2.44	0.01	0.120
Age group	1	6.49	0.89	0.012 *
Behaviour subtype × age group	2.60	3.92	0.02	0.012 *
Behaviour subtype × gender	2.60	3.53	0.02	0.020 *
Age group × gender	1	0.08	0.00	0.777
Behaviour subtype × age group × gender	2.60	2.68	0.01	0.055

Note: F: ANOVA F ratio; df: degrees of freedom; ηp²: partial eta squared effect size. * p < 0.05; *** p < 0.001.

shows main effects and interactions from the mixed ANOVA conducted on teacher-reports. In most cases, significant results had medium to large effect sizes (greater than self-reports).

Table 4 shows that there was a main effect of behaviour subtype for all four behaviours, but the finding for solitary behaviour should be treated with caution due to the very small effect size and low significance. Age group and gender both had an effect on teacher-reports of prosocial and aggressive behaviours, although interactions between these and behaviour subtypes yielded small effect sizes. Age group also had an effect on teacher-reports of solitary behaviour and victimisation experiences. Gender also had an effect on different forms of victimisation but should be treated with caution due to the low effect size and low significance.

Bonferroni post hoc tests were used to understand significant main effects, and plots and contrasts to interpret significant interactions. These are discussed in the following sections.

3.2.2. Significant Main Effects and Significant Interactions of Age Group and Behaviour Subtype in Teacher-Reports (Based on Table 4)

Teachers reported higher levels of prosocial behaviour in the younger age group (EMM = 3.99, SE = 0.07) than children in the older age group (EMM = 3.24, SE = 0.07). Teacher-reports of aggression were higher in the older age group (EMM = 2.16, SE = 0.09) than in the younger age group (EMM = 1.61, SE = 0.09). This was also the case for victimisation, where teacher-reports for the older age group children (EMM = 1.75, SE = 0.07) were higher than younger age group children (EMM = 1.51, SE = 0.06).

For solitary behaviour, post hoc tests (see

Table 5. EMMs and SEs for behaviour subtype × age group (teacher-reports).

	EMM	SE	EMM	SE	EMM	SE
	Whole sample		Younger		Older
	Solitary behaviour
Contextual solitude	1.76	0.07	1.85 ***f	0.10	1.66 ***f	0.11
Shyness	1.74 **a	0.06	1.51 ***fg	0.09	1.97 ***fg	0.11
Unsociable	1.68	0.07	1.80 ***gh	0.10	1.57 ***gh	0.11
Active isolation	1.54 **a	0.67	1.23 ***fh	0.09	1.86 ***fh	0.10
	Prosocial behaviour
Sharing	3.92 ***ad	0.07	4.10 *fg	0.10	3.72 *fg	0.10
Caring	3.85 ***be	0.06	4.38 *gh	0.08	3.31 *gh	0.09
Helping	3.64 ***cde	0.07	4.10 *f	0.10	3.18 *f	0.10
Including	3.07 ***abc	0.09	3.39 *h	0.12	2.76 *h	0.13
	Aggressive behaviour
Indirect relational	2.16 ***ab	0.08	1.73 ***f	0.11	2.38 ***f	0.08
Direct relational	2.05 ***cd	0.08	1.72 ***g	0.11	2.60 ***g	0.08
Verbal	1.77 ***ace	0.07	1.43 ***h	0.10	2.08 ***h	0.07
Physical	1.57 ***bde	0.07	1.54 ***fgh	0.09	1.60 ***fgh	0.07
	Victimisation
Indirect relational	1.96 ***abc	0.08	1.69 *fg**h	0.11	2.32 *fg**h	0.11
Direct relational	0.61 ***ad	0.07	1.56 *f	0.09	1.72 *f	0.10
Verbal	1.49 ***b	0.06	1.44 **h	0.08	1.54 **h	0.08
Physical	1.43 ***cd	0.05	1.36 *g	0.07	1.49 *g	0.08

Note: * p < 0.05; ** p < 0.01; *** p < 0.001. Letters a–e denote where there are significant differences between means within each behaviour type. Letters f–h show results from contrasts where the younger/older means were significantly different across behaviour subtypes.

) showed that active isolation was reported less than shyness but there were no significant differences with or between the other subtypes of solitary behaviour. For prosocial, including was significantly lower than all other subtypes. For prosocial, helping was also significantly lower than sharing and caring. There were higher reports of relational aggression than non-relational aggression, and physical aggression was statistically lower than all other subtypes of aggression. For victimisation, teacher-reports of indirect relational victimisation were higher than other subtypes.

Table 5 also shows post hoc test results from an interaction between behaviour subtype and age group for all four behaviours. For solitary, teacher-reports of contextual solitude and unsociable were similar in both age groups; and reports of shyness and active isolation were similar in both age groups. These showed opposite patterns in each age group, with the former higher for the younger group and the latter higher in the older group. For prosocial, there were significant differences in terms of: sharing from helping, sharing from caring,; and caring from including across the two age groups. Caring was the highest teacher-reported behaviour in the younger group and sharing was highest among the older group. For aggressive behaviour, there was a significant difference between physical aggression and the other aggressive behaviours, with greater differences in the older age group, than the younger age group. For victimisation, teacher-reports of indirect relational victimisation were significantly higher than the other subtypes for the older age group. Whilst these were also the highest in the younger age group, there was a smaller difference between this and the other victimisation subtypes.

3.2.3. Significant Main Effects of Gender and Significant Interactions with Age Group and Behaviour Subtype in Teacher-Reports (Based on Table 4)

For prosocial behaviour, there were higher teacher-reports of prosocial behaviour in girls (EMM = 3.79, SE = 0.07) than boys (EMM = 3.44, SE = 0.07).

For aggressive behaviour (see Figure 3), teacher-reports of relational aggression were higher for girls than boys, and physical aggression was lower for girls than boys. There was little difference in teacher-reports of verbal aggression for boys and girls. For victimisation (see Figure 3), the effect size reports of physical victimisation were lower than the other victimisation subtype for girls. For boys, physical victimisation occurred more than verbal.

For prosocial behaviour, all subtypes were higher in the younger age group than the older age group, but this difference was greater for boys than girls (with the exception of including). Teacher-reports of aggressive behaviour were higher in the older age group but the differences between boys and girls were higher for relational than for non-relational aggression (where reports were higher in girls). This result should be treated with caution due to a small effect size. A breakdown of results can be seen in Figure 4.

3.3. Correlations Between Self- and Teacher-Reports of Behaviour

Spearman’s correlations were run to compare individual self- and teacher-reports (the third research question), split by age group and gender, across the 16 individual behaviours. The significant correlations reported below were generally weak (see

Table 6. Spearman correlations (r_s) between self- and teacher-reports.

Behaviour	Overall rs (n = 193)	Younger (n = 102)	Older (n = 91)	Boys (n = 96)	Girls (n = 97)
Solitary behaviour
Shyness	−0.09	−0.15	−0.01	−0.05	−0.17
Unsociable	0.15 *	0.25 *	0.04	−0.01	0.32 ***
Active isolation	0.03	0.13	−0.10	0.03	0.03
Contextual solitude	0.01	−0.01	0.12	0.12	0.04
Prosocial behaviour
Sharing	0.10	0.11	0.03	0.02	0.19
Caring	−0.08	−0.03	0.17	−0.01	−0.10
Helping	−0.56	0.05	0.11	−0.06	−0.09
Including	−0.26	0.05	−0.06	0.03	−0.02
Aggressive behaviour
Direct relational	0.08	0.09	0.12	0.03	0.17
Indirect relational	−0.01	0.06	0.14	0.01	−0.03
Verbal	0.06	0.12	0.11	0.13	−0.08
Physical	0.20 ***	0.27 **	0.12	0.24 **	−0.00
Victimisation
Direct relational	0.15 *	0.15	0.16	0.09	0.22 *
Indirect relational	0.16 *	0.19	0.07	0.07	0.25 *
Verbal	0.03	0.01	0.10	−0.15	0.20
Physical	0.03	0.12	−0.08	−0.01	−0.01

Note. * p < 0.05; ** p < 0.01; *** p < 0.001

).

For solitary behaviour, unsociable reports were significantly and positively correlated among younger children but not older children. They were also significantly and positively correlated between teacher-reports and self-reports for girls but not boys. For prosocial behaviour, there were no significant correlations between children’s self-reports and teacher-reports. For aggressive behaviour, there were significant positive associations between self- and teacher-reports of physical aggression, although this only held for the younger age group when the sample was split by age, and boys when split by gender. For reports of victimisation there was a low significant positive correlation between teacher- and self-reports of direct and indirect relational victimisation only, which only held for girls and not boys.

4. Discussion

4.1. Overview of Findings

The findings from this exploratory study make an important contribution to understanding self-reports of behaviour in young children. It was unique in looking across four areas of behaviour—solitary, prosocial, and aggressive behaviours; and victimisation experiences—and was interested in understanding the patterns within these self-reports and how these compared with teacher-reports (between subtypes and across age group and gender). The findings show that that young children provide unique insights into their own behaviours. This is discussed by each behaviour concept below.

4.2. Reports of Solitary Behaviour

4.2.1. Self and Teacher-Reports of Solitary Behaviour

The current study shows that young children provide important insights on their own solitary behaviour. There was a small effect of behaviour subtype on children’s self-reports where children’s reports of unsociable behaviour (reported the most) and active isolation (reported the least) were significantly different from each other. This partially supports previous research which suggested that 3–6-year-olds can distinguish between subtypes of solitary behaviour in hypothetical scenarios [54]. However, it may also suggest that there is either some overlap in children’s display of solitary behaviour or generalisation in their reports. It is possible that this is because unsociability and active isolation are particularly salient motivations for being solitary, and that young children may find it difficult to distinguish these from the other less salient behaviours of shyness and contextual solitude. For teacher-reports, there were age group differences which did not occur for self-reports. Contextual solitude and unsociable behaviour were reported by teachers as higher in the younger age group than among older children. This, along with the lack of agreement between self- and teacher-reports when correlating individual responses, is in line with previous research with slightly older children [32]. This demonstrates that children provide a different perspective to their teachers when reporting on these behaviours. This may be due to situational factors. For instance, it is possible that unsociable behaviour and contextual solitude are more salient to teachers in the younger age group because of the play-based classroom environment [25,26], but may still occur for children in other school contexts, potentially out of sight of teachers [28]. In addition, children’s own interpretation of their solitary experience may differ from those around them.

4.2.2. The Role of Gender in Reports of Solitary Behaviour

The idea that children offer a unique perspective on their solitary behaviour is also evident when looking at gender differences. Whilst teacher-reports were consistent with previous literature in not showing any gender differences [11], it is noteworthy that there was a small effect of gender on self-reports, where boys self-reported higher levels of solitary behaviour than girls. This suggests that boys’ reports may be unaffected by social expectations and that some solitary behaviours (e.g., shyness) are seen as more acceptable in girls [11]. When correlating individual reports of solitary behaviour, there was some agreement in reports of unsociable behaviour for girls only, adding weight to the idea that teachers may have less awareness of solitary behaviour in boys. This is somewhat concerning given the associated negative outcomes for boys [82]. Given that teacher-reports of aggressive behaviour were higher in boys (see Section 4.4), it is possible that the saliency of aggressive behaviours overshadows their awareness of when boys are alone. In addition, teachers have been found to report higher quality relationships with girls than boys [83]. Perhaps this is reflected in the teacher-reports where they have an increased awareness of less salient behaviours in girls.

4.3. Reports of Prosocial Behaviour

4.3.1. Self and Teacher-Reports of Prosocial Behaviour

The unique insight and understanding of young children are also evident when looking at the findings for prosocial behaviour. As in previous research with older children, [57], there was very little agreement in the individual reports of prosocial behaviour from children and their teachers. Furthermore, there were differences when comparing overall patterns. Our findings were similar to previous research looking at how children distinguish between prosocial behaviours in hypothetical peers [56]. Caring was reported the least frequently across the whole sample. In addition, whilst helping was reported as significantly higher than caring or including, there were no other significant differences between the self-reports of prosocial behaviour, suggesting that this age group may find it difficult to distinguish between some prosocial behaviours. These findings differed from teacher-reports, where there was greater differentiation. Including and helping were reported as less prevalent than sharing and caring. As such, it is possible that children generalise in their accounts of prosocial behaviour in comparison to teachers.

That said, children’s understanding of sharing behaviour seemed to differ from other prosocial behaviours, as the older age group reported less sharing than the younger age group (but more of other prosocial subtypes). As children progress through school, there are fewer opportunities within the classroom for free play activities [25,26], and therefore, perhaps fewer opportunities to share, which seems to be evident in children’s self-reports. It may be that the older group’s understanding of sharing differs from their perceptions of the other prosocial behaviours, which was not evident in teacher-reports.

4.3.2. Age Group and Gender Differences in Reports of Prosocial Behaviour

Self-reports of prosocial behaviour displayed different patterns to teacher-reports when looking at age group and gender differences. Whilst teacher-reports of prosocial behaviour were lower in the older group for all behaviour subtypes, the opposite pattern was evident in self-reports (with the exception of sharing). It is possible that self-reports are influenced by an increased awareness of social expectations as children become older [33]. An alternative explanation is that prosocial behaviour increases [5] and becomes more varied with age [84], where children show more caring, helping, and including (but less sharing), and that this is reflected in children’s self-reports. It is also possible that situational biases influence teacher-reports and that the change in play-based settings [25,26] means they witness fewer prosocial behaviours in the classroom, which may occur elsewhere in school [55].

Teacher-reports were similar to research showing higher levels of prosocial behaviour in girls than boys [7], but there were no gender differences in self-reports. Whilst it is possible that boys over-report their prosocial behaviour [57], this may also be because young children tend to not hold gender stereotypes about prosocial behaviour [58] compared to teachers [57]. Therefore, young children may be able to provide less biased reports of their prosocial behaviour than their teachers.

4.4. Reports of Aggressive Behaviour

4.4.1. Self and Teacher-Reports of Aggressive Behaviour

Our findings in relation to aggressive behaviour suggest that self-reports make an important contribution. Overall, there was some agreement when correlating self-reports and teacher-reports of physical aggression, and this was reported as occurring less frequently by both children and their teachers. Similarly, both self- and teacher-reports showed that relational aggression was the most common form of aggression (although patterns differed between direct and indirect relational aggression). Young children have been found to be equally accepting of physical and relational aggression [85]. Therefore, these reports seem to accurately reflect the aggressive behaviour in this cohort, rather than being affected by their perceptions of social acceptance. However, the lack of overall correlations, when comparing self- and teacher-reports of aggressive behaviour, and the consideration of age group and gender differences, indicate that there may be some complexities in these reports.

4.4.2. Age Group and Gender Differences in Reports of Aggressive Behaviour

Findings showed that self-reports of aggressive behaviour were lower in the older age group, but the opposite pattern was evident in teacher-reports. In addition, the self-report and teacher-report agreement relating to physical aggression, was only significant in the younger age group. This shows some disagreement between teachers and children when reporting on aggressive behaviour [5,64].

Looking at self-reports specifically, on one hand, our findings support previous research that overt aggressive behaviour decreases with age [5], as children’s cognitive and language abilities develop [43,44], or that children have a decreased acceptance of aggressive acts [85]. However, it is also possible that older children are more selective in their reports of their own aggressive behaviour and that increasing social desirability [68] directly biases children’s responses.

An analysis of gender differences also raised the same questions, particularly in relation to girls. In line with previous research [57], we found that boys self-reported higher levels of aggressive behaviour than girls (particularly in the older age group). Perhaps girls engage in less aggressive acts because it is seen as less socially acceptable for girls to show some aggressive behaviours [58,86], and girls are more aware of and affected by the potential consequences of aggressive behaviour [28]. However, it is also possible that girls’ self-reports are directly affected by the saliency of the behaviours and the associated social desirability. Previous research has shown that relational aggression occurs more than other forms of aggression in girls [69,70]. Therefore, it may be that girls’ self-reports are lower because they think that their display of this behaviour is less obvious to others. This idea is supported by the finding that teachers reported higher relational aggression for girls (particularly in the older age group), which was not present in self-reports.

It is also interesting that boys self-reported higher levels of solitary (see Section 4.2) and aggressive behaviour than girls. Previous research has shown relationships between aggression and solitary behaviour [87], which supports the idea of both higher solitary and aggressive behaviour reports amongst boys and warrants further investigation into behaviour profiles.

4.5. Reports of Victimisation Experiences

4.5.1. Age Group and Gender Differences in Self- and Teacher-Reports of Victimisation Experiences

When looking at the effect of age group and gender on victimisation overall, unlike the other behaviours, our research showed similar findings across children’s self-reports and teacher-reports. Both indicated that experiences of victimisation were higher in the older age group than younger age group. Whilst the cross-sectional nature of this research meant that it was not possible to draw direct comparison, this aligns with ideas from previous research that has suggested that the stability of victimisation occurs later in early childhood [88] than in our sample. That said, this was a much smaller effect size in self-reports than teachers, also highlighting the variability in children’s reports. Neither self-reports nor teacher-reports showed significant differences between boys and girls, suggesting that boys and girls are equally victimised, also aligning with previous research [69].

4.5.2. Comparing Self and Teacher-Reports of Individual Victimisation Experiences

Despite the patterns in effects of age group and gender, our findings indicated that children’s self-reports of victimisation experiences are unique and individualised. When looking across subtypes, there did not seem to be any clear patterns in the self-reports, and it is possible that these reflect the varied (and sometimes brief [64]) victimisation experiences that children may encounter. It is possible that their perspectives are influenced by factors such as their social and emotional learning, or friendships [30,31]. In contrast, the teacher-reports of victimisation experiences showed similar patterns to the teacher-reports of aggressive behaviour. For instance, they reported higher levels of direct-relational victimisation than physical and verbal victimisation. This suggests that children’s self-reports may offer a deeper insight to their victimisation experiences than teachers are able to provide. It is possible that teachers generalise across aggression and victimisation, rather than fully understanding children’s own interpretations of these experiences. This idea is supported by the finding, in line with earlier research [38,65], that there was little to no agreement between self- and teacher-reports of victimisation experiences for individual children. Where agreement did occur, this was only the case for relational victimisation amongst girls.

4.6. Summary of Findings Across Behaviours

Our findings highlight the importance of including children’s self-reports when studying their behaviour. For solitary behaviour, findings suggested more generalisation in teacher-reports than self-reports and less teacher awareness of solitary behaviour amongst boys. For prosocial behaviour, there was some evidence to suggest that children differentiated caring (younger group) and sharing (older group) from other forms of prosocial behaviour in their self-reports, but there is more generalisation between the other subtypes. Whilst there may be some over-reporting amongst children, particularly as they become older, there are also potential situational and gender biases in teacher-reports. Our findings also show that there is scope to consider self-reports of children’s aggressive behaviour more closely. On one hand, there is some agreement between teachers and children about the display of aggressive behaviour. However, on the other hand, our findings showed self-reports of less frequent aggressive behaviour from older children, and girls. There is scope to understand more fully whether this is reflective of their behaviour perceptions or biases in their self-reports. Finally, whilst children and teachers agree that victimisation experiences are generally higher in the older age group and occur equally across boys and girls, our findings also suggest that children offer unique views on the complexities of their own different types of victimisation experiences. Comparing the effect sizes of age and gender across self- and teacher-reports shows that these are more important in teacher-reports than self-reports. This further demonstrates the unique perspective that children can provide, and suggests that children offer an important additional perspective to their teachers when asked about their behaviours at school [28]. This may occur for a multitude of reasons, such as self-reports being influenced by social expectations [33] and desirability [68], situational [24] or gender biases [57] amongst teachers, or other individual differences amongst children [31]; however, this emphasises the need to take a multi-informant approach.

The current research has highlighted the importance of understanding that children’s self-reports and perceptions of their own behaviour may differ from others. Not only are these useful to consider in future research but can also make an important contribution to practitioners working with young children, both in terms of day-to-day interactions and when providing reports about pupils to others (e.g., parents/carers). In addition, there is scope to consider the current findings in policymaking, such as recognising children’s unique reports in school behaviour management policies.

4.7. Strengths, Limitations, and Future Directions

A particular strength of this research was the use of innovative and engaging animations in collecting young children’s self-reports. As with all research with young children, this approach was both time- and resource-intensive, which meant that our sample was limited to 193 children. Whilst this prevented some further analysis, the sample was adequate to provide reliable findings. As mentioned, Bonferroni corrections were not applied due to the exploratory nature of the research. Where appropriate, we exercised caution with results, but it is possible that some findings may have resulted from multiple testing. Therefore, the use of a larger sample and statistical adjustments would help to demonstrate greater statistical power whilst controlling for type 1 errors.

We collected reports on the same 16 individual behaviours across self-reports and teachers, meaning that comparisons could be made to inform our exploration of self-reports. For methodological reasons, a three-point scale was used for self-reports and a five-point scale for teacher-reports. This did not impact upon the robustness of statistical analysis, but there is scope to make use of the same scale to look specifically at agreement scores in future research. In relation to the reliability of the data, research has shown moderate agreement between teacher-reports and other adults in school when using the method in the current study [23]. Other research has shown test–retest reliability in 5–7-year-old children’s self-reports over 7–10 days, using other interactive methods [89]. However, given the potential situational impact on behaviour reports [24], it would be useful to carry out specific reliability and stability tests in future research. It is also worth noting that, teacher-reports often focus on the behaviour they witness in class, rather than other environments such as the playground [55]. Others in school, such as peers and teaching assistants, could provide this insight that teachers may not have [37], and there is scope to include these in future research. In addition, the cross-sectional nature of the current study meant that we were unable to look at the same children over time. It would be useful to study self-reports longitudinally, to consider whether age-group differences in reports still occur.

The findings from this study highlight several areas of possible future research. Firstly, self-reports of solitary behaviour and victimisation showed particularly unique perspectives from young children and there is scope to explore this further. In addition, given the potential negative outcomes for boys, there would be value in delving deeper into their self-reporting of solitary and aggressive behaviour as a behaviour profile [90]. Exploration of other self-report profiles, such as prosocial and aggressive behaviours [91] and aggression and victimisation [38], may also expand upon the unique perspective that children’s self-reports have been shown to provide. For teacher-reports only, there were comparatively low reports of including behaviour. Given the rise of interventions, such as the ‘Buddy Bus-stop’ or ‘Buddy-Bench’ [92], this is a surprising finding. Including is not traditionally incorporated into studies of prosocial behaviour [61], and considering the findings that reports of active isolation were comparatively low in both teacher-reports and children’s self-reports of solitary behaviour, there is scope for more exploration about how these may relate to each other. Whilst a key strength of the current study was the separate exploration of subtypes of each behaviour, the development of profiles with so many individual behaviours would be complex. Given the reliability of collating teacher-reports and children’s generalisations across some subtypes of behaviour, there is scope to look at this further and collapse some components to explore behaviour profiles.

There is also value in considering how self-reports and perspectives of behaviour relate to children’s outcomes and wellbeing. Exploration of how children think others see their behaviour could also help to understand how far social desirability and expectations influence both their own behaviours and reports.