A mixed-method study comparing drawing and interview data and their association with self-report involvement in bullying events

Bullying continues to be a social issue affecting millions of students of all ages worldwide. Research on bullying seems to be dominated by quantitative research approaches employed standardized categories and measures, ultimately limiting our knowledge about children’s own view on bullying. Our research follows another direction, aiming to explore the representation of bullying in a sample of Italian primary school children by using and comparing the functioning of two qualitative research instruments: interviews, and children’s drawings. In addition, aided by quantitative analyses, we aimed to investigate whether students’ involvement in different bullying roles (as bullies, victims, or defenders), as measured by self-assessment, correlated with different characteristics of the representation of bullying emerging from children’s drawings and interviews. We recruited a convenient sample of 640 primary school students (mean age = 9.44; SD = 0.67), 53.3% of whom were male. The results showed that all forms of bullying, i.e., physical, verbal, and social bullying, could be identified in interview and drawing data, although references to all types of bullying were more frequent in interview data. In terms of bullying criteria, the presence of a power imbalance between the bully and the victim was most frequently detected in both the interview data and the drawing data, while repetition was more easily detected in the interview data. The interview data showed that sadness was the most frequently reported victim emotions, followed by fear, anger, and lack of emotion. The drawing data showed a similar pattern, although victims were more frequently described as lacking emotions compared to the interview data. In both interview and drawing data, age and female gender were positively associated with references to verbal bullying, and negatively associated with references to physical bullying. Additionally, bully/victim children were more likely than uninvolved children to depict physical bullying in the drawings, while this association was not detected in interview data. In summary, our study shows that, compared with drawings, interviews tend to provide a more comprehensive view of children’s own representation of bullying, while drawing data tend to show stronger connections with children’s current personal experiences of bullying.


School is a context in which various forms of victimization can occur (Longobardi et al., 20172019Badenes-Ribera et al., 2022). Bullying continues to be a social issue affecting millions of students of all ages worldwide (Craig et al., 2009Ossa et al., 2021) which tends to be associated with poorer developmental and academic outcomes for affected children (Moore et al., 2017Prino et al., 2019Fabris et al., 2021). Bullying is usually defined as a frequent, repeated, and intentional form of aggression involving an imbalance of power or strength between the bully and the victim (Olweus and Limber, 2010) which may be exerted in different forms, including physical aggression (e.g., being hit, kicked, pushed, shoved), verbal aggression (e.g. being insulted or called nasty and hurtful names or threatened) or social exclusion (e.g., being ignored or excluded from peer groups). Bullying is also considered a group phenomenon in which one or more individuals (bullies) repeatedly and intentionally attack, humiliate, or exclude others (victims) who have difficulty fighting back (Salmivalli, 2010). The social scene of bullying is complex, and peers may participate not only as victims, bullies, or bullies-victims, but also play other roles. There are assistants to the bully who join the ringleaders to attack a victim; reinforcers who are not actively involved in the bullying but are instrumental to the actions of the bully; defenders who actively intervene and try to stop the bullying (e.g., demand teacher intervention or try to comfort the victim); and bystanders who know what is happening but do not take sides with either the bully or the victim (Salmivalli, 2010).

Although peer aggression usually peaks in early adolescence (Perry et al., 1988), forms of bullying can also occur in primary school, and some studies have found certain gender differences in the prevalence of involvement. In particular, males appear to be at heightened risk of being involved in bullying, both as bullies and as victims, while females tend to report more often forms of indirect victimization (Iossi Silva et al., 2013Ang et al., 2018Twardowska-Staszek et al., 2018Jiménez, 2019) and are more likely to engage in forms of indirect bullying such as active social exclusion (Ang et al., 2018).

The majority of research currently conducted on the topic of children’s involvement in bullying appears to favor quantitative studies (Patton et al., 2017Marengo et al., 2021Samara et al., 2021). Quantitative studies based on self-report questionnaires allow us to increase our knowledge in large samples in a comparable way. However, they have the limitation of not revealing the subjective experiences of the children involved. Unlike quantitative surveys, qualitative research is typically inductive and therefore lends itself to an in-depth exploration of the perspectives, perceptions, and experiences of children involved in bullying (Bosacki et al., 2006Patton et al., 2017). This is a central aspect of bullying research because it allows researchers to examine more closely how children perceive and define bullying, which has concrete implications for intervention and prevention strategies. Along these lines, qualitative research allows us to examine children’s representations of bullying, integrate and extend data from quantitative research, and thus inform researchers and practitioners about intervention and prevention strategies (Torrance, 2000Espelage and Asidao, 2001Patton et al., 2017). In practice, as several authors point out (Bosacki et al., 2006), quantitative research forces the child to answer questions designed and suggested by researchers, whereas qualitative research allows children to express their own perspectives and highlight the aspects of the phenomenon that are most important to them. In doing this, we may be more able recognize that there is no single, common definition of bullying and that the definition of bullying, and thus the perception of the phenomenon, may vary between students and adults, such as researchers, school staff, and teachers (Eriksen, 2018). Still, use of qualitative research is not without limitations, as it is typically more time consuming and requires more resources than quantitative research (e.g., personnel performing interviews, or the coding of collected data; Carter and Henderson, 2005). Lack of anonymity may also be another issue possibly introducing bias in the assessment procedure in terms of both lowering proneness to respond, as well as the need to do it in socially desirable way (Bergen and Labonté, 2020).

A common data collection approach in qualitative research is the use of in-person interviews (Silverman, 2016). Using interviews with primary school children, Guerin and Hennessy (2002)found that verbal and physical bullying were the most common forms of bullying in children’s narratives, followed by forms of bullying characterized by threats, spreading rumors, and social exclusion. However, in contrast to the definitions proposed by researchers, it appears that repetition, intent, and lack of provocation are not central to definitions of bullying by students (Madsen, 1996Guerin and Hennessy, 2002Monks and Smith, 2006Naylor et al., 2006), while harm inflicted on victims is a salient feature in their definition of bullying (Madsen, 1996Naylor et al., 2006).

Studies using interviews or open-ended questions also report age and gender differences in children’s representation of bullying. As for age, young children tend to differentiate only between non-aggressive and aggressive acts, viewing the latter as bullying even when they do not involve bullying behavior (e.g., children of equal power fighting over a misunderstanding); in turn adolescents and adults have a more complex understanding of bullying, successfully distinguishing between direct and indirect forms of physical, relational/social, and verbal bullying behaviors (Smith et al., 2002Monks and Smith, 2006Naylor et al., 2006). For example, in the school context, children are more likely to refer to direct victimization (physical and verbal) compared to their teachers, who in turn tend to refer also to indirect forms of bullying (e.g., social exclusion; Naylor et al., 2006).

When compared with males, females’ representations of bullying appear to focus more on the impact of bullying and the emotional well-being of the victims; in turn, males are more likely to describe observable behaviors that may occur in bullying incidents (Naylor et al., 2006Byrne et al., 2016). Some data show that females tend to report more verbal abuse than males (Naylor et al., 2006). In addition, Naylor et al. (2006) report that females tend to report social exclusion as a form of bullying more often than males; however, some studies do not support this finding (Guerin and Hennessy, 2002Smith et al., 2002).

Interviews are not the only qualitative techniques used in studying children’s representation of bullying among primary school students. Some surveys have used children’s drawings to identify children’s representations of bullying (Bosacki et al., 2006Patton et al., 2017). Children’s drawing appears to be a useful tool for research because it allows us to examine the representations and perceptions that children exhibit toward a particular topic of inquiry (Bozzato et al., 2021). Drawing is considered an attractive and entertaining activity for children (Kukkonen and Chang-Kredl, 2018). From a methodological perspective, children’s drawing could be an investigative tool that benefits children who have difficulty with verbal expression, and through drawing, the child can incorporate elements that are important to him or her in terms of representing the phenomena he or she depicts (DiCarlo et al., 2000MacPhail and Kinchin, 2004). In addition, several studies point to the importance of children’s drawings as a tool for assessing psychological well-being and the quality of interpersonal relationships (Bombi et al., 2007Laghi et al., 2013Potchebutzky et al., 2020Kallitsoglou et al., 2022).

Research on bullying conducted using the drawing technique shows some interesting findings, which we will summarize here. The vast majority of primary school children tend to draw the victim-perpetrator dyad, while children only begin to draw more than two people in the scene as they get older (Bosacki et al., 2006). It appears that children tend to draw the bullying scene protagonists with their own gender, while only a smaller percentage (10%) draw mixed-gender scenarios (in which a male typically bullies a female) (Bosacki et al., 2006). Children usually draw bullies either larger or the same size as the victim, while it is rare for the victim to be drawn larger than the bully (Bosacki et al., 2006Slee and Skrzypiec, 2015). Most children draw facial expressions for both the bully and the victim (Bosacki et al., 2006). For the bullies, the majority draw positive facial expressions, while only between 6 and 38% of them draw negative facial expressions (Bosacki et al., 2006Slee and Skrzypiec, 2015). Regarding victims, the majority of them are presented with a negative facial expression and to a lesser extent with a neutral or positive facial expression (Bosacki et al., 2006Slee and Skrzypiec, 2015). Many children also draw “speech bubbles” or verbal comments, and this appears to characterize younger children in particular (Bosacki et al., 2006). However, as Bosacki et al. (2006) note, not only do the depictions of verbal comments decrease as children get older, but older children are more likely to portray the victim as silent during bullying events. According to Andreou and Bonoti’s (2010) survey, nearly half of the children draw themselves in the bullying scene, as victim, bully, or defender, but not as helper or reinforcer of the bully. Girls tend to draw themselves into more verbal victimization scenes than boys, while boys tend to draw themselves as engaging physical acts of bullying. Furthermore, Andreou and Bonoti’s (2010) analysis shows that physical, verbal, or mixed (both physical and verbal) forms of victimization appear in the drawings, while other forms of violence, such as attacks on property or social exclusion, are not depicted.

One aspect that we believe is insufficiently explored in the literature is whether experiences with bullying (as victim, aggressor, or bystander) may be associated with children’s representation of bullying in some way. In this direction, evidence suggests that the experience of peer victimization does not appear to be associated with children’s definition of bullying in interviews (Monks and Smith, 2006Naylor et al., 2006). However, when confronted with vignettes depicting bullying, bullies tend not to recognize these aggressive behaviors as bullying (Monks and Smith, 2006). This could be because bullies’ moral restraint leads them to minimize negative emotions (such as shame and guilt) and emphasize positive reactions to bullying (Ortega et al., 2002). This would result in aggressive acts being less recognized and defined by bullies as bullying behavior (Monks and Smith, 2006). In a large sample of adolescents, Byrne et al. (2016) found that students who had experienced peer victimization were more likely to discuss the emotional impact of bullying on the victim in their definition of bullying compared to those who had not been victimized.

Regarding the relationship between self-reported bullying involvement and drawing in childhood, we are aware of only two studies that have attempted to examine possible correlates in primary school children (Andreou and Bonoti, 2010Slee and Skrzypiec, 2015). Andreou and Bonoti (2010) examined the correlation between self-report and bullying design and found a weak correlation. Slee and Skrzypiec (2015) found that children who were bullied tended to make more detailed drawings and depict less spatial distance between the figures of the victim and the bully. No significant relationship was found between the frequency of victimization and the size of the bully or victim and some graphic indices traditionally associated with emotional well-being, such as the size of the drawing and the weight of the lines. Overall, then, there is a need to explore the relationship between self-report and qualitative research instruments (particularly interviews and children’s drawings) to understand primary school children’s representation bullying. Furthermore, there is limited evidence of comparisons between different qualitative approaches such as drawing and interviewing to understand whether both instruments can be considered informative and whether they converge or diverge in terms of the information they provide about the child’s experience.

The present study

Research on bullying seems to be dominated by quantitative research approaches, thus disregarding important information about children’s representation of bullying and their involvement in the phenomenon. The present study seeks to fill this gap using a mixed method approach. Using two qualitative research approaches, namely the interview and children’s drawings, we collected data about the representation of bullying (and the characteristics that define it) in a sample of Italian primary school children. Based on collected data, and aided by quantitative methods, we sought to answer multiple research questions. First, we sought to determine if children’s representations of bullying emerging from interview and drawing data differed in significant way. We based this comparison on a set of bullying characteristics naturally emerging from the data, including the type of bullying behaviors enacted by the bully (e.g., physical, verbal, and relational bullying), their compatibility with commonly used criteria for bullying (i.e., repetition, power imbalance, and intentionality), the emotional and behavioral response of the victims, and the presence of other individuals (e.g., teachers and other children). Secondly, we wanted to understand if the children’s representations of bullying observed in their drawing and interview would be related to the demographic characteristics of the children, as well as their direct involvement in bullying episodes as measured using a self-report questionnaire. Thus, asking the children to describe their personal representation of bullying through their drawing and interview data, and self-report about their involvement bullying episodes, allowed us to examine how these experiences were associated with their view on bullying. It is noteworthy that, because of the lack of previous studies exploring the first research question (i.e., are there differences in bullying representations between drawing and interview data?), we considered this aim of the study as eminently explorative. In regard to our second research question (i.e., are gender, and age, and bullying experiences related to children’s representations of bullying?), tentative hypotheses may be derived from previous literature based on interview and drawing data. More specifically, we hypothesized that gender differences might emerge as regards the type of bullying described by children (Naylor et al., 2006Slee and Skrzypiec, 2015), with a prevalence of references to physical bullying being more frequent among males, and references to verbal bullying appearing more often among females (e.g., Naylor et al., 2006Andreou and Bonoti, 2010); we also expected victims of bullying to be more likely to refer to direct forms of bullying (i.e., physical and verbal aggression) when compared to students who had not been involved in bullying (Naylor et al., 2006). Following studies based on self-report data we expected that age might also show some associations with the type of bullying mentioned, with a decline in the mentioning of physical bullying and an increase in references to verbal bullying with increasing age (e.g., Marengo et al., 2019).

Materials and methods


We recruited a convenience sample of 640 primary school students attending grade 4 to 5 in 7 different public primary schools located in North-West Italy. The mean age was 9.44 years (SD = 0.61; range = 8–12) and 53.3% of recruited students were male. All recruited students were fluent in Italian language, and none of the children presented diagnoses related to intellectual deficits or forms of psychopathology that would compromise their ability to participate in the research.


The aims of the research were presented in the classroom to the students by the one of the researchers. Participation in the research was on a voluntary basis and no reward was provided to the children, their families or the school. Participating children were administered a protocol that included, in order, the production of a drawing relating to their experience of bullying, a semi-structured interview and a questionnaire relating to their experience of involvement in bullying as a victim, aggressor or bully-victim. Typically, all assessments (i.e., drawing, interview, and self-report questionnaire) were performed on the same day for all students of a specific classroom. Before test administration, the children had the opportunity to familiarize themselves with the researchers. The researchers who administered the protocol were psychologists with experience in the field of developmental and school psychology, who had received training in child drawing administration and interpretation, and had extensive research experience.

Ethical considerations

The study was approved by the IRB of the University of Turin (protocol no. 291061), and was undertaken in accordance with the indications of the Italian Association of Psychology (AIP) and the Helsinki Convention. After obtaining approval from the school headmaster, informed consent for participation in the research was obtained from both parents and children. The absence of informed consent from parents and children precluded the latter from participating in the research.


Children’s drawings

We involved students in a bullying drawing task in which children were asked to draw a picture portraying what bullying meant to them using the following standardized stem: “Please draw what bullying is like for you.” The children were given a white A4 sheet of paper and 12 colored crayons. No time limit was given to the children; however, children typically completed the task in less than an hour. Children completed the task in the classroom along with their peers; however, school desks were separated to avoid mutual interference. In contrast to other research (Bosacki et al., 2006), we did not ask the child to refer to his or her own experiences with bullying, but to describe what bullying was like for him or her by drawing it. In our opinion, this approach was instrumental in allowing the children to draw a more spontaneous representation of what he or she understood bullying to be.


Following the protocols used by Guerin and Hennessy (2002) and Bosacki et al. (2006), the authors developed a semi-structured interview designed to capture the children’s definition of bullying. During the interview, the authors asked the children what bullying means to them, what actions define bullying (i.e., types of bullying), what behaviors the victims engage in and what they feel emotionally when they are attacked, and whether other people (including children and teachers) are usually present when bullying takes place. Interviews were manually transcribed for later analyses.

Adolescent peer relations instrument

Children’s involvement in bullying and victimization was measured by administering an Italian adaptation of the adolescent peer relations instrument (APRI; Marsh et al. 2011) for the Italian context. The APRI is a psychometrically validated instrument that can be used to assess involvement in bullying behaviors as bullies and victims among school-aged children; although initially designed for use in adolescent samples, the APRI has also shown adequate functioning among children attending primary school (Finger et al., 2008). The Italian version of the APRI has shown good psychometric properties, as well as theoretically coherent associations with external criteria, including age, gender, internalizing and externalizing symptoms, and student-relationship quality, and students’ social status in the classroom (Marengo et al., 20192021). The APRI consists of two sections, namely the bully and victimization sections, that can be combined to assess students’ involvement in bullying behaviors as a bully, victim, or bully-victim. The bully section consists of 18 items allowing for the scoring of three subscales representing three types of bullying, namely verbal (example item: “I made rude comments to a student”), social (example item: “I got my friends to turn against a student”), and physical (example item: “I hit or kicked a student hard”) bullying. Similarly, the APRI victim section consists of 18 items allowing for the scoring of three subscales respectively representing verbal (example item; “I was called names I didn’t like”), social (example item: “A student ignored me when they were with their friends”), and physical (example item: “I was hit or kicked hard”) victimization. Items are rated using a six-point Likert scale (1 = Never, 2 = Sometimes, 3 = Once or twice a month, 4 = Once a week, 5 = Several times a week, 6 = Daily). Based on responses to each subscale we generated three categorical variables grouping participants based on their involvement in each form of bulling/victimization, that is a distinction was made between uninvolved students and those involved as bully, victim, or bully/victims in each form of bullying (i.e., verbal, physical, and social bullying). For each type of bullying, in order to be identified as either bullies or victims, students needed to have indicated “Sometimes” or a higher frequency of involvement to at least one of the bullying or victimization items. Students were categorized as bully/victims if they responded “Sometimes” or a higher frequency of involvement to at least one item assessing bullying behaviors, and one item assessing victimization. Finally, uninvolved students were identified among those responding “Never” to all administered items.

Data analysis

Content analysis of interview and drawing data

In order to detect relevant characteristics of bullying in collected data, a content analysis was conducted to develop a coding framework for subsequent analysis of the interview and drawing data. Please note that in looking into the data for such characteristics we followed a hybrid approach: first, based on a review of the literature, we determined a set of areas of interest, which we identified as being the following: (1) the specific type of bullying represented in the data; (2) the depiction/mentioning of specific criteria for bullying (i.e., repetition, power imbalance, and intentionality); (3) the behaviors and emotions shown by the victims; and finally, (4) the presence and behavior of other individuals on the scene. As a second step, for each of these areas of interest, we followed an inductive approach to let the characteristics emerge from the data. More specifically, the interview and drawing data were reviewed by one researcher in order to identify those characteristics reflecting differences in the type of bullying event described, the reference to theoretical criteria for bullying, the reactions and emotions of the victims involved, and the presence of other individuals in the scene.

In line with past qualitative studies examining children’s drawings of bullying (e.g., Bosacki et al., 2006), the bullying drawings were inspected for evidence of the presence of characters (graphical representations of one or more bullies or victims, as well as other people, including other children, and the teacher); size differences between the bully and the victim, single vs. multiple bullying scenes, the graphical depiction of verbal aggression (e.g., voice or speech bubbles and thought bubbles), physical bullying (e.g., kicking and punching); and of social/relational bullying (e.g., the exclusion/isolation of the victim, and the spreading of rumors against the victim, as depicted through voice or speech bubbles). Compliance with theoretical criteria for bullying (i.e., repetition, power imbalance, and intentionality) was determined based on combinations of the aforementioned characteristics (for example, a difference in size between the bully and the victim was considered indicative of a power imbalance; the presence of multiple bullying scenes representing the same characters was considered indicative of repetition over time; word bubbles).

A similar approach was employed in examining interview data. However, instead of looking for graphical representations of the aforementioned characteristics, we searched the interview transcripts for verbal references indicating the presence of characters, theoretical bullying criteria (i.e., repetition, power imbalance, and intentionality), the victim’s emotions and behavior in responding to the aggression by the bully, the type of aggression, and the involvement of other people beyond the bully and the victim during the bullying event.

The detected characteristics were then adapted for use as a coding framework. The reliability of the classifications was tested by checking the correspondence between the coded characteristics and an additional independent coding performed by a second researcher using the same coding framework. Independent coding was performed on a random sample representing 10% of the original interview and drawing data set. The percent agreement between two independent coders was calculated, with 70% agreement considered the minimum acceptable level of agreement. Results showed moderate-to-high agreement between coders, with the coding of sadness in the drawings showing the lowest agreement (78.6% agreement, Cohen’s K = 0.45) and coding of physical bullying in the interview data showing the strongest agreement (82.5% agreement, Cohen’s K = 0.71). Of the identified characteristics, only those that occurred in at least 5% of the sample were selected for further analysis (see Table 1). Figure 1 provides a diagram including example interview and drawing data, as well as a schematization of the procedure used to code the bullying characteristics.

table 1
error: Content is protected!!