Bullying in Primary School Children: The Relationship between Victimization and Perception of Being a Victim

This study aims to investigate victimization of bullying in primary school children, as well as its relationship with children’s perception of being a victim. In a sample of 4646 students from 3rd to 6th grade, we evaluated children’s victimization and cybervictimization behaviors, and children were also asked whether they had been victims of bullying or cyberbullying. From the participants, 36.7% were victims, and 4.4% cybervictims. In addition, 24.2% had a perception of being a victim, and 4.9% a perception of being a cybervictim. On the other hand, 56.9% of victims of traditional bullying had no perception of being a victim. The victimization behaviors of traditional bullying associated with a higher perception of being a victim were threats, while physical and direct verbal aggression implied a lower perception of being a victim. The results suggest the most frequent victimization behaviors may be normalized.


Introduction
Bullying is a type of aggressive behavior that implies intentionality, repetition and an imbalance of power between the aggressor and the victim, in a way that the victim is incapable of defending against the aggressor [1]. Cyberbullying occurs when bullying behaviors are carried out using electronic devices [2], situation in which the repetition, the imbalance of power and the roles of the participants are not as obvious as in traditional bullying [3]. For example, one aggression with electronic devices may be multiplied with only one intervention of the perpetrator [4]. Bullying may be divided into different types (physical, verbal and relational) each including direct and indirect forms of aggression, depending on whether the aggressor is concealed (indirect) or not (direct) [5,6]. In the present article we focus on the role of victims of bullying and cyberbullying, which implies notable negative consequences in their psychosocial adjustment [7]. To analyze this role we considered, on the one side, children's perception of being a victim (the child admits to having been a victim or not), and on the other side, the victimization behaviors the child informs to have received. In this way, we intend to analyze victimization behaviors in primary school students as a function of sex and grade, as well as the relationship between victimization behaviors and the perception of having been a victim.

Experience of Bullying and Cyberbullying
An international report of UNESCO [8] informs of a global percentage of bullying of 32% at the age of 11 years. However, bullying prevalence varies as a function of the instruments used to measure

Perception of Being a Victim
Some students may receive different victimization behaviors with a certain frequency and still not consider themselves as being victims. Juvonen et al. [24] explain that those students who had a victim perception were evaluated worse in psychological well-being than those who did not consider themselves victimized. In another study [25], they found that victims were not conscious of the rejection they suffer from other children. In this sense, it is important that studies consider not only the victimization behaviors that children receive, but also whether children have the perception of being victimized, as it is useful for professionals and families to raise awareness and to alert about the danger of bullying behaviors [26]. Even so, only a few studies on harassment have taken into account the perception of the victims when identifying them. One of those is the study by Schuster [27], who found that, among 22 students (11-13 years old) who according to their classmates and teachers suffered bullying only 7 identified themselves as victims (32%). Another study, by Hwang et al. [28], found that in students aged 12 to 15 the percentage of victims (according to their classmates) who had a perception of being a victim was 16.2% (42 out of 259). Therefore, some studies have examined the perception of being a victim from the perspective of the victim compared to the view of the classmates, but we are unaware of studies exploring the perception of being a victim in relation to the victimization behaviors reported by the same child. This analysis could allow us to understand better how the different types of victimization behaviors are related to the perception of being a victim. On the other hand, Solberg and Olweus [29] argue that young children could have a more extensive conception of bullying than older children (including actions that are not intentional or not repeated), while other authors defend that younger students exclude indirect types of bullying [5,6]. In any case, the results of Solberg and Olweus suggest that boys and girls with low frequencies of bullying are less likely to consider that they are victims of bullying.

The Present Study
Although being bullied in primary school might have consequences in secondary school, and even adulthood [30], only a few studies on bullying and cyberbullying are focused on primary education, and they usually cover just the last cycle of this educational stage (5th and 6th grade). The present research also intends to provide data about bullying and cyberbullying in the middle cycle (3rd and 4th grade). Moreover, few studies have focused on the relationship between victimization behaviors and the perception of being a victim, which is important because the perception of being a victim plays an important role in the recognition of the bullying experience as well as in its consequences [28].
The present study has two main objectives: First, to study bullying and cyberbullying victimization behaviors in primary school students, and to analyze possible grade and sex differences, including the less studied 3rd and 4th grades; second, to examine the relationship between victimization behaviors and the students' perception of being a victim.
In relation to the first objective, and according to the scientific literature reviewed above, we can formulate the following hypotheses for the sample of children from 3rd to 6th grade of primary school: (a) Boys will have higher victimization rates in traditional bullying compared to girls [14,16]; (b) Victimization rates in traditional bullying will decrease according to age [13,14]; (c) Girls will have higher cybervictimization rates compared to boys [18,19]; Finally, it was not possible to establish a clear hypothesis relating cyberbullying and age.
Regarding the second objective, we could not make a hypothesis, as we are not aware of prior research studying the relationship between self-reported victimization behaviors and having a perception of being a victim. Therefore, this part will be exploratory.
With regard to the socioeconomic status, as we collected data on a representative sample of the students from 3rd to 6th grade of primary school in Catalonia, that took into account the type of school of the children as well as the school area (see Procedure section), we believe the sample is representative of the different socioeconomic statuses that exist in this area of Spain.
In order to ensure that our sample size was sufficient to detect an adequate effect, we computed a post hoc power analyses using G * Power 3.1 Software [31]. Results revealed that our total sample size of 4646 participants was sufficient, in the case of the ANOVA test, to detect with a 0.99 power a medium effect size (f = 0.25).

Instruments
It is worth noting that right before the administration of the questionnaires, a definition of bullying and cyberbullying was provided to the children by the researchers of the project. Children also had the opportunity to ask questions. Moreover, a definition was also written at the beginning of the questionnaires. The following instruments were administered to the participants, by order of presentation.
(a) Bullying Questionnaire The European Bullying Intervention Project Questionnaire (EBIPQ) [32] was used. It is an instrument of 14 items (7 for victimization and 7 for aggression) in which children are asked what situations they have lived in the last two months, evaluated on a Likert scale with 5 response options: 0 = No; 1 = Yes, once or twice; 2 = Yes, once or twice a month; 3 = Yes, about once a week; 4 = Yes, more than once a week. In the present study, only the victimization scale has been taken into account. Reliability, assessed through Chronbach's Alpha was 0.81 on the victimization scale.

(b) Cyberbullying Questionnaire
A reduced version of the European Cyberbullying Intervention Project Questionnaire (ECIPQ) [32] was used. The original instrument consists of 22 items (11 for cybervictimization and 11 for cyberaggression) evaluated on the same Likert scale mentioned for the bullying questionnaire. In our version for primary education (see Appendix A), some items were eliminated or merged, while others were maintained, finally using 12 items (6 cybervictimization and 6 cyberaggression). In this study, only cybervictimization items were taken into account. Reliability, assessed through Chronbach's Alpha on the cybervictimization scale, was 0.75.

(c) Perception of being a victim and cybervictim
We asked children whether they have been a victim, aggressor or observer of bullying or cyberbullying. In relation to the perception of being a victim, we asked "Have you been a victim of bullying?" and "Have you been a victim of cyberbullying?" Children had to mark yes or no in each question.

Procedure
First, the Department of Education of the Catalan Government was contacted, who gave permission to conduct the study. To select the educational centers, a representative random sample of students from 3rd to 6th grade of primary school in Catalonia was carried out, stratified according to the type of school (public vs. private or concerted), and size (fewer than one class per grade, one class per grade, or more than one class per grade). Furthermore, in the case of public schools, the territorial area of the Department of Education was also considered as a criterion of representativeness. The school management teams were contacted to explain the project and request participation based on: (a) an explanatory letter of the study; (b) a letter of support from the Department of Education; and (c) consent from the school to confirm participation. A total of 41 schools participated in the study (data were collected between December 2018 and April 2019).
The questionnaires were administered mostly in paper, and only 10 group classes responded to it online. The families were informed of the objectives of the study and their informed consent was obtained. The researchers from the project administered the questionnaires in most cases, but five schools preferred to administer them on their own. After being provided with an explanation of bullying and cyberbullying, children chose to respond the questionnaires in Catalan or Spanish. Researchers from the project monitored that children answered correctly the questionnaires, and gave support to those who asked for help. At no time were the children asked for their names. After completing the questionnaires, children put them in a sealed envelope to preserve anonymity.
This study was approved by the Ethical and Biosecurity Research Committee of the University of Girona (code: CEBRU0016-2018).

Data Analysis
Data were analyzed using the SPSS 25 statistical program. To classify the participants into victims or cyber victims, the criteria by Romera et al. [19] was used, who considered as victims those participants with a minimum frequency of "Once or twice a month" in any of the EBIPQ or ECIPQ victimization items. On the other hand, in the line of García-Fernández et al. [33], participants with a minimum frequency of "More than once a week" in some item have also been considered as frequent victims or frequent cybervictims. Finally, the items of the EBIPQ questionnaire were classified into direct aggressive behaviors (items 1, 2 and 4) or indirect/relational behaviors (items 3, 5-7).
In order to compare victimization behaviors as a function of sex and grade, the mean scores of each type of behavior were used and an ANOVA of grade (4) × sex (2) was performed, also analyzing the simple effects of sex and grade. The partial eta squared (η p 2 ) was taken as a measure of the effect.
To analyze the effects of grade and sex on the perception of the victim, the chi-square test was used, using Cramer's V as a measure of the effect. The same test was used to analyze the relationship between victimization behaviors and perception of being a victim.

Victims of Bullying and Cyberbullying
From the participants, 36.7% met the criteria established for victims of traditional bullying and 4.4% the criteria for cybervictims. On the other side, 17.4% of the participants were frequent victims of traditional bullying, while 1.7% of them were frequent cybervictims. Table 1 shows the victims of bullying and cyberbullying according to sex and grade. Table 2 shows the mean scores for victimization and cybervictimization items, so they represent for each sex and grade the mean frequency of the victimization and cybervictimization behaviors.   The results showed that 10.3% of victims of traditional bullying were also victims of cyberbullying. The prevalence of cyberbullying was much lower (of only 1%) among those participants who were not victims of traditional bullying. On the other hand, among the victims of cyberbullying, 85.8% were also victims of traditional bullying, while the prevalence of traditional bullying among participants who were not cybervictims was 34.1%.
The results in traditional bullying (means of victimization scores) showed a significant effect of sex (see Table 2 The results of the 2-way ANOVA for cyberbullying behaviors (mean scores) showed a significant effect of sex with higher scores for boys (see Table 2), a significant effect of grade (F(3, 4196) = 5.96; p < 0.001; η p 2 = 0.004), and also an interaction between both variables (F(3, 4196) = 4.90, p = 0.002; η p 2 = 0.003). When analyzing the simple effects of sex in each of the grades, differences were observed between boys and girls at 3rd and 4th, but not at 5th or 6th grade (see Table 2). The analysis of the simple effects of grade for each sex indicated a grade effect for girls (F(3, 4196) = 8.38; p < 0.001, η p 2 = 0.006) but not for boys (p = 0.063). Therefore, cybervictimization remained stable in boys throughout primary school, while increased in girls in the last years of this stage. Table 3 shows the percentage of victims as a function of the number of victimization behaviors. From the total amount of victims of traditional bullying, more than half were victims in more than one item (60.5%). Similarly, 44.3% were frequent victims in more than one item. Most of the cybervictims and frequent cybervictims reported victimization behaviors just in one item (56.8% and 58.1% respectively).  Table 4 shows direct and indirect victimization behaviors. A two-way ANOVA of grade (4) × sex (2) was performed to analyze the effects of those variables. For direct victimization, there was an effect of sex and grade (F(3, 4309) = 34.55; p < 0. 001, η p 2 = 0.023). The analysis of simple effects showed sex differences in each grade (see Table 2 In summary, boys showed more direct victimization behaviors than girls, and in both sexes these behaviors tended to decrease from 3rd to 6th grade. For indirect victimization (see Table 4), a sex effect was also found with higher scores for girls (see Table 4). Similarly, a grade effect was found (F(3, 4186) = 29.97, p < 0.001, η p 2 = 0.021), but there was no interaction between sex and grade (p = 0.607). When analyzing the simple effects of grade for each sex, differences were found both in girls (F(1, 4186) = 13.38, p < 0.001, η p 2 = 0.009) and boys (F(1, 4186) = 17.27, p < 0.001, η p 2 = 0.012), with a decrease in both. Finally, the analysis of simple sex effects in each grade showed that sex differences in indirect victimization occurred only in 6th grade, with higher scores for girls. In Appendix B (Table A1)

Perception of Being a Victim
From all participants, 24.2% had the perception of having been a victim of traditional bullying, and 4.9% reported having been victims of cyberbullying. Table 5 shows the participants with a perception of being a victim as a function of grade and sex. We did not find a relationship between sex and the perception of being a victim of traditional bullying (using the Chi-square test) in the total sample (p = 0.355), and only at 5th grade when dividing it by grades (3rd: p = 0.903; 4th: p = 0.536; 5th: χ 2 (1, n = 1102) = 7.71, p = 0.006; Cramer's V = 0.084; 6th: p = 0.818). However, there was a relationship between perception of being a victim and grade (χ 2 (3, n = 4574) = 31.18, p < 0.001; Cramer's V = 0.083). This relationship was significant both for boys (χ 2 (3, n = 2209) = 19.75, p < 0.001; Cramer's V = 0.095) and girls (χ 2 (3, n = 2144) = 16.66, p = 0.001; Cramer's V = 0.088), with a reduction in the perception of being a victims as the grade increased.

Relation between Victimization and Perception of Being a Victim
When analyzing the relationship between the victimization behaviors reported in the EBIPQ questionnaire (on traditional bullying) and the perception of being a victim, a significant relationship was found between both variables (See Table 6), although the majority of victims (56.9%) according to the EBIPQ do not have the perception of having been a victim of bullying. This relationship remained stable throughout the different grades; the percentages of victims who did not have a perception of being a victim were 55.6% at 3rd, 61.5% at 4th, 54.5% at 5th, and 56.12% at 6th grade. Regarding the relationship between frequent victimization in traditional bullying and the perception of the victim, it was also statistically significant (See Table 6). In this case, frequent victims who did not have a perception of being a cybervictim were 44.6%, being quite stable throughout the grades (46.6% at 3rd, 51.7% at 4th, 43.3% at 5th, and 40.1% at 6th). When analyzing the relationship between victimization and the perception of the victim as a function of the type of victimization behavior (see Table 7), we observed that the type of victimization behavior where the subjects had a greater perception of being a victim was being threatened, followed by exclusion/being ignored. On the other side, the behavior with the lowest perception of the victim was direct verbal aggression, and the second with the lowest perception was physical aggression, which are among the behaviors with a highest frequency (See Appendix Table A1).
The analysis of the relationship between the perception of being a victim of traditional bullying and the number of items in which participants had a suffered victimization (see Table 8) showed that the more items in which children reported victimization, the more likely they had a perception of being a victim.  Table 9 shows the relationship between cybervictims (according to the ECIPQ) and the perception of being a cybervictim. The Chi-square test shows a relationship between both variables. Even so, 67% of the cybervictims did not have the perception of being a cybervictim. Besides, a significant relationship was also found between being a frequent cybervictim and having the perception of being a cybervictim. The percentage of frequent cybervictims who did not have a perception of being a cybervictim was of 58.9%.

Discussion
This study aimed to analyze children's victimization of bullying and cyberbullying in primary school (grades 3rd to 6th) as a function of grade and sex. Also, it intends to study the relationship between children's victimization behaviors and their perception of being a victim.

Victimization in Traditional Bullying
We found 36.7% of victims of traditional bullying in children from 3rd to 6th grade (and 17.4% of frequent victims). These percentages are slightly higher than those indicated by UNESCO [8] globally at the age of 11 (32%), and also slightly higher than the report by Romera et al. [19] on 5th and 6th grade in Spain (around 30%). The traditional bullying victimization instruments and criteria in our study were equivalent to those of Romera et al., so the differences can possibly be explained by a greater victimization in our study in 3rd and 4th grades compared to 5th and 6th grades. Hence, our data point to a decrease in traditional bullying throughout primary school in both sexes, in line with other studies [8,11,14], and higher rates than in secondary school. Thus, our hypothesis about a decrease of victimization rates according to age has been supported. It is possible that bullying-related behaviors changes along primary school, but this change could also be explained by age-related changes in the conception of bullying [29].
Regarding sex differences, we found a higher victimization in traditional bullying in boys than in girls (except in 6th grade), similarly to other studies [8,14,16,34], placing the differences already from 3rd grade. Hence, although these data support our hypothesis of greater victimization in boys than in girls, this is not the case in 6th grade. Besides, the lack of sex differences in 6th grade could indicate a change in trend between primary and secondary education. In the study by Romera et al. [19], sex differences in traditional victimization were very low for primary education (girls showed 2% more victimization) but increased in secondary education (around 6-8% more for girls).
We found a greater direct victimization in boys than in girls, a difference that was significant in all grades. In contrast, girls showed more indirect victimization behaviors than boys, although these differences were only significant in 6th grade. These data are in line with some authors who suggest that boys are more involved in physical forms of aggression and girls in indirect forms [10]. Beyond these descriptive data, several questions emerge regarding the observed differences between boys and girls in the different grades and in relation to the different forms of bullying. Our study did not used the appropriate instruments to allow a discussion about this issue, but we consider that analyzing this phenomenon from a gender perspective would permit a better understanding of the issue [35].

Cybervictimization
We found 4.4% of prevalence of cybervictims (and 1.7% of frequent cybervictims), percentages lower than those found previously in other parts of Spain [21,22], and specifically in Andalusia, where the percentages in 5th and 6th of primary education varied from 7.5% to 10% [19]. On the other hand, while in our study the vast majority of cyberbullying victims were also victims of traditional bullying (85.8%), a very low percentage of victims of traditional bullying were also victims of cyberbullying (10.3%). Other studies already pointed out that in primary school cyberbullying is related to problems with traditional bullying [36], and in the present study we highlight the high probability of suffering traditional bullying among cybervictims. This might imply a lack of convenience of interventions only aimed at cybervictimization [37], especially in primary education.
With regard to sex differences in cybervictimization, our hypothesis has not been confirmed, as we did not observe more cybervictimization in girls compared to boys. Contrarily, we observed more cybervictimization in boys than in girls from 3rd and 4th grades. If we take into account both the effects of grade and sex, we found that cybervictimization remained stable in boys throughout primary education, while it increased from 4th grade in girls. So more cybervictimization was found for boys compared to girls in grades 3rd and 4th, but no sex differences existed in grades 5th or 6th, where an increase was found for girls, especially in grade 6. These results differ somewhat from other investigations in which a higher prevalence of cybervictimization was found for girls with older samples [8,16,19]. Thus, sex differences in cybervictimization could undergo a change in trend throughout primary and secondary education, going from being more frequent in boys first and in girls later. This suggests that teachers should be especially attentive to cybervictimization for boys in the earlier years of primary school, and according to what other studies suggest [38], it is important to implement prevention and intervention programs to help children understand the risks of communication technologies and to deal with cyberaggressions. Finally, in our study most cybervictims were victims of only one type of behavior, whereas most victims of traditional bullying were victims of more than one behavior.

Relation between Victimization and Perception of Being a Victim in Traditional Bullying
In our study we found that the perception of being a victim (of traditional bullying) is reduced as the grade increases, in line with previous results [39,40]. Considering that the question about perception of being a victim refers to any time point, this reduction is counterintuitive (because older participants have had a longer period for having suffered bullying), and might indicate that younger children have a broader conception of bullying than older children [41]. Nevertheless, it could also be interpreted as a greater reluctance in older children to accept that they have been a victim. In this sense, some school bullying prevention programs include public recognition of the prevalence of bullying, so that the victim does not attribute it so much to oneself and more to contextual conditions [24]. It is also possible that in the case of the youngest children (3rd grade), a different conception or understanding of bullying may have influenced the results.
Our results show that only 43.1% of victims of traditional bullying (according to children's informed victimization behaviors) have a perception of being a victim (and 55.4% in the case of frequent victims). These percentages are higher than in the studies by Schuster [27] and Hwang et al. [28], so a greater congruence between the experience of victimization and the perception of being a victim appears when the victimization is evaluated through self-report (as in our case) than through peers (as in the other mentioned studies).
Our study highlights that the perception of being a victim is related to the type of victimization behaviors. Thus, the most frequent victimization behaviors (physical aggression and direct verbal aggression) had a lower perception of being a victim, perhaps indicating a normalization of those behaviors. In this direction, Cuadrado [42] states that many children and adolescents internalize and normalize many indirect verbal and physical abuses. Perhaps this could also explain why more than half of the victims in our study did not have a perception of being a victim, and only when the number of victimization behaviors increased, or if they suffered a more frequent victimization, this perception increased. Solberg and Olweus [29] also found that boys and girls with low frequencies of bullying were less likely to consider themselves as victims.

Relationship between Cybervictimization and Perception of Being a Cybervictim
The perception of being a cybervictim was higher in boys compared to girls at 3rd and 4th grades, but higher in girls at 6th grade, following a trend to decrease in boys throughout primary school and to increase in girls. These data are similar to our results on cybervictimization prevalence. Besides, we observed that only 33% of cybervictims (and 41.1% of frequent cybervictims) had a perception of being a cybervictim, which is lower than in perception of being a victim in traditional bullying. This is especially worrisome considering that attacks with electronic devices can multiply and lengthen over time with a single action by the aggressor [4]. Thus, it seems necessary to work in primary schools on the concept of being a victim of bullying, and especially of cyberbullying.

Limitations
Regarding the limitations of the study, when we asked about the perception of being a victim of bullying, we did not ask how long ago it happened, which could have helped to interpret our results more adequately. In addition, the instruments used to measure victimization (EBIQP) and cybervictimization (ECIPQ) do not directly address the issues of intentionality and imbalance of power inherent in the definition of bullying. Moreover, if we had used sociometric data, or measured the victimization perceived by the peers, our data would have been more comprehensive. Finally, as we commented earlier, we cannot discuss up to which point the observed sex differences are related to gender socialization-see Navarro [43] for a review of gender studies on cyberbullying.

Conclusions
In the present study we found a reduction with grade of both children's victimization behaviors (of traditional bullying) and of children's perception of (traditional) bullying. Future research should try to disentangle whether these age-related changes are an effect of behavioral or conceptual changes. It is possible that young children have a broader concept of bullying compared to older children, but also that older children are more reluctant to accept that they have been bullied. Regarding sex differences in traditional bullying, boys were more victimized than girls except in 6th grade. Also, in line with prior research, boys suffered more direct victimization (physical and verbal aggressions) than girls in all grades, while girls in 6th grade suffered more indirect victimization than boys. In any case, direct victimization was more frequent than indirect victimization in both sexes.
Cybervictimization was stable in boys but increased in girls along the primary school years. As it was more frequent in boys only at 3rd and 4th grade, and prior studies had found a higher prevalence for girls in older samples [8,19], cybervictimization could undergo a sex-related change of tendency from primary to secondary education. On the other hand, our results highlight that cybervictims in primary school have a high likelihood of suffering from traditional bullying, an aspect that should be considered in interventions designed to address cyberbullying.
We also found that the victimization behaviors of traditional bullying that were associated with a higher perception of being a victim were threats; on the other hand, physical and direct verbal aggression, which were among the behaviors with the highest frequency, were associated with a lower perception of being a victim. This might indicate a normalization of these aggressive behaviors in primary school children, which is an important finding. According to Thornberg [44], children often view bullying as a normal consequence of deviance related to the victim or fight for social status, and thus they might interpret it as ordinary and trivial and reduce their motives for intervention. Future studies could analyze further which factors contribute to the normalization of these behaviors, and also to understand its consequences on children. More effort must be made to understand how children perceive victimization so that education professionals and families can reduce the widespread tolerance and consequent normalization of bullying [26].
From the Theory of system justification [45] it is argued that it is painful to accept that one is living an injustice situation, which leads some people to interpret this situation less negatively; this might help them to feel better, but at the same time reduces the likelihood of taking action and rejecting injustice, especially if they view the situation as inevitable. This could (at least in part) explain why we found that only about half of frequent victims/cybervictims have the perception of having been a victim/cybervictim. Thornberg [44] suggests that bullying might create a victim career on the victim, which starts with a social construction of being deviant or marginalized, and which implies a social devaluation of the victim.
In this regard, our results suggest the need for psychoeducational interventions to work on children's conceptions of being a victim of bullying (e.g., to help valuing themselves) and against the acceptance and normalization of aggressive behaviors. This seems especially necessary in virtual contexts, where we found a lower perception of being a victim. School bullying represents an important health problem for the victims [14,15]. In this regard, educational centers should promote the rejection of these behaviors and offer explicit support for those who suffer victimization behaviors, as it would make it easier for them to seek help [46].
Funding: This study was carried out thanks to the funding of BARÇA FOUNDATION.