Bullying and Cyberbullying in Primary School: The Impact of Gender and Student Academic Performance

(1) Background: This study examines school bullying and cyberbullying, which are unfortunate realities in our schools that require interdisciplinary, multidimensional educational interventions in order to enhance the quality of the educational process. (2) Methods: The study analyzes the results of the application of the School Harassment Questionnaire (CAES) to 494 students in the sixth year of primary school, aged between 11 and 13, in the Spanish region of Castilla-La Mancha. The results are presented for each dimension and item in the instrument, then analyzed by gender and student academic performance. (3) Results: The results indicated statistically significant differences in the variables examined, particularly in the variable Skills for Conflict Resolution. (4) Conclusions: The use of this instrument and results such as those in this study could enhance teachers’ awareness of the reality of their classrooms, from the school climate to the relational conflicts that may be present. This allows teachers to look into preventive action to encourage the comprehensive development of all students, with tutorial action being essential.


Introduction
In Spain, there is no unified legal framework that establishes the procedure to be followed by schools in situations of peer abuse. However, there are guidelines for prevention and support [1,2]. In addition, recent legislation also included a new crime of harassment indicated in Organic Law 1/2015, 30 March, which modified Organic Law 10/1995, 23 November, of the Criminal Code [3]. This new regulatory framework establishes a new type of harassment within crimes against freedom which is intended to respond to behavior that, although serious, could not be classified as coercion or threats. It covers cases in which, without necessarily having the explicit announcement of the intention to cause harm (threats), or the direct use of violence to restrict the freedom of the victim (coercion), there are repeated behaviors that undermine the freedom and perceived safety of the victim, who is subjected to constant persecution or surveillance, repeated name-calling, or other continuous acts of harassment.
Prevention and intervention protocols for bullying are the preserve of each autonomous community in Spain and are regulated differently in each case. When we talk about bullying or mistreatment between peers, we must consider deliberate, repeated physical or verbal aggression, social or psychological rejection, sexual or discriminatory assault, and harassment via the internet and social networks in which there are prevalent expressions of superiority, domination, and intimidation. Olweus (1993) [4] defines this term as that intentional and repeated behavior from students that causes harm to another, with an imbalance of power between the bully and the victim.
Bullying is influenced by internal and external factors, which are elements to bear in mind [5][6][7], especially in the planning and implementation of preventive strategies. It is a bullying to look at their impact on equality in education. Gender can be an important factor in the type of aggression in schools [41,42] as well as outside them [43]. We also examine academic performance as a variable, which is a key construct in schoolchildren's adaptive processes in education [44] and, more specifically, in situations of peer harassment [45][46][47]. This study aims to analyze the extent to which these variables can be predictors of bullying.
Bullying can originate from conflict in social, family, and school settings [6], hence the need to approach these phenomena of mistreatment between equals from an ecologicalsystemic perspective [7,30]. Due to the broad nature of the phenomenon of bullying, this study considers the impact that the socialization of gender and students' academic performance may have on how bullying and cyberbullying occur. Some research has included gender [48][49][50][51] and academic performance [51][52][53] variables in the understanding of bullying and cyberbullying. The current study analyzes the possible impact of these variables during primary education, which is the main theoretical hypothesis the study examines using a questionnaire designed and constructed to identify situations of peer abuse.
The general objective of this quantitative study is to analyze what sixth-year primaryschool students in Castilla-La Mancha (Spain) know about bullying by administering a specific evaluation questionnaire for this construct. The main strength of the study is that it will provide information about bullying and cyberbullying in primary-school students, as the vast majority of studies have been performed with secondary-school students. This information can also be important in the transition between educational stages and schools. Our specific objectives are to analyze the results based on two modulating variables, the gender of the students and their average academic performance across all areas of the curriculum.

Participants
We used stratified sampling to select a probabilistic sample of 494 schoolchildren in the sixth year of primary school, aged between 11 and 13 years old, in the Autonomous Community of Castilla-La Mancha (Spain). We chose this grade/year because it is the final year before students begin secondary education and move to secondary schools. It is a time of multiple physical, psychological, and social changes that could be key in the appearance of school aggression [24] because there are situations of imbalance that could lead to problems with school harmony and situations of peer abuse [6,40].
The participating sample represented 2.27% of the school population in this region for this school year [54]. This age range is similar to that used in other studies on this topic [9,13,14]. The students in our study attended 14 primary schools throughout the region, proportionally distributed according to the number of students in the sixth grade, whether the school was in an urban (296 students, 59.9%) or rural (198 students, 40.1%) areas, and whether the school was state-funded or not, public (409 students, 82.8%) and private or semi-private (85 students, 17.2%). The distribution of the sample by gender was 240 boys (48.6%) and 254 girls (51.4%). Finally, it should be noted that this sample was the same as the one used for the validation of the applied questionnaire [55].

Variables
The terminological basis for this study comes from the Ministry of Education, Culture, and Sport Resolution from 18 January 2017, which published an action protocol for situations of bullying in schools in Castilla-La Mancha [31]. The evaluation of student knowledge about the concepts of bullying and cyberbullying, the main study variable, started from three dimensions with their respective sub-dimensions: School setting: Urban (U) or Rural (R), schools in towns with more than 10,000 inhabitants were considered to be in an urban setting.

Instrument
We used the School Bullying Questionnaire (CAES) to collect the data for the study. The tool contains 40 items in order to learn what the students know about the terms bullying and cyberbullying, what they think about these concepts from the perspectives of victims, aggressors, and witnesses, and what personal and social skills they have for conflict resolution.
Content validity was ensured via expert judgment through a panel of 11 educational guidance specialist teachers with at least 10 years of experience and who had specific training in bullying, were able to participate, and worked in the same region in which the study took place. Three-quarters (77.5%) of the items had content validity indices above 0.80, none were below 0.25, and the Kappa index was 0.80. An exploratory factor analysis gave a percentage of total explained variance of 63.83% and a factorial structure similar to that designed initially, using the orthogonal varimax rotation method according to Kaiser and no items appearing correlating above 0.40 in two or more factors. According to the confirmatory factor analysis, the correlations of the items with their respective latent variables or dimensions were adequate, and there was a degree of fit of the plausible model, with CMIN values of 37.23, p = 0.067, RMSA 0.04, CFI 0.92, and TLI 0.91, confirming the fit of the established structure to the data. In terms of reliability analysis, Cronbach's Alpha for the items was 0.76 and for the dimensions 0.84 (Conceptualization 0.94, Perceptions 0.79, and Conflict resolution skills 0.78), both of which were high and significant [55].
The structure of the instrument is displayed in Table 1. Each item is evaluated using a five-point Likert-type scale between 1 and 5: Completely disagree (1), Disagree (2), Neither agree nor disagree (3), Agree (4), and Completely agree (5). The instrument takes about 45 min to complete; we also used an additional data sheet to collect information related to the variables gender and academic performance.

Procedure
The study took place during 2019 and 2020, following the questionnaire being administered in person. Teachers from each participating school provided information about student gender and performance. Before the study began, we sought approval and informed consent from parents or guardians, the school authorities, and the children. We ensured the anonymity of the responses and the confidentiality of all data collected, with published results not containing any school identifying information.

Data Analysis
First, we calculated the descriptive statistics for each item and dimension. Subsequently, we performed comparisons of means to determine whether there were statistically significant differences according to gender and academic performance. Lastly, we compared these two variables to each other via multivariate analysis and then compared them in terms of school type and setting. All statistical analysis was done using SPSS version 24.0, software license for teaching provided by the University of Castilla-La Mancha (UCLM).

Results
The results are presented according to the general study objective, examining what sixth year primary-school students in Castilla-La Mancha know about the concept of bullying and cyberbullying, and the two specific objectives, examining the results according to the gender of the students and their average academic performance in all areas of the curriculum. Table 2 shows the results of the student responses to the questionnaire.   Table 2 shows that there were high scores for the Conceptualization dimension, particularly for Item 1, the concept of bullying in the school bullying subdimension, which had the highest mean score (4.30; SD = 0.70) in the instrument. The Conceptualization dimension also includes the Cyberbullying subdimension, which had lower scores than the Bullying subdimension, the lowest being Item 6, knowledge of the dangers of social networks (2.66; SD = 0.95).
The highest scoring item in the Perceptions dimension was Item 34, about witnessing rejection or isolation on social networks or the internet, with a mean value of 2.37 (SD = 1.28). It is worth noting that Items 14 ("I have been a victim of touching"), 23 ("I have touched colleagues"), and 33 ("Have you ever seen a colleague being sexually harassed") had scores of 1 (SD = 0.00) as all of the participants gave those items the lowest score possible. Due to the lack of variation in the results for these three items, they were eliminated from the subsequent analyses.
The highest score in the Skills dimension was in Item 40, related to having friends in and out of school, with a mean of 4.00 (SD = 0.87). The lowest scoring items in the skills dimension were Items 36 ("I like to make decisions") and 38 ("I like to be the leader in my group of friends") with means of 3.01 (SD = 1.28) and 3.00 (SD = 1.23), respectively.
Once the results from the questionnaire were calculated, we analyzed them in terms of gender and academic performance. The distribution of the sample was 240 (48.6%) boys and 254 (51.4%) girls. We used a t-test for independent samples to determine differences in the results with respect to this variable. The results are shown in Table 3.
We found statistically significant differences in the Conceptualization dimension in Items 1 (p = 0.002, d = 0.  Table 3, with statistical significance, p < 0.05, in items 9,13,19,20,21,22,24,25,26,27,28,30, and Total. In this regard, it is worth noting item 11, which had a p value = 0.029 following the chi-square test, whereas in Table 3 the t-test produced a value of 0.072. There were no statistically significant values for the remaining items in this dimension.  41), and in the total dimension (p = 0.000, d = 2.02). All of these differences indicated girls scoring higher than boys, suggesting that girls have better personal and social skills for managing conflict resolution than their male counterparts. Finally, we also found a statistically significant difference in the overall score of the questionnaire (p < 0.001, d = 1.08), with the girls scoring higher than the boys. It is important to highlight the large effect size. Finally, a multivariate analysis between the sex variable and each dimension and the total score in the questionnaire confirmed these statistically significant differences, with values of p < 0.001 (Concept dimension F = 876.79, p = 0.000; Perception dimension F = 1662.43, p = 0.000; Skills dimension F = 1165.13, p = 0.000; Total F = 2842.42, p = 0.000). Student gender is a variable to bear in mind both at the preventive level and in specific interventions in situations of bullying and cyberbullying [48][49][50][51]56].
The final variable used to examine the results was student academic performance, which is shown in Table 4. To determine whether there were statistically significant differences between the schoolchildren in the instrument as a whole, we transformed the variable academic performance into a categorical variable divided into three levels: Low (Fail and Satisfactory, n = 156, 31.6%), Medium (Good, n = 142, 28.7%), and High (Notable and Outstanding, n = 196, 39.7%). The average score in each of the categories was 2.60 (SD = 0.33) for the Low level, 2.76 (SD = 0.24) for the Medium level, and 2.92 (SD = 0.34) for the High level. The mean value for the three categories was 2.77 (SD = 0.34). Table 5 shows the results of the ANOVA for this variable. After determining the statistical significance of the differences between means, we compared between groups using the Bonferroni test (Direction column).
The ANOVA (Table 5) indicated multiple statistically significant differences. For the Conceptualization dimension, they were in Items 1, 2, 3, 5, 6, 7, and 8 and for the total dimension. In general, in this dimension, students with medium and high levels of performance exhibited greater knowledge of the concepts of bullying and cyberbullying. In the Perceptions dimension, there were significant differences in Items 9 and 10 (Victim subdimension) for high-achieving students compared to medium-performing students. There were also differences in Items 20, 24, and 26 (Aggressor subdimension) with variable differences between the three types of performance. Finally, there were also differences in Items 29, 30, and 34 (Witness subdimension), although no clear pattern, with differences in favor of the three levels of performance. We found no statistically significant differences for the overall Perceptions dimension. Due to the non-normal distribution of the items in this Perceptions dimension in relation to the academic performance variable, we performed a non-parametric chi-square test. The results gave values similar to those in Table 5, with statistical significance, p < 0.05, for items 9, 10, 24, 26, 29, 30, and 34. It is worth highlighting items 13 and 20, with p values of 0.003 and 0.000, respectively, following the chi-square test, compared to the values following the t-test shown in Table 5 of 0.170 and 0.054, respectively. The results for the remaining items indicated non-statistically significant results. Academic performance has an influence on situations of bullying and cyberbullying [51][52][53]. We did find statistically significant differences in the Skills dimension in Items 35, 36, 37, 38, 39, and 40 and in the overall dimension, with high-achieving children scoring higher. From a general standpoint, these schoolchildren would have better personal and social skills for conflict resolution than their medium-and low-performing peers. After performing a multivariate analysis between the academic performance variable and each dimension and the total score in the questionnaire, these statistically significant differences were confirmed, with p < 0.001 (Concept dimension F = 1194.46, p = 0.000; Dimension of perception F = 1809.08, p = 0.000; Skills dimension F = 1750.25, p = 0.000; Total F = 4245.04, p = 0.000).
Finally, in the Table 6, we found the following results when relating each dimension to gender and academic performance after comparing averages.  To conclude this section, a multilevel analysis was carried out considering school type and setting with the scores in each dimension and the total score for the questionnaire, gender, and academic performance. The results indicated that all the factors of the model had a significant influence on the dependent variables, the scores in each dimension, and the overall instrument score. In other words, both the level 1 independent variablesgender and academic performance-and the level 2 variables-school type and settingcontributed to the variance of the students' questionnaire results (p < 0.05), although the level 1 variables explained more of the variance of the results.

Discussion
The processes of bullying and cyberbullying can cause serious problems in the schoolchildren who are the victims of them [57,58]. Many studies have reported the various psychological and emotional problems that could arise [34,37,38,[59][60][61]. This means it is essential to have action protocols in schools [62]. In this study, we highlighted the existing protocol in the Spanish Autonomous Community of Castilla-La Mancha and its embodiment in an evaluation instrument. This may be interesting for the generalization of a national protocol on school bullying that seeks to unify action criteria, including evaluation, for all of the autonomous communities in Spain.
In addition, strategies need to be developed along with dynamic content for schoolchildren to help them develop conflict resolution skills and emotional intelligence [27]. Working with the peer group is essential [63] in order to ensure that they can get along with each other [64] and avoid self-concealment, which sometimes occurs in cases of harassment [56]. Teaching work is of outstanding importance, and the initial and continuous training of teachers in specific strategies and activities is essential to prevent and intervene in relationship problems between peers [65] from the perspective of victims, aggressors, or witnesses. Empowering students with strategies and knowledge that facilitate detection and intervention before these situations become serious is essential in order to prevent them in the future as well as to properly deal with them in the present. Giving students knowledge and assessing them from an early age will promote prevention of these situations as well as intervention, bearing in mind the importance of risk and protection factors [66]. This has been the fundamental purpose of this study, work in schools promotes well-being in childhood and adolescence [67].
In terms of the results and our general objective with this study, we found the highest scores in the Skills dimension, which emphasizes the need for tutorial action work with students. The results were also high for the Conceptualization dimension, although it should be noted that the Cyberbullying subdimension had lower mean scores than the Bullying subdimension. Within the Perceptions dimension, the highest-scoring item was Item 34, related to witnessing rejection or isolation on social networks or the internet. It is also worth noting that all of the participants stated never having suffered, perpetrated, or witnessed conduct related to sexual abuse. In the Skills dimension, having friends inside and outside the school environment (Item 40) had the highest mean score.
In relation to our specific objectives, despite a balance between the genders in the sample, we did find statistically significant differences by gender with moderate to large effect sizes. In Conceptualization and Skills, girls demonstrated greater knowledge of the concepts of bullying and cyberbullying and had better personal and social skills in terms of conflict resolution than boys. In the Perceptions dimension, boys scored higher in the three subdimensions, Victim, Aggressor, and Witness. Other studies have reported similar findings of differences between the genders [27,[68][69][70][71][72][73][74]. The gender of students has been found to be a fundamental variable to consider when studying bullying [56], including cyberbullying [75,76], particularly with regard to internalization and externalization of violent behaviors and attitudes [77], and it is important to include the analysis of the greater vulnerability to bullying in LGTB students [78].
In terms of academic performance, we grouped the results into three categories, high, medium, and low. According to the ANOVA, there were generally statistically significant differences in the Conceptualization dimension in the medium and high levels, compared to low-level students. High-and medium-performing students demonstrated greater knowledge of the concepts of bullying and cyberbullying. In the Victim subdimension of the Perceptions dimension, there were some statistically significant differences between highachieving children and medium performers. In the other two subdimensions, Aggressor and Witness, there were notable differences between the three performance levels. In the Skills dimension, high-level students had higher scores, demonstrating more personal and social skills for conflict resolution than students who performed at medium and low levels. The academic performance of those involved is a variable to consider in bullying and cyberbullying behaviors [15,72,[79][80][81][82][83], whether one is considering victims or bullies [84]. Bullying hinders the teaching and learning processes in schools since the students involved could have very different levels of school performance, especially where there is physical violence [85], and therefore it should not be considered as something external to school dynamics [75].
On comparing the results of the questionnaire with the gender of the students and their performance, higher-performing boys and girls demonstrated greater knowledge of the conceptual terms and greater personal and social skills. There was a disparity in the results for the perceptions dimension, with girls demonstrating lower perceptions as victims, aggressors or witnesses, and greater personal and social skills. In addition, the multilevel analysis indicated that the school type and setting also had an impact on the results, although to a lesser degree than gender and academic performance. In short, our study confirms the impact of these variables and the importance of considering them when looking at the phenomena of bullying and cyberbullying, in line with conclusions from other studies [66].
One of the limitations of our study is its focus on the region of Castilla-La Mancha, and future work could broaden the scope to other areas, although the conceptual basis of the regulatory framework is similar in other regions such as Andalucía. Expanding this research to other regions will also allow the study of how Items 14, 23, and 33 operate, which are related to sexual abuse from the point of view of the victim, aggressor, and witness. Furthermore, a new broader application of the instrument will allow us to carry out a new confirmatory factor analysis and see whether the non-normal distribution is maintained for the items in dimension 2, Perceptions.
Finally, we would like to emphasize the importance of tutorial action in schools [86], including anti-bullying programs [87]. Topics such as social skills, group dynamics, decisionmaking, self-esteem and self-concept, and emotional intelligence are essential in the teaching and learning processes [88]. The creation of positive links between students promotes peaceful coexistence based on mutual respect and equal opportunities.  Data Availability Statement: Due to the anonymity and confidentiality of the data obtained, the authors have not reported any of the data obtained, the purpose of which is exclusively the development of this research.