Validating the Cisneros Self-test on Bullying in Dominican Students

Bullying is a relevant concept in education. In the Dominican Republic, the high bullying rates call for research focused on this big issue. Bullying is composed of different dimensions and it is significantly associated with other factors such as prosocial behavior, emotional instability, suicidal ideation, or intimidation. Therefore, there is a need for an instrument that can analyze the level of bullying as a single dimension, such as the Cisneros Self-test on Bullying (Oñate & Piñuel, 2005). This instrument analyzes the intensity of bullying and values its relevance to the different factors mentioned above. Our main goal was to test the psychometric properties of this instrument, including internal consistency and factorial and nomological validity in 531 Dominican secondary students. To validate it, we utilized two instruments: the Emotional Instability Scale (Caprara & Pastorelli, 1993) and the Prosocial Conduct Questionnaire (Caprara & Pastorelli, 1993). Several competitive structural models were estimated, and the results supported the reliability and, factorial and nomological validity of this instrument. Therefore, we can conclude that Cisneros’s scale is a good, reliable, and valid instrument to measure global bullying in the Dominican Republic. We highlight its influence on future interventions considering the relevance that bullying has on many other factors such as aggressiveness, prosocial behavior, or suicide ideation.

t present, bullying is an issue of concern to society, both in terms of the acts it entails and its consequences.According to PISA (2018), this is a big concern in the Dominican Republic where bullying rates are especially high.More than 20% of students report having suffered frequently bullying and these rates have increased considerably between 2015 and 2018.
International studies that address it contemplate different dimensions (Baridon & Martin, 2014;Schultze-Krumbholz et al., 2015).Olweus (1980) offered an initial approach to violent behavior among classmates in Norwegian school spaces, by observing the mistreatment and abuse continuously produced between classmates (Román & Murillo, 2011;Sargin, 2017).There is great difficulty in precisely defining and specifying the concept of violence.It is a complex phenomenon (Baridon & Martin, 2014) with different possibilities to address: verbal and physical violence, and even socially isolating the victim (Marín & Reidl, 2013).
School violence includes "a wide range of actions that are intended to cause harm and that alter the institutional environment to a greater or lesser extent" (Martínez-Otero, 2005, p. 35).It refers to behaviors that violate the educational purpose of the school, including antisocial behavior, aggression, bullying, and violence itself (Sargin, 2017).Bullying manifests itself in the form of interpersonal violence caused by a perversion of relationships between equals that cease to be equal and symmetrical, becoming unbalanced and regulated by a dominationsubmission scheme (Olweus, 1980).
There are three conditions that distinguish an incident of bullying from other violent behaviors that can occur within the school context (Save the Children, 2013): (1) intentionality on the part of the bully; (2) repetition in time; and (3) power imbalance based on physical, psychological, or social inequality.Olweus (1991) adds that (4) it must appear in most cases without prior provocation by the victim.In addition, more recently, what would be considered a fifth condition has been proposed: (5) most of the time it is aimed at a single student (Garaigordobil & Oñederra, 2010).
Thus, different forms of bullying can be considered (Jara et al., 2017); we could distinguish among three categories (Le Menestrel, 2020;Rose et al., 2016).In the first place, physical harassment refers to direct aggressive actions towards the person or their possessions (Caballo et al., 2011).Second, verbal harassment, is more related to threats, insults, slander, blackmail, and/or cruel teasing related to some characteristic of the harassed individual.This is the most frequent category of aggression (Baridon & Martin, 2014;Del Barrio et al., 2007).And thirdly, social harassment is related to the isolation, exclusion, and/or social marginalization of a person, who gradually moves away from his peers, until he is left without any external support (Mateu-Martínez et al., 2017).To these three, a fourth category could be added (Garaigordobil, Martínez-Valderrey, Páez, et al., 2015;Jara et al., 2017): the psychological (Garaigordobil, Martínez-Valderrey, Páez et al., 2015;Garaigordobil & Oñederra, 2010).Psychological harassment refers to all those actions whose objective is to undermine the victim's self-esteem, increasing the feeling of insecurity, fear, and even the feeling of guilt and powerlessness in the face of a problematic situation.It is related to those attacks that have an effect on the mind of the harassed.This last category is particularly relevant since it is present in the rest of the categories of bullying (Garaigordobil, Martínez-Valderrey, Páez, et al., 2015;Garaigordobil & Oñederra, 2010).The relevance of self-esteem  Balluerka et al. (2023) show how self-esteem mediates the relationship between bullying and anxiety and depression.On the other hand, Álvarez et al. (2022) found that the self-esteem of bullying victims is lower than that of non-victims or bullies, in addition to having greater symptoms of depression and emotional and behavioural difficulties.
Given the complexity of the construct, making a good measurement is not easy.The Cisneros Self-test on Bullying (Oñate & Piñuel, 2005) is a general measure of bullying (Piñuel & Oñate, 2006).The Cisneros Report, which gives its name to the instrument, is the most extensive study carried out in Spain to measure bullying.Specifically, the Cisneros X Report (Oñate & Piñuel, 2007) included 24,990 subjects from 7 to 17 years of age from all over the country.This scale includes measures of harassment: offensive behavior towards the victim; exclusion: social isolation; intimidation: physical or psychological threat; and aggression: physical or psychological harm.It provides a final global measure of bullying that can help to know the scope of this social and educational problem.
Despite its wide use in this context, the validations of the instrument are very scarce (Cassiani-Miranda et al., 2014;Cepeda-Cuervo et al., 2008).Through this research, a validation of the instrument is carried out in a different sample from that of previous studies, to prove its psychometric properties.Therefore, the aim of this study is to test the psychometric properties of the Cisneros Self-Test on Bullying, including internal consistency and factorial and nomological validity, at the secondary level of education in the Dominican Republic. ´

Participants and Procedure
The participants of this study are students from the six mandatory years of secondary education from San Pedro de Macorís, Educational District 05-03 of La Romana (Dominican Republic).We chose five centers that fit the criteria: to have all secondary grades (from 1 st to 6 th ) in the same district (La Romana).In each educational center, the participating students were randomly selected according to their distribution.The directors from each center were contacted by sending them the informative and consent letter to get both, from the center and the parents.Taking into account the population data of the five public centers located in the area, and after a stratified probabilistic sampling, by school and educational level, a total of 531 students were obtained as the final sample (Table 1).The mean age of the participants was 15.48 years (SD = 1.74), and they were between 13 and 21 years old.Of the 531 students, 227 were male (42.7%) and 304 were female (57.3%).50.5% of the students lived with both parents, 37.5% only with one of the parents, and the remaining 12% were distributed in an extended family, reconstituted, and/or with grandparents.
The questionnaires were applied in the classrooms, or a place arranged by the school according to availability.For this, it was ensured that the conditions of the spaces were optimal, quiet, and comfortable.Moreover, participation was completely voluntary, and the answers were anonymous.Participants were duly informed about the procedure for filling out the instruments, the conditions of anonymity and confidentiality of the survey, so they could be confident and their answers as sincere and honest as possible.Given that the survey was taken in the classroom setting the amount of missing data was negligible, less than 1%.The study was conducted following the guidelines of the Declaration of Helsinki and was approved by the Ethics Committee of the University of Valencia (Spain).

Instruments
Participants completed some socio-demographic indicators previously presented in the sample description, together with several psychological scales.Subscales and total scores are calculated as the mean of the different indicators.Any reverse-coded item has been previously reversed to get the total score: 1) Cisneros Self-test on Bullying (Piñuel & Oñate, 2006).Composed of 50 items, it is a self-test integrated into the AVE School Harassment and Violence Questionnaire (Piñuel & Oñate, 2006).The AVE is an essential tool, aimed at bullied people, to prevent, identify, treat, and diagnose bullying, school abuse and the psychological damage most frequently associated with these behaviors.Using a 94-item self-report questionnaire, two global indexes are obtained: bullying index and Intensity of bullying.The Cisneros Self-test on Bullying (Oñate & Piñuel, 2005) would correspond to this first, with 50 items used in this research, composed of 4 global factors of harassment: Harassment, Intimidation, Exclusion and Aggression.The instrument is designed in a Likert-type scale format from one to three (never to many times).The original version of the Global Bullying Index obtained a Cronbach's alpha of .96(Oñate & Piñuel, 2005).This will be the questionnaire object of validation in this study.
2) Emotional Instability Scale (Caprara & Pastorelli, 1993) in the Spanish version of Del Barrio et al. (2001).It consists of 20 items that describe adolescent behavior measuring lack of control, and low capacity to contain impulsivity and emotionality.The instrument is designed in a Likert-type scale format from one to three (never to often).The Spanish version of the Emotional Instability Scale obtained a Cronbach's alpha of .74(Del Barrio et al., 2001).The internal consistency of the test in this study, for the single dimension of emotional instability, was α = .81and a one-factor CFA also fitted the data well:  2 77 = 213.93,p < .001,RMSEA = .058[.049 -.067], CFI = .943.
3) Prosocial Conduct Questionnaire (Caprara & Pastorelli, 1993).We used the Spanish version adapted by Del Barrio et al. (2001).Initially, it had 15 items, of which 10 offer a description of children's behavior about altruism, trust, and sympathy, and five are control items.The instrument is designed in a Likert-type scale format from one to three, from never to often.The Cronbach's alpha of the scale in other Spanish-speaking samples ranges between .60 and .65 (Mestre et al. 2001;Mestre et al. 2002).The scale provides a reliability estimate for prosocial behavior with an alpha of .67 and, again, a CFA with one factor fitted the data reasonably well:  2 35 = 119.18,p < .001,RMSEA = .067[.054 -.081], CFI = .900. 4) Scale of Physical and Verbal Aggressiveness (Caprara & Pastorelli, 1993), in the Spanish version by Del Barrio et al. (2001).Of the 20 items on this scale, 15 describe the behavior of children referred to hurting others either physically or verbally, and five are control items.The instrument is designed in a Likert-type scale format from one to three (never to often).The Spanish version of the Scale of Physical and Verbal Aggressiveness obtained a Cronbach's alpha of .84(Del Barrio et al., 2001).The internal consistency estimates of the test were α = .84for verbal aggressiveness, and α = .80for physical aggressiveness.A two-factor CFA had excellent fit:  2 89 = 247.65,p < .001,RMSEA = .058[.049 -.067], CFI = .968. 5) Inventory of Positive and Negative Suicidal Ideation (PANSI; Osman et al., 1998).It is a scale adapted to Spanish by Villalobos-Galvis (2010).It has 14 items: six measure positive suicidal ideation or protective factors as self-control, and eight measure negative suicide ideation or risk factors (social exclusion).The instrument is designed in a Likert-type scale format from one to four (never to always).The Spanish version of the Inventory of Positive and Negative Suicidal Ideation obtained a Cronbach's alpha over 0.8 (Villalobos-Galvis, 2010).In this study, Cronbach's alpha was .70 for positive suicidal ideation and .89for negative suicidal ideation.A two-factor solution CFA model fitted the data very well:  2 76 = 242.89,p < .001,RMSEA = .064[.055 -.073], CFI = .979.

Statistical Analyses
The analyses included internal consistency estimates for items and scale (i.e.Cronbach's alpha, items-total corrected correlations, and Composite Reliability (CRI); Raykov, 1997).Correlations were used to test for nomological validity.
The factorial validity of the Cisneros Self-test on Bullying was tested via a set of competitive structural models (Confirmatory Factor Analyses, CFA).These competitive models are alternative theoretical and or empirical structures of the scale.The models tested (Figure 1), according to the literature and the original structure of the scale, were: 1) One-factor model.One factor of bullying underlying the 50 items of the Cisneros Selftest on Bullying was specified.This model assumes a general trait construct of bullying.This is a baseline model tested for parsimony.
2) Four-factor model.Four correlated factors underlying the 50 items of the Cisneros Selftest on Bullying: harassment, intimidation, social exclusion, and aggressiveness.This is the actual first-order structure defended by Piñuel and Oñate (2006).
3) A second-order factor model, with four first-order bullying dimensions (harassment, intimidation, social exclusion, and aggressiveness) and a second-order factor of bullying.This model represents the first and second-order structure defended by the authors of the scale (Piñuel & Oñate, 2006).

Figure 1. Structural equation models specified and tested for the Cisneros Self-test on Bullying
The plausibility of these confirmatory models has been assessed using several fit criteria (Hu & Bentler, 1999;Tanaka, 1993)

Factorial Validity
As stated in the method section three competitive factor structures were tested.These models were: 1 one-factor model, 2) four-factor model, and, 3second-order factor model (Figure 1).Table 2 presents fit indexes for these three a priori models.The general fit indexes showed an adequate fit for all three models, but model 1 offers greater parsimony, and it is therefore retained.The standardized factor loadings are presented in Table 3.All factor loadings of the items in the single bullying factor were statistically significant (p < .01).In general, standardized factor loadings >.4 are considered adequate.Standardized loadings ranged from a minimum of .186(item 1) to a maximum of .836(item 47).The mean standardized loading was .672(SD = .143).

Internal Consistency
Cronbach's alpha was computed for the factor of bullying with an estimate of .931.CRI was also computed with a value of .977.As can be seen in Table 4, the means are close to the midpoint of the response scale (1.5) in most of the items.In general, the homogeneity of the items is adequate.Only two items get values less than .30(items 1 and 6).Ignoring these two, the homogeneity values range from a minimum of .30(items 4 and 5) to a maximum of .66(item 44).

Nomological Validity
The nomological validity of the scale was established by correlating it with constructs theoretically linked to it in the literature.That is, establishing reasonable relationships with the theoretical network or nomological validity.The bullying dimension obtained a statistically significant correlation with all the constructs, being positively related in the case of physical and verbal aggressiveness, emotional instability, and negative suicidal ideation; and negatively in the case of prosocial behavior and positive suicidal ideation.All correlations are in Table 5.These relationships are in line with those traditionally found in the literature.

Discussion
The importance of having quality instruments of measurement is unquestionable.If professionals do not perform good assessments, they cannot perform good interventions.If we apply this to areas as sensitive as bullying, the relevance increases.A good instrument for measuring bullying should be able to detect the level of bullying that a bullied person is suffering, considering and taking into account the complexity of the construct.In this sense, the Cisneros Self-test on Bullying (Oñate & Piñuel, 2005) has been used in what, to date, is the most extensive study carried out in Spain to measure bullying.Specifically, the Cisneros X Report (Oñate & Piñuel, 2007) included 24,990 subjects from 7 to 17 years of age from all over the country, and in which it was shown that 23.2% of the students had ever been bullied.It is a measure of bullying, within the two dimensions that make up the School Harassment and Violence AVE Questionnaire (Piñuel & Oñate, 2006): bullying and intimidation.This study aimed to test the psychometric properties of Cisneros Self-test on Bullying for a population of particular interest, due to the high rates of bullying in their classrooms: secondary students in the Dominican Republic.There are several studies that present a high rate of bullying in public (Parada et al., 2017), and recognize the Dominican Republic as the third most violent country in Latin America, only behind Argentina and Ecuador (Román & Murillo, 2011).The current validation offers results of internal consistency, factorial, and nomological validity in a sample of Dominican secondary school students.The results show a clear factorial structure, satisfactory reliability, and good validity when applied to students.The factorial validity of the scale was tested using competitive structural models (Confirmatory Factor Analysis, CFA).Among the three models tested, a one-factor model emerged as the one that best fitted and offered the most parsimony.Therefore, a structure of a bullying factor for the Cisneros Self-test on Bullying is proposed, instead of the four-factor structure proposed by the original authors, and traditionally used.Current results are important since they show that there is a general dimension of bullying that does not differentiate between types of violence.In other words, all types of bullying are most likely exerted together.The findings of this study are especially relevant because it demonstrates the quality and usefulness of the Cisneros Self-test on Bullying for measuring this construct.Therefore, it certifies that it is a precise and simple measure, which contemplates the different types of bullying, very useful for detecting and assessing any case of bullying.In addition, taking into account the widespread use that this scale has in Spanish-speaking countries, the results of this study question the traditional use that has been given to its dimensions.Taking this into account, the difficulty of differentiating between types of bullying could be raised, at least with the use of this specific scale.
The factor loadings for the bullying dimension were large, so the factorial validity of the scale with a single factor of bullying has been supported by the data.Only item 1 ("They don't speak to me") presents a low and not significant factor loading but without impact on the internal consistency of the scale.The analyses provide evidence of good levels of internal consistency for the scale, adequate Cronbach's alpha and CRI, in line with those found in other investigations (Córdova et al, 2016;Garaigordobil, Martínez-Valderrey, & Aliri;2015;González & Guerrero, 2016;Martínez et al., 2017;Merlyn & Díaz, 2012;Quiroz, 2013;Robles, 2016;Rodríguez & Grijalva, 2017;Serrano et al., 2017;Suclla et al., 2015).
Nomological validity was evaluated by relating the scale of bullying to factors traditionally involved in it.The results were in line with previous studies (for example, Caprara et al., 2017;Fredrick & Demaray, 2018;Hesapcioglu et al., 2018;Mestre et al., 2012;Padilla-Walker et al., 2018;Quarshie et al., 2023), with correlations statistically significant, high and positive among bullying and physical and verbal aggressiveness, emotional instability and negative suicidal ideation; and negative between bullying and prosocial behavior and positive suicidal ideation.
Having a sample from a single educational district in the Dominican Republic is the main limitation of this research.In the future, it would be convenient to expand the study sample, both within the country and with the participation of other Spanish-speaking countries.In addition, the information collected comes from self-reports.It would be interesting to have research from other informants (e.g., parents and teachers).Finally, it would be interesting to support these results with longitudinal data.
This article provides evidence of the existence of a general dimension of bullying, in which it is difficult to distinguish the type of specific violent behavior of the bully.These results may stimulate the initiation or continuation of future research that tries to provide insight into the conceptual delimitation of bullying.In addition, it is the first full validation with confirmatory analysis of the Cisneros Self-test on Bullying.This contribution to the measurement of bullying may be of particular interest in the detection and intervention of bullying since it highlights the importance of conceiving bullying as a single dimension.It would be useful to address this question in future research.This could emphasize the influence of bullying on some factors and provide information to detect intervention strategies and programs that would help to avoid bullying, decrease aggressiveness or suicide ideation, and improve prosocial behavior.This is especially relevant in the context of the Dominican Republic, due to the high rates of bullying at schools (Parada et al., 2017).Analyzing which factors could intervene in this behavior and how to cope with them, in addition to being able to have a global measure of bullying could help, not only to elaborate intervention programs to decrease the rates but as well to prevent it.
is clear.On one hand : (a) chi-square statistic; (b) the comparative fit index (CFI;Bentler, 1990), with values of more than .90as indicative of adequate fit; (c) the root mean squared error of approximation (RMSEA;Steiger & Lind, 1980) of .05 or less as indicative of adequate fit.The RMSEA uses errors of prediction and measurement to assess the degree of match between the hypothesized and true models.Given the ordinal nature of the data WLSMV (weighted least square mean and variance corrected) estimation method was used(Finney & DiStefano, 2013).All statistical analyses were conducted with IBM SPSS Statistics, v. 20.0 and Mplus 8.2.

Table 1 .
Distribution of the sample by schools.

Table 2 .
Fit indexes for the structural equation models for the Cisneros Self-test on Bullying

Table 3 .
Factor loadings for the one-factor model of the Cisneros Self-test on Bullying

Table 4 .
Means, standard deviations, item homogeneity, and alpha if item deleted of the Cisneros Self-test on Bullying

Table 5 .
Correlations between the dimensions of the Cisneros Self-test onBullying and  aggression, Prosocial behavior, Emotional instability, and Suicidal Ideations.