Cognitive behavioral therapy for genderbased violence prevention: Evidence from a randomized trial in Mexico

Magaloni Kerpel, Beatriz; Dal Poggetto, Sofía Marinkovic; Murphy, Tommy E.; Pucci, Florencia; Fernández, Beatriz Serra Working Paper Cognitive behavioral therapy for genderbased violence prevention: Evidence from a randomized trial in Mexico IDB Working Paper Series, No. IDB-WP-1702 Provided in Cooperation with: Inter-American Development Bank (IDB), Washington, DC Suggested Citation: Magaloni Kerpel, Beatriz; Dal Poggetto, Sofía Marinkovic; Murphy, Tommy E.; Pucci, Florencia; Fernández, Beatriz Serra (2025) : Cognitive behavioral therapy for genderbased violence prevention: Evidence from a randomized trial in Mexico, IDB Working Paper Series, No. IDB-WP-1702, Inter-American Development Bank (IDB), Washington, DC, https://doi.org/10.18235/0013519 This Version is available at: https://hdl.handle.net/10419/324787 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/3.0/igo/ Cognitive Behavioral Therapy for Genderbased Violence Prevention: Evidence from a Randomized Trial in Mexico Beatriz Magaloni-Kerpel Sofía Marinkovic Dal Poggetto Tommy E. Murphy Florencia Pucci Beatriz Serra Fernández WORKING PAPER No IDB-WP-1702 InterA merican Development Bank Gender and Diversity Division May 2025 * Stanford Universit y ** Universidad de San Andrés and REDDES *** University of Wisconsin-Madison & REDDES **** REDDES Cognitive Behavioral Therapy for Genderbased Violence Prevention: Evidence from a Randomized Trial in Mexico Beatriz Magaloni-Kerpel* Sofía Marinkovic Dal Poggetto** Tommy E. Murphy** Florencia Pucci*** Beatriz Serra Fernández**** InterA merican Development Bank Gender and Diversity Division May 2025 Cataloging-in-Publication data provided by the Inter-American Development Bank Felipe Herrera Library Cognitive behavioral therapy for gender-based violence prevention: evidence from a randomized trial in Mexico / Beatriz Magaloni-Kerpel, Sofía Marinkovic Dal Poggetto, Tommy E. Murphy, Florencia Pucci, Beatriz Serra Fernández. p. cm. — (IDB Working Paper Series ; 1702) Includes bibliographical references. 1. Gender-based violence-Mexico-Prevention. 2. Violence-Prevention-Mexico. 3. Cognitive therapy-Mexico. I. Magaloni, Beatriz. II. Marinkovic Dal Poggetto, Sofía. III. Murphy, Tommy E. IV. Pucci, Florencia. V. Serra Fernández, Beatriz. VI. Inter-American Development Bank. Gender and Diversity Division. VII. Series. IDB-WP-1702 http://www.iadb.org Copyright © 2025 Inter-American Development Bank ("IDB"). This work is subject to a Creative Commons license CC BY 3.0 IGO (https://creativecommons.org/licenses/by/3.0/igo/legalcode). The terms and conditions indicated in the URL link must be met and the respective recognition must be granted to the IDB. Further to section 8 of the above license, any mediation relating to disputes arising under such license shall be conducted in accordance with the WIPO Mediation Rules. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to the United Nations Commission on International Trade Law (UNCITRAL) rules. The use of the IDB's name for any purpose other than for attribution, and the use of IDB's logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this license. Note that the URL link includes terms and conditions that are an integral part of this license. The opinions expressed in this work are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent. Abstract* Cognitive behavioral therapy (CBT) has become a powerful and effective tool to deal with violence in many at-risk areas in the world. However, its use for genderbased violence (GBV) and dating violence, although promising, has been limited and used as a response service for survivors, rather than for prevention. To understand to what extent such interventions can help provide tools and skills to young people in their impressionable years to produce behavioral changes that prevent GBV, we carried out such an intervention among high school students in the municipality of Ecatepec in Mexico. We assessed the intervention with a randomized control trial. We introduce the novelty of collecting objective measures from automated neuropsychological tests to explore whether CBT might be functioning through the development of subjects’ executive functions. Results from this intervention fail to show any clear change in self-reported violence. They do show, however, impacts on executive functions related to violence, such as emotional recognition and inhibitory control skills. JEL classifications: J16, I31, Z18, H43 Keywords: Gender-based violence, Violence prevention, Cognitive-behavioral therapy (CBT), Executive functions, Mexico * This project was conducted by the Mexican NGO Redes Sociales para el Desarrollo A.C. (REDDES) in collaboration with the Poverty, Governance, and Violence Laboratory (PovGov) at Stanford University. We are grateful for the funding provided by the Gender and Diversity Knowledge Initiative (GDLab) and the Citizen Security Division of the Inter-American Development Bank. We also deeply appreciate the valuable feedback from Andrés Moya and Amber Peterman, both advisers at GDLab. Our thanks extend to Araceli Sanz Martin at the Institute of Neuro-sciences at the University of Guadalajara for her guidance on neuropsychological tests, and to Eduardo Verduzco García Ruiz for his support in developing the interface to automate these tests on mobile devices. Additionally, we acknowledge the contribution of Enrique Ariza Moedano at the Centro Integral de Educación y Salud, who played a key role in implementing the intervention in Ecatepec de Morelos. 2 1. Introduction Rates of gender-based violence (GBV) in Mexico are alarmingly high. According to INEGI (2021), in 2021 roughly 70 percent of women (15-year-old or older) had experienced a violent situation at least once in their lifetime. Although gender violence is observed in many public and private spheres, one of its most pervasive manifestations takes place in intimate partner relationships. The patterns of interaction that occur in these relationships are learned and consolidated from an early age in dating relationships. The National Survey on Dating Violence (ENVINOV 2008) shows that dating violence is a social problem in Mexico, indicating that 76 percent of young people of both sexes between the ages of 15 and 24 who are in a dating relationship have suffered emotional violence (insults, humiliation, threats), 15 percent have suffered physical violence, and 16.5 percent sexual violence. Several studies show that dating violence is bidirectional (Straus, 2004; O’Leary & Smith-Slep, 2003; Rey-Anacona, 2013), with men reporting greater victimization and women greater perpetration (Cortés-Ayala et al., 2015). However, the most severe physical aggressions are from men to women (O’Leary et al., 2008). Therefore, it is of utmost importance to develop actions that can help prevent these harmful dynamics in relationships from the beginning during the dating period in adolescence. For this, it is necessary to recognize that behind violent actions in relationships, there are different drivers that somewhat operate interdependently at different levels. At an individual level, those drivers are lack of self-esteem, anxiety and depression, substance abuse and/or impulsivity, and lack of selfcontrol and/or emotional regulation. At a community level, factors such as social isolation, lack of support, or weak law enforcement usually contribute. Finally, at a societal level, gender norms and roles may encourage male dominance and control over women. Because of this multifaceted nature of GBV, interventions that work at different levels are more likely to be effective. Cognitive behavioral therapy (CBT) is one such intervention. It can promote understanding and regulation of emotions associated with situations where gender beliefs are involved. Furthermore, it encourages critical analysis of thought distortions associated with gender stereotypes and norms. It also trains people in the use of skills and tools for self-control and self-regulation, practicing alternative behaviors and violence-free relationships. There is now substantial evidence that CBT group-based interventions are effective in addressing violence, crime and recidivism reduction among young people at-risk (Heller et al. 2017; Blattman et al., 2017, 2022). Nevertheless, 3 CBT interventions aimed at addressing GBV have predominantly been reactive, focusing on survivors (Archer 2000; Babcock et al. 2004; Feder & Wilson 2008), and their effectiveness in lowand middle-income countries has been infrequently assessed (Gupta et al. 2017). In terms of their results, these studies fail to be entirely satisfactory in different dimensions, most notably, they rarely explore the mechanisms through which CBT contributes to prevent GBV. One plausible mechanism involves what are known in the literature as “executive functions” (Barkley 2012), those that are required to maintain an appropriate problem-solving set for attainment of a future goal. Some studies suggest, for example, that aggressive individuals lack elements of these functions, such as cognitive flexibility or inhibition (Burgess 2020). This project assessed whether an intervention based on CBT is capable of changing gender attitudes and behaviors in teenagers in an area of Mexico with high prevalence of GBV. Briefly, we ask whether CBT can contribute to prevent GBV and whether the therapy might be affecting executive functions. Furthermore, our experimental design allows us to address if the intervention generates spillovers in subjects not directly treated, which would facilitate scalability. 2. Related Literature 2.1 CBT-based Interventions CBT is a type of psychotherapy that focuses on the relationship between thoughts, emotions, and behaviors. It posits that our thoughts, beliefs, and attitudes influence how we feel and behave in certain situations. CBT considers the situations or events that trigger a person’s emotional and behavioral responses. These can include external factors such as stressful events, social interactions, or life transitions and postulates that a person’s thoughts and beliefs about a situation can influence their emotional and behavioral responses. If a man, for example, believes he should be more intelligent than any woman, because men are superior to women, in a situation where she performs better, he may feel threatened by the woman’s intelligence, he may think he is incompetent or worthless, may feel anxious or depressed about it, or frustrated and angry, and can act violently against her. CBT-based interventions have shown to be effective tools for violence reduction and prevention (Heller et al., 2017; Blattman et al., 2017, 2022). Although there have been many CBT group interventions, none has addressed GBV and dating violence prevention in older adolescent (high school) ages. Interventions have only been reactive for GBV (Murphy et al., 2020; Berg 4 Nesset et al., 2021; Romero-Martínez et al., 2016, 2021), although experts have pointed out its potential as a prevention tool (Crooks et al., 2007). This could be particularly important if carried out in adolescents or young adults if we consider the long tradition in social psychology (e.g., Mannheim, 1952; Krosnick & Awin, 1989) that suggests older adolescents and young adults (up to circa 25 years) are in their “impressionable years”; beliefs and attitudes in that age range are particularly malleable and hence can have long-term impacts. 2.2 The Executive Functions Connection CBT programs like the one we implemented aim to influence and reduce violent behavior by the identification and challenging of negative sexist thoughts and beliefs, developing empathy and respect for others, working on emotional awareness and regulation, building communication and conflict resolution skills, and practicing non-violent behaviors, using anger management techniques, and coping skills to manage difficult emotions. Therefore, the mechanisms through which it is believed the intervention may be effective for violence prevention are through sexist beliefs changes (cognitive restructuring) and through the acquisition of several cognitive skills for behavior regulation. It is well established that poor executive functions correlate with increased violence and aggression (Broomhall, 2005). There are a variety of definitions of executive functions, each highlighting different aspects of them (Barkley, 2012: 4), but most scholars would agree that they are neuropsychological functions involving the ability to maintain an appropriate problem-solving set for attainment of a future goal (Welsh & Pennington, 1988: 201-202), and include qualities such as working memory, processing speed, inhibition, and cognitive flexibility. The review by Burgess (2020) suggests that aggressive individuals lack cognitive flexibility and inhibition. Interventions involving skills training and behavior modification (i.e., practicing skills that help to delay gratification, improving verbal skills to deescalate situations, reasoning through future consequences, and inhibiting impulses) may be able to help those who are at a higher risk for future violence or aggression. The question is whether it is possible to exercise an improvement in this type of skills and executive functions through a group intervention. There are already a few examples of interventions for violence reduction showing successful results in executive functions modification. For 5 example, the Contexto Program is a community-based intervention, run at the Universidad de Valencia, based on the ecological model framework that aims to reduce risk factors and increase protective factors for intimate partner violence offenders, promoting the change of beliefs and attitudes that maintain gender violence, offering alternatives to the men who participate in the program, with the aim that they can establish healthy couple relationships based on respect and free of violence. The evaluation of potential neuropsychological changes (emotional decoding, perspective taking, emotional empathy and cognitive flexibility) due to this program showed that increases were observed in cognitive empathy (emotional decoding and perspective taking) and in cognitive flexibility (Romero-Martínez et al., 2016). Further research (Romero-Martínez et al., 2021) pointed out that when the program was complemented with specific cognitive training the intimate partner violence (IPV) perpetrators improved their processing speed and cognitive flexibility, and that these participants presented lower risk of recidivism after the intervention. In addition, Ron‑Grajales et al. (2021) showed that mindfulness meditation training improved inhibitory control and, therefore, may have a positive effect on mitigating violent behavior in young offenders. An effective way to evaluate the impact of the intervention on gender and dating violence prevention can be through the assessment of changes in sexist beliefs, as well as through changes in executive functions related to the control of violent responses. 2.3 Interventions in Schools There are several randomized control trials (RCTs) done in schools where CBT therapy is given as a treatment but only a few center their attention to GBV. Olatunbosun (2022) uses a CBT intervention to reduce bullying, that might include gender-related issues but not as a main result. Miller et al. (2011) and Van Starrenburg et al. (2017) use CBT therapy in schools as a preventive intervention for child anxiety. In a similar direction, Manassis et al. (2010), studies the impact of CBT therapy in schools for anxious or depressive symptoms. There are some RCTs that focus on gender attitudes, hence not directly on GBV (yet they could have an impact on the prevention of GBV), but they use treatments other than CBT. Dhar et al. (2022), for example, evaluates an intervention based on gender equality discussions in the classroom that aimed to reduce the support for societal norms that restrict women’s and girls’ opportunities, and found that the program made attitudes 12 Table 2. Analytical Instruments for Workshop Evaluation Risk factor associated with violent behavior Intervention component to address risk factor Analytical Instrument for risk factor assessment Description Self-reported questionnaire Automated neuropsychological test Sexist Beliefs Critical Review of Thoughts & Beliefs Brief Ambivalent Sexist Inventory (BASI) 12 items with two subscales: Hostile Sexism (HS): 6 items (Cronbach’s α = .85) subscale to measure traditional sexism oriented toward the perception of inferiority of the other sex. Benevolent Sexism (BS): 6 items (Cronbach’s α = .73) subscale to assess sexist attitudes with a positive affective tone that allows men to behave pro-socially, although it helps justify and maintain gender inequality. Impulsivity Mindfulness State Impulsivity Scale (SIS) 20 items (Cronbach’s α = .884) that assess impulsive behaviors and tendencies across three subscales: Motor Impulsivity or Automatism: measures impulsive actions and behaviors, such as acting without thinking, being restless, or having difficulty sitting still. Attentional Impulsivity: assesses difficulties in maintaining focus and attention, including distractibility and difficulties with concentration. Non-planning Impulsivity or Gratification: evaluates impulsive decision-making and lack of future planning or consideration of consequences. Emotional Stop Signal Test (ESST) Participants must rapidly indicate the sex (man or woman) of serially presented images (112 in total) of human faces showing different emotions on the same number of occasions (joy, fear, anger, or neutral) by keypress, except on trials with a visual stop-signal (red frame). It assesses inhibitory response capacity by establishing the respondent’s ability to respond correctly to the stimulus and their ability to inhibit their response when necessary. It is assessed by the proportion of successful trials. Lack of emotional recognition/empathy Emotional recognition skills Reading the Mind in the Eyes Test (RMET) Participants must choose out of the seven words of the primary emotions (joy, fear, disgust, anger, sadness, surprise, neutral) the one that describes how the person in each of the 42 sequentially presented pictures of the region of the eyes - including eyebrows and part of the nose - of men, women, and children (sex equally represented) is feeling. Higher scores indicate better ability to read and interpret emotions and mental states from the eyes. 13 Table 2., continued Risk factor associated with violent behavior Intervention component to address risk factor Analytical Instrument for risk factor assessment Description Self-reported questionnaire Automated neuropsychological test Lack of e motional regulation Emotional regulation skills Emotional Regulation Questionnaire (ERQ-CA-9) 9 items (Cronbach’s α = .830), that assess two main strategies of emotional regulation separately: Cognitive Reappraisal: This strategy involves changing one's interpretation or perspective of a situation to alter the emotional response. It includes techniques such as reframing, finding silver linings, or focusing on positive aspects of a situation. Expressive Suppression: This strategy involves inhibiting or suppressing the outward expression of emotions. It may include efforts to hide or conceal emotions from others. Emotional Stop Signal Test (ESST) Participants must rapidly indicate the sex (man or woman) of serially presented images (112 in total) of human faces showing different emotions on the same number of occasions (joy, fear, anger, or neutral) by keypress, except on trials with a visual stop-signal (red frame). It assesses emotional regulation capacity by comparing how the emotion shown in the test affects the respondent’s ability to act correctly to the stimulus and their ability to inhibit their response when necessary. Lack of c ognitive flexibility/Problem -Solving Problem-Solving Skills Interpersonal Conflict Resolution Scale (ICRS) 22 items (Cronbach’s α = .810) structured around three factors for conflict resolution: Collaborative means: the tendency to use cooperative and constructive strategies in resolving conflicts (negotiation, open communication, finding solutions that satisfy all parties involved, teamwork and mutual problem-solving) Passive means: the inclination to avoid conflict or to give in to the demands of others. Aggressive means: the use of confrontational or forceful strategies in dealing with conflict (threats, manipulation, or assertive behavior to dominate or win the conflict, often disregarding the perspectives or needs of others) Wisconsin Card Sorting Test (WCST) Participants are asked to match cards based on different criteria (such as color, shape, or number), but the matching rule changes periodically without warning. The participant must use feedback to adjust their strategy and discover the new rule. The test measures skills like problem-solving, adaptability, and the ability to learn from mistakes. 14 Table 2., continued Risk factor associated with violent behavior Intervention component to address risk factor Analytical Instrument for risk factor assessment Description Self-reported questionnaire Automated neuropsychological test Scale of Risk of Violence Assessment for Adolescents (SRVAA) 25 items (Cronbach’s α = .836) structured around six factors: Negative Self-concept: evaluates the adolescent's self-perception, including feelings of worthlessness, low self-esteem, and negative attitudes towards oneself. Lack of Self-control: measures the adolescent’s difficulty in managing impulses and emotions. Traits of Self-harm: assesses tendencies toward self-destructive behaviors, such as self-injury or suicidal thoughts. Proactive Violence: measures calculated, goal-directed violent behavior. Verbal Violence: evaluates the use of aggressive language, threats, or insults as a form of violence. Reactive Violence: measures violent behavior that occurs in response to perceived threats or provocations. Social Desirability Index (SDI) 13 item short form of the Marlowe-Crowne Social Desirability Scale (Cronbach’s α = .764) 15 Table 2 summarizes the analytical instruments used to measure the impact on each of the risk factors targeted by the components of the intervention. For a more detailed description of each instrument, see Annex 1. Annex 2 includes all the self-reported questionnaires with all their items and the Likert scale for each, both in Spanish, the language in which they were applied during the intervention, and in English. In addition to the instruments used to evaluate the factors targeted by the intervention, two additional self-report questionnaires were administered to complete the assessment. These include the Scale of Risk of Violence Assessment for Adolescents (SRVAA), which aims to determine if the intervention can reduce violence risk, and a Social Desirability Index, which assesses the extent to which students’ responses may be susceptible to bias originating from their beliefs about socially desirable answers. The SRVAA is based on WHO definitions of violence risk factors and includes six subscales: Negative Self-concept, Lack of Self-control, Traits of Self-harm, Proactive Violence, Verbal Violence, and Reactive Violence. It evaluates the frequency of these violent behaviors. All self-reported questionnaires, except the State Impulsivity Scale (SIS), have been validated on Mexican adolescent populations prior to this intervention. The SIS has been validated using a Spanish adolescent population. 3.3 Taking the Baseline and Endline During the first three weeks of September 2023, a group of evaluators accessed each classroom of first-year students in all schools participating in the study at tutoring time and administered a baseline survey for all outcome variables. Through a battery of modems, a secure Internet network was shared with the students, to which they connected with their mobile phones. Once connected, access to the evaluation was shared with them through a QR code. Each subject introduced personal details, and then the data collection took place, which lasted between 20 to 30 minutes. Before carrying out the data collection, schools were provided with an informed consent form to send to parents. Parents were asked to return the signed sheet if they agreed for their children to participate in the evaluation. Only students for whom parental authorization was available were included in the data collection, yet the great majority of them received the authorization (86.1 percent). For the endline data collection, we followed the same procedure as the baseline. However, the schools unexpectedly concluded the semester early, under the directive of educational authorities to offer additional support to struggling students. Consequently, it was not feasible to assess 16 all students who had initially provided informed consent. Only 55.4 percent of the students who took part in the baseline completed the endline data collection at the conclusion of the intervention. 3.4 Experimental Design As previously mentioned, seven schools consented to participate in the experiment. Among these, five schools operate on both morning and afternoon shifts, while the remaining two schools have a single shift. Within each shift, the number of courses offered by the schools ranges from 3 to 8 (see Table 3 below). The workshops were conducted at the course level. However, to evaluate potential spillovers in information transmission among students, we chose the school-shift as the unit of randomization rather than the individual course. Consequently, our cluster sample size is N=12, representing the total number of school-shifts included in the study. The randomization process was conducted in two steps: Step 1: The initial phase involved assigning 4 out of the 12 school-shifts to the pure control group. None of the courses within these 4 school-shifts received the workshops during the intervention. Step 2: For the 8 school-shifts not assigned to the pure control group, each course within these shifts was randomly assigned to either the treatment or control group. To ensure a balanced distribution of courses between the two groups, we used pre-determined block sizes for randomization. For instance, in school-shifts with 6 courses, 3 were randomly assigned to the treatment group and 3 to the control group. For school-shifts with an odd number of courses, the blocked randomization was balanced across both morning and afternoon shifts. For example, in School 3, the morning shift had 4 treated courses, while the afternoon shift had 3 treated courses. This 2-step procedure left us with 8 school-shifts that contained both treated and untreated courses. The untreated courses within these 8 school-shifts formed the potentially contaminated control group. Due to the variation in the number of courses per school-shift, there was a risk of having unbalanced group sizes for the treatment, contaminated control, and pure control groups. It is important to note that the randomization resulted in 21 courses being assigned to the treatment group, 22 courses to the contaminated control group, and 20 courses to the pure control group. 17 3.5 Potential Spillovers As mentioned earlier, the random assignment assumed that potential spillovers occur at the schoolshift level. If this assumption holds, the true effectiveness of the CBT workshop is captured by the total effect of the treatment on both treated and non-treated individuals within shifts that have at least one treated course (Baird et al., 2018). This means we must consider both the direct effects on treated individuals and the indirect effects on non-treated individuals. We refer to this combined impact as the Total Causal Effect (TCE). Formally, TCE = Direct Effect of CBT workshop + Indirect Effect of CBT workshop If the treatment positively influences the outcomes of non-treated individuals in the same shift, comparing the average outcomes of the treated group to the contaminated control group will underestimate the TCE. To address this, we compare the average outcomes of the contaminated control group to the pure control group to isolate the Indirect Effect of the CBT workshop. Similarly, we compare the average outcomes of the treated group to the pure control group to measure the Direct Effect of the CBT workshop. 4. Results 4.1 Workshop Attendance Attendance at the workshops was 77.5 percent of all sessions taught. Figure 1 shows the percentage of attendance at each of the workshop sessions for each of the eight treatment school-shifts. Although the workshop was compulsory and held during school hours, several factors affected attendance. Firstly, there was little control over attendance, especially in one shift-school. Secondly, the workshop ended close to the winter holidays, and schools decided to end their classes earlier than planned to provide extra instruction to students who had fallen behind. In the end, there were fewer students present in the last sessions of all groups as attendance was not mandatory. Additionally, two groups had only 11 sessions and another group only had 9 sessions out of 12. Finally, in one school, only a small number of students were able to participate in the endline data collection. Table 3 shows the number of students participating in the survey at the baseline and the endline, and the comparison of the number of students evaluated by treatment arm specified by sex are presented in Table 4. 18 Figure 1. Percentage of Attendance at Each of the Workshop Sessions for Each of the Eight Treatment School-shifts Table 3. Courses and Students Participating in the Study School Morning shift Afternoon shift (1) Course s (2) Total students (3) Students at baseline (4) Students at endline (5) Courses (6) Total students (7) Students at baseline (8) Students at endline 1 0 0 0 0 6 286 212 35 2 5 244 229 93 5 246 192 94 3 7 295 253 166 7 364 298 157 4 5 174 161 101 5 215 190 139 5 8 286 248 146 0 0 0 0 6 3 100 97 63 3 73 69 45 7 5 115 104 75 4 86 85 70 Total 1214 1092 644 1270 1046 540 Table 4. Comparison of Students by Sex and Treatment Arm Baseline. 2138 Endline: 1184 (55.4% of baseline) Pure control: 646 Contaminated control: 804 Treatment: 688 (32.6%) Pure control: 422 (35.6%) Contaminated control: 373 (31.5%) Treatment: 389 (32.9%) Girls: 312 (48.3%) Boys: 334 (51.7%) Girls: 413 (51.4%) Boys: 391 (48.6%) Girls: 356 (51.7%) Boys: 332 (48.3%) Girls: 220 (52.1%) Boys: 202 (47.9%) Girls: 192 (51.5%) Boys: 181 (48.5%) Girls: 215 (55.3%) Boys: 174 (44.7%) 19 4.2 Baseline Results The number of students participating in baseline data collection at each school is included in columns 3 and 6 in Table 3. These students are those whose parents signed the informed consent sheet and who attended class on the day of the evaluation. The percentage of students participating corresponds to 87 percent of enrolled students. Table 5 shows results for all students, and Table 6 separates results for female and male students. Additional baseline information is included in Annex 5. A statistical difference in baseline between the control and treatment groups is only reported for two variables: “Score BASI” and “Score Passive Means.” For “Score BASI” this applies to both males and females, while for “Score Passive Means,” the difference is only recorded for males. In addition, for “Score Aggressive Means,” there is a statistically significant difference for males. For the other variables in the baseline, there are no significant differences. Table 5. Baseline Comparisons for All Relevant Variables: All Students Variables Treatment Pure Control Difference p-value Age 15.301 15.312 -0.011 0.921 Sex 0.515 0.488 0.027 0.306 % Right RMET 0.549 0.553 -0.004 0.606 % Right ESST 0.638 0.635 0.003 0.704 Count Correct Go 57.136 56.385 0.751 0.420 Count Correct Stop 14.669 15.500 -0.831 0.046 Score BASI 34.242 36.167 -1.925 0.001 Score SIS 20.275 20.140 0.135 0.823 Score Attentional 7.338 7.192 0.146 0.550 Score Automatism 6.132 6.192 -0.061 0.764 Score Gratification 6.805 6.755 0.050 0.833 Score Reevaluation 16.602 16.886 -0.283 0.282 Score Suppression 12.444 12.488 -0.044 0.831 Score SRVAA 34.292 34.057 0.235 0.563 Score Negative Self-concept 11.541 11.314 0.226 0.260 Score Lack of Self-control 5.011 5.057 -0.046 0.577 Score Traits of Self-harm 5.136 5.156 -0.020 0.811 Score Proactive Violence 4.753 4.771 -0.018 0.803 Score Verbal Violence 4.180 4.150 0.030 0.690 Score Reactive Violence 3.672 3.609 0.062 0.334 Score SDI 7.889 7.963 -0.074 0.551 Score ICRS 36.965 35.710 1.255 0.104 Score Aggressive Means 6.185 6.106 0.080 0.776 Score Collaborative Means 18.550 18.070 0.480 0.296 Score Passive Means 12.230 11.534 0.696 0.042 20 Table 6. Baseline Comparisons for All Relevant Variables by Gender Female students Male students Variables Treatment Pure Control Diff p-value Treatment Pure Control Diff p-value Age 15.239 15.249 -0.010 0.949 15.367 15.372 -0.006 0.972 % Right RMET 0.564 0.566 -0.002 0.877 0.533 0.541 -0.008 0.464 % Right ESST 0.634 0.619 0.015 0.172 0.643 0.651 -0.008 0.474 Count Correct Go 56.723 53.939 2.784 0.037 57.577 58.703 -1.126 0.382 Count Correct Stop 14.909 15.684 -0.775 0.174 14.412 15.326 -0.914 0.133 Score BASI 32.375 34.040 -1.665 0.029 36.227 38.190 -1.963 0.022 Score SIS 19.819 20.635 -0.816 0.344 20.763 19.658 1.105 0.194 Score Attentional 7.153 7.274 -0.122 0.728 7.537 7.113 0.424 0.215 Score Automatism 5.901 6.423 -0.522 0.066 6.379 5.969 0.410 0.155 Score Gratification 6.766 6.939 -0.173 0.608 6.847 6.577 0.271 0.421 Score Reevaluation 17.383 17.628 -0.245 0.481 15.765 16.179 -0.413 0.288 Score Suppression 12.975 13.059 -0.084 0.763 11.875 11.943 -0.069 0.816 Score SRVAA 34.998 35.111 -0.114 0.847 33.545 33.054 0.491 0.374 Score Negative Self-concept 12.575 12.380 0.196 0.511 10.445 10.301 0.144 0.560 Score Lack of Self-control 4.983 5.060 -0.077 0.507 5.041 5.054 -0.013 0.909 Score Traits of Self-harm 5.231 5.289 -0.059 0.623 5.036 5.029 0.007 0.948 Score Proactive Violence 4.534 4.663 -0.129 0.194 4.985 4.874 0.111 0.295 Score Verbal Violence 4.211 4.235 -0.024 0.826 4.147 4.069 0.078 0.456 Score Reactive Violence 3.464 3.485 -0.021 0.781 3.892 3.728 0.164 0.110 Score SDI 7.654 7.742 -0.088 0.611 8.138 8.172 -0.034 0.847 Score ICRS 37.364 36.563 0.801 0.456 36.533 34.887 1.645 0.140 Score Aggressive Means 5.882 6.380 -0.498 0.214 6.514 5.841 0.672 0.088 Score Collaborative Means 19.161 18.358 0.803 0.205 17.889 17.793 0.096 0.885 Score Passive Means 12.322 11.826 0.496 0.301 12.130 11.253 0.877 0.072 In addition to comparing the control and treatment groups, analyzes were conducted based on age and school shift. No significant differences were found in any of the measured variables concerning these factors. Differences were indeed observed, however, between boys and girls across several variables, as outlined below. In the baseline assessments, our population’s BASI questionnaire (sexist beliefs) responses indicate similar levels to those of other Latin American countries and align with the prevailing literature, which suggests that men typically exhibit significantly higher levels of both types of sexism compared to women (Rollero et al., 2014; Martínez-Baquero & Vallejo-Medina, 2024). 21 Regarding the SIS (impulsivity questionnaire), our population’s scores surpass those of the population in which the scale was validated. Notably, there are no significant initial score differences between males and females across the total scale or its measured dimensions. Analysis of ERQCA-9 (emotional regulation questionnaire) baseline data reveals that male students tend to score higher in emotional regulation through expressive suppression, while female students score significantly higher in cognitive reappraisal, suggesting superior emotional regulation strategies among females, consistent with existing literature (Gross & John, 2003). Findings from the baseline ICRS (conflict resolution) assessment indicate no remarkable distinctions in the use of aggressive and passive methods for emotional regulation between boys and girls. However, girls tend to employ collaborative approaches to conflict resolution more frequently, aligning with prior research by Shute and Charlton (2006). Conversely, a study by Fariña et al. (2021) showed no significant gender variations in conflict resolution styles. Our population’s scores on the violence risk scale are somewhat lower than those found in Muñoz’s scale validation work among Mexican adolescents. Gender differentiation reveals significantly higher total scores among girls. Notably, girls score higher in negative self-concept and self-injury traits, while boys score higher in reactive and proactive violence. There were no significant differences by sex for the lack of self-control and verbal violence subscales. The relationship between each of the evaluated variables and violent behaviors, based on findings from previous research, is described in the Introduction and in greater detail in Annex 1. But the data collected allows us to explore whether there is a relationship between the variables measured at baseline, particularly in relation to the risk of violence and its subscales. Table 7 presents Pearson correlation values between pairs of study variables, highlighting positive correlations in blue and negative correlations in orange, with more intense colors indicating stronger correlations. Naturally, the subscales show strong positive correlations with the total impulsivity (SIS) and violence (SVRAA) scales, as these are constructed by summing the scores of their respective subscales. But a notable finding is the negative correlation between self-reported variables associated with negative concepts (all subscales of violence risk assessment, impulsivity, and the use of aggressive means to resolve conflicts) and the Social Desirability Index. This suggests that participants tend to respond in a socially desirable manner, potentially underestimating the scores obtained in these variables. At the same time, the Social Desirability Index shows a positive correlation with cognitive reappraisal, indicating a possible overestimation in this dimension. These 28 4.3.4 Neuropsychological Tests Results Tables 10 and 11 present the results for the Automated Neuropsychological Tests. As stated in Section 3, an improvement in the performance on any of these tests is implied by the achievement of a higher score and coefficients in this table should be interpreted as percentage points. The findings indicate that engagement in the workshop has a positive effect on the emotional recognition abilities among participants. Table 10. Neuropsychological Tests Results: Main Effects (Model 1) (1) (2) (3) Emotional recognition (RMET) Impulsivity (ESST) Cognitive flexibility (WCST) Treatment 0.040* 0.014 0.008 p-value (0.022) (0.214) (0.616) Adjusted p-value for MHT [0.068] [0.321] [0.617] Female 0.015 0.002 0.003 p-value (0.188) (0.799) (0.784) Constant -0.063*** -0.010 0.586*** p-value (0.002) (0.189) (0.000) N 816 844 927 In parenthesis, p-values corresponding to clustered SE at the school-shift level, adjusted by small number of clusters using reg_sandwich command in Stata. In brackets, p-values corrected for Multiple Hypothesis Testing using Anderson’s sharpened q-values. *if Anderson’s q-values<.1; ** if Anderson’s q-values<.05; *** if Anderson’s q-values<.01 Table 11. Neuropsychological Tests Results: Effects by Gender (Model 2) (1) (2) (3) Emotional recognition (RMET) Impulsivity (ESST) Cognitive flexibility (WCST) Treatment 0.022 -0.005 -0.006 p-value (0.207) (0.706) (0.776) Adjusted p-value for MHT [0.415] [0.928] [0.928] Female -0.001 -0.015 -0.011 p-value (0.928) (0.286) (0.559) Female*Treatment 0.033 0.036 0.028 p-value (0.128) (0.041) (0.174) Adjusted p-value for MHT [0.386] [0.165] [0.415] Constant -0.055** -0.001 0.593*** p-value (0.007) (0.875) (0.000) N 816 844 927 In parenthesis, p-values corresponding to clustered SE at the school-shift level, adjusted by small number of clusters using reg_sandwich command in Stata. In brackets, p-values corrected for Multiple Hypothesis Testing using Anderson’s sharpened q-values. *if Anderson’s q-values<.1; ** if Anderson’s q-values.<.05; *** if Anderson’s q-values<.01 29 4.3.5 A Comment on Robustness Checks To ensure the robustness of our findings, we investigate whether social desirability bias may have influenced the treatment effects by incorporating interactions with a social desirability index (SDI). We estimate both the main model and a fully interacted model with SDI, testing for any significant impact on the treatment effects. As the results show no significant coefficients for these interactions, the complete tables are provided in Annex 7 for reference. This approach allows us to address the potential concern of social desirability bias without detracting from the main findings. 4.3.6 Statistical Analysis Using a Less Restrictive Approach This section presents a less restrictive statistical analysis to further explore the potential effects of our intervention. Initially, our experimental design randomized at the school shift level while also implementing classroom-level randomization. Within treated school shifts, only a subset of classrooms was randomly assigned to the intervention group. However, since no spillover effects were detected in the control classrooms within treated school shifts, these classrooms are included in the analysis, increasing the number of observations. Furthermore, the study population consists of first-semester high school students who recently transitioned from various middle schools. Given that no significant baseline differences were found in the measured variables across different school shifts, a less restrictive analytical approach is justified. In this analysis, standard errors are clustered at the classroom level, under the assumption that observations within a classroom are correlated, while those across different school shifts are not necessarily so. Accordingly, the following section presents statistical results using the classroom as the unit of randomization, with all classrooms that did not receive the workshop grouped into the control group. This approach assumes that all school shifts are equivalent and clusters errors at the classroom level. Tables 12 and 13 provide questionnaire results, while Tables 14 and 15 present the results of the neuropsychological tests under these revised specifications. In this more comprehensive analysis, we have observed statistical improvements in both emotional recognition, as assessed by the RMET neuropsychological test, and inhibitory control, as measured by the ESST. Additionally, students who participated in violence prevention workshops showed enhanced use of emotional regulation strategies, according to the ERQ-CA-9 questionnaire, compared to those who did not attend these workshops. 30 These results suggest that the workshop activities positively influence key variables associated with violent behaviors. Consequently, these workshops may represent an effective, practical, and easily implementable strategy for preventing gender-based and dating violence within this population. 31 Table 12. Questionnaires Results: Main Effects (Model 1) (1) (2) (3) (4) (5) (6) (7) (8) Sexist Beliefs (BASI) Impulsivity (SIS - All) Emotional Regulation (ERQ -CA-9 - Cognitive Reappraisal) Emotional Regulation (ERQ -CA-9 - Expressive Suppression) Cognitive flexibility (ICRS - Aggressive mea ns) Cognitive flexibility (ICRS - Collaborative means) Cognitive flexibility (ICRS - Passive means) Assessment of risk of violence (SVRAA - All) Treatment 2.065* 0.716 1.046* 0.672* 0.169 0.591 -0.549 0.515 p-value (0.011) (0.444) (0.021) (0.029) (0.723) (0.365) (0.228) (0.472) Adjusted p-value for MHT [0.077] [0.540] [0.077] [0.077] [0.724] [0.540] [0.457] [0.540] Female -0.340 0.798 0.515 0.107 -0.095 0.546 0.140 -0.177 p-value (0.619) (0.343) (0.204) (0.712) (0.793) (0.363) (0.737) (0.730) Constant -2.120*** 0.227 -1.986*** -1.117*** 0.435 -2.270*** -0.646** 1.732*** p-value (0.000) (0.725) (0.000) (0.000) (0.201) (0.000) (0.040) (0.002) N 2.065* 0.716 1.046* 0.672* 0.169 0.591 -0.549 0.515 In parenthesis, p-values corresponding to clustered SE at the classroom level, adjusted by small number of clusters using reg_sandwich command in Stata. In brackets, p-values corrected for Multiple Hypothesis Testing using Anderson’s sharpened q-values. *if Anderson’s q-values<.1; ** if Anderson’s q-values<.05; *** if Anderson’s q-values<.01 32 Table 13. Questionnaires Results: Effects by Gender (Model 2) (1) (2) (6) (7) (8) (9) (10) (11) Sexist Beliefs (BASI) Impulsivity (SIS - All) Emotional Regulation (ERQ -CA9 - Cognitive Reappraisal) Emotional Regulation (ERQ -CA9 - Expressive Suppression) Cognitive flexibility (ICRS - Aggressive means) Cognitive flexibility (ICRS - Collaborative means) Cognitive flexibility (ICRS - Passive means) Assessment of risk of violence (SVRAA - All) Treatment 2.648 0.169 1.081 0.416 0.053 -0.005 -0.590 1.105 p-value (0.026) (0.859) (0.106) (0.378) (0.930) (0.995) (0.423) (0.268) Adjusted p-value for MHT [0.424] [0.996] [0.825] [0.825] [0.996] [0.996] [0.825] [0.825] Female 0.020 0.452 0.536 -0.055 -0.169 0.170 0.114 0.179 p-value (0.980) (0.695) (0.271) (0.876) (0.686) (0.815) (0.803) (0.789) Female*Treatment -1.082 1.017 -0.065 0.481 0.214 1.097 0.075 -1.087 p-value (0.481) (0.515) (0.942) (0.435) (0.797) (0.396) (0.939) (0.298) Adjusted p-value for MHT [0.825] [0.825] [0.996] [0.825] [0.996] [0.825] [0.996] [0.825] Constant -2.306*** 0.410 -1.997*** -1.033*** 0.474 -2.071*** -0.632** 1.547*** p-value (0.000) (0.573) (0.000) (0.001) (0.188) (0.000) (0.050) (0.009) N 2.648 0.169 1.081 0.416 0.053 -0.005 -0.590 1.105 In parenthesis, p-values corresponding to clustered SE at the classroom level, adjusted by small number of clusters using reg_sandwich command in Stata. In brackets, p-values corrected for Multiple Hypothesis Testing using Anderson’s sharpened q-values. *if Anderson’s q-values<.1; ** if Anderson’s q-values<.05; *** if Anderson’s q-values<.01 33 Table 14. Neuropsychological Tests Results: Main Effects (Model 1) (1) (2) (3) Emotional recognition (RMET) Impulsivity (ESST) Cognitive flexibility (WCST) Treatment 0.046** 0.017* 0.015 p-value (0.004) (0.066) (0.167) Adjusted p-value for MHT [0.014] [0.099] [0.167] Female 0.009 -0.009 0.006 p-value (0.268) (0.210) (0.385) Constant -0.066*** -0.006 0.578*** p-value (0.000) (0.338) (0.000) N 0.046** 0.017* 0.015 In parenthesis, p-values corresponding to clustered SE at the classroom level, adjusted by small number of clusters using reg_sandwich command in Stata. In brackets, p-values corrected for Multiple Hypothesis Testing using Anderson’s sharpened q-values. *if Anderson’s q-values<.1; ** if Anderson’s q-values<.05; *** if Anderson’s q-values<.01 Table 15. Neuropsychological Tests Results: Effects by Gender (Model 2) (1) (2) (3) Emotional recognition (RMET) Impulsivity (ESST) Cognitive flexibility (WCST) Treatment 0.027 -0.007 0.006 p-value (0.194) (0.542) (0.680) Adjusted p-value for MHT [0.305] [0.650] [0.680] Female -0.002 -0.024*** 0.000 p-value (0.793) (0.008) (0.981) Female*Treatment 0.034 0.045*** 0.017 p-value (0.065) (0.001) (0.203) Adjusted p-value for MHT [0.195] [0.008] [0.305] Constant -0.060*** 0.001 0.581*** p-value (0.000) (0.824) (0.000) N 0.027 -0.007 0.006 In parenthesis, p-values corresponding to clustered SE at the classroom level, adjusted by small number of clusters using reg_sandwich command in Stata. In brackets, p-values corrected for Multiple Hypothesis Testing using Anderson’s sharpened q-values. *if Anderson’s q-values<.1; ** if Anderson’s q-values<.05; *** if Anderson’s q-values<.01 4.3.7 Possible Impact of the Intervention on Grades To evaluate the potential impact of the intervention on student academic performance, we requested the grades for each subject taken by students in both the control and treatment groups from the schools at the end of the semester. It was not possible to obtain differential grade data, meaning we could not compare individual differences before and after the intervention, which would have provided greater statistical power. This is because there are no prior grades available for these groups, as they are first-semester high school students. After analyzing the grades, we did not detect any significant differences between the control and treatment groups, nor were there any differences in performance between male and female students. 34 4.4 Discussion 4.4.1 Analysis of the Results: Impact of the CBT-Based Intervention The findings of this research suggest cognitive-behavioral therapy (CBT)-based intervention can potentially improve emotional recognition, inhibitory control, and emotional regulation among high school students. While other assessed dimensions—such as gender beliefs, risk of violence, and cognitive flexibility—did not show statistically significant improvements, the coefficients obtained bear the expected signs and consistent magnitudes. This consistency suggests a possible positive directional effect that could become more apparent with a larger sample size. It is important to highlight that progress was observed in two of the three dimensions of executive functions assessed through neuropsychological tests, which rely on objective measures of student performance rather than subjective self-reports. Because of this, there are reasons to believe that interventions of this sort have the potential to generate objective improvements in critical dimensions associated with violent behaviors such as inhibitory control, impulsivity, and emotional recognition. As there is a close relationship between executive functions and violent behaviors, an improvement in students’' self-control capacity could translate into a significant reduction in such behaviors. The significant improvement in emotional recognition is a key advancement, given that this skill is essential for fostering empathy and emotional awareness. Emotional recognition allows students to better identify and understand the emotions of others, which, in turn, contributes to greater empathy, social understanding, and potentially more positive interpersonal relationships. Similarly, the reported improvement in emotional regulation, measured through questionnaires, indicates that students perceive greater control and management of their emotions after participating in the workshop. This aspect is fundamental, as appropriate emotional regulation is a determining factor in preventing aggressive and violent behaviors. In summary, these results are encouraging, as they suggest that the CBT-based intervention has a positive impact on the development of essential skills related to self-control, empathy, and emotional regulation. These skills are not only linked to a reduction in violent behaviors but also promote a healthier and more constructive social environment. 35 5. Some Lessons from the Intervention, beyond Impacts 5.1 Workshop Attendance Perhaps the biggest difficulty encountered was ineffective communication with schools, which resulted in incomplete interventions or problems collecting information. As pointed out earlier, the intervention consisted of 12 50-minute sessions, with the possibility of four additional sessions if time allowed during the term. The original plan was for all groups in all schools to have time for 14-16 sessions. However, due to various school-related circumstances, such as holidays falling on specific days of the week, school activities leading to class suspensions, and unforeseen school closures, the number of weeks was reduced. Sometimes, for example, schools did not notify if classes were cancelled. Two groups did have sufficient time for additional sessions, but they were not conducted to maintain intervention homogeneity. 5.2 Positive Feedback from Facilitators and Students This large-scale experience of conducting the workshop in several educational institutions simultaneously with more than 900 children allowed us to test the program. One important aspect is the feedback we received from those who took part in the experience. On the one hand, we have the impression of facilitators. First, they all agreed that the workshop was good (60 percent) or very good (40 percent) and felt it had a positive impact on the students as well as having a theoretical and methodological framework, with a structured session design, addressing the topics with diverse activities, and being based on evidence-based therapies. Second, even though the groups consisted of 30 to 50 students, an atmosphere of trust and a safe space was created in all workshops. Facilitators were able to verify that the content and examples resonated with their experience, so that at the end of the implementation, significant changes in the dynamics of the groups were perceived on a general level and, in some cases, it was possible to see how the content was materialized in actions. Third, most of the students were engaged during the sessions and, as the sessions progressed, facilitators noticed changes such as a higher level of understanding in their discourse regarding the importance of paying attention to the present moment, emotional identification/regulation, social issues related to gender, and visualizing the consequences of their actions in the long run. In addition, students who were interested in the workshop mentioned they shared it with their family. Fourth, facilitators agreed that there was good organization and support and that the weekly sessions with the whole team of facilitators were 36 useful to review together the content of each session and to share the difficulties and incidences they encountered. In addition, feedback from students was also quite encouraging. In the last session of the intervention, a link to a satisfaction survey about the workshop was shared with the participants in a Google Form. A total of 383 responses were recorded, and 87 percent of the participants who responded considered that the workshop they had just received was excellent (47 percent) or good (40 percent), while 11 percent thought it was mediocre and 2 percent considered it poor. Ninetythree percent completely agreed (31 percent) or agreed (62 percent) that the workshop met their expectations, and the remaining 7 percent considered that it did not meet their expectations. Eighty-eight percent responded that they were very likely (34 percent) or likely (54 percent) to put into practice what they had learned in the workshop. Ninety percent of those surveyed considered that the workshop activities were fun and engaging. Ninety-six percent found the content of the workshop easy to understand and felt comfortable, and approximately the same percentage considered that they were able to reflect on what they had seen in the workshop through the experiences of their daily lives, and that it helped them. For the participants, the main lessons obtained in the workshop were related to the knowledge, control, and regulation of their emotions, to their ability to realize what is happening to them and to their problem-solving abilities. Finally, 93 percent of those surveyed would recommend the workshop. These responses are positive feedback from students, but because nearly 50 percent of the students did not complete the feedback survey, it is necessary to consider that participation in the survey is endogenous and responses may be biased. 5.3 Lessons and Challenges This first experience in schools has allowed us to understand many important elements in the implementation of the workshop. One is that the intervention can be developed on a large scale and implemented (with appropriate adaptations) in very different contexts throughout the country. This was in part possible because data collection was performed using a platform that could be used on any device such as desktop, tablet or mobile to seamlessly complete each evaluation step was easy and worked very well, and that it has been possible to carry out neuropsychological tests in the field, outside the laboratory environment. 37 We were also able to identify, however, a series of challenges. First, the experience in the schools allowed us to identify some ways to improve the way the workshops were conducted. On the one hand, some activities seem to be too theoretical and became a bit boring for the students, so it might be useful in the future to give sessions a more practical approach, using examples relevant to the students’ social context. In addition, some games or more interactive activities can be added to the sessions to better capture students’ attention. On the other hand, the workshop covered a lot of topics in a very short time. Facilitators felt that there were too many activities planned for one session that could not be carried out due to lack of time. It might be desirable to define which contents should be prioritized and which are less important. The lack of time made it impossible to review and share the content of previous sessions. For all these reasons, some topics should be shortened or more sessions should be added to the intervention. Second, there were problems regarding student participation and managing student activities. Two main problems were mentioned by the facilitators. First, since attendance was mandatory, some students refused to participate in the activities, which had a negative impact on the course atmosphere. Second, having such large groups was detrimental to participation and freedom of expression. In particular, the large number of students in the classroom made the mindfulness exercises difficult. Although having courses of 30 or 50 students facilitates workshops and reduces the cost of conducting them, it seems to be very costly in terms of classroom experience. Perhaps it would be convenient to reduce the number of students in order to have more personalized attention, to motivate those who are not so interested in the program, and to have more enriching discussions. Lastly, there were problems coordinating with schools. We consider that there was a lack of coordination and communication with the schools to mark times and places, as well as a formal presentation of the facilitators with the counsellors in front of the group. This lack of a formal presentation by an educational authority may have diminished the importance of the workshop, and the lack of communication with the school counsellors reduced the support from the academic and administrative staff of the schools. 5.4 Some Ideas for Improvement From the experience studied, there are at least five modifications that can easily be made to improve the intervention: 44 Versions: problems, solutions, and considerations”. Anales de Psicología/Annals of Psychology, 39(2), 304-313. Kazemi, N., Nemati, L., Yadegari, S. L., & Abdous, F. (2023); Predicting social cognition based on self-differentiation and cognitive flexibility in students. Social Psychology Research, 12(48), 103-116. Krosnick, J. A., & D. F. Alwin (1989); “Aging and Susceptibility to Attitude Change,” Journal of Personality and Social Psychology, 57: 416-425. Kurt, A. A., & Gündüz, B. (2020); The investigation of relationship between irrational relationship beliefs, cognitive flexibility and differentiation of self in young adults. Cukurova University Faculty of Education Journal, 49(1), 28-44. Laca, F. A., Alzate, R., Sánchez, M., Verdugo, J. C., & Guzmán, J. (2006); Communication and conflict in young Mexican students: messages and attitudes. Conflict Resolution Quarterly, 24(1), 31-54. Lawrence, K., Campbell, R., & Skuse, D. (2015); Age, gender, and puberty influence the development of facial emotion recognition. Frontiers in psychology, 6, 145042. Leone, R. M., Crane, C. A., Parrott, D. J., & Eckhardt, C. I. (2016); Problematic drinking, impulsivity, and physical IPV perpetration: A dyadic analysis. Psychology of Addictive Behaviors, 30(3), 356. León-Ramírez, B., Ferrando, P.J. (2013); Assessing sexism in a sample of Mexican students: A validity analysis based on the Ambivalent Sexism Inventory. Anuario de Psicología, 43(3), 335-347. Logan, G. D., & W. B. Cowan (1984); “On the ability to inhibit thought and action: A theory of an act of control,” Psychological Review, 91(3), 295–327. Madrona-Bonastre, R., Sanz-Barbero, B., Pérez-Martínez, V., Abiétar, D. G., Sánchez-Martínez, F., Forcadell-Díez, L., ... & Vives-Cases, C. (2023); Sexismo y violencia de pareja en adolescentes. Gaceta Sanitaria, 37. Maldonado, R. C., DiLillo, D., & Hoffman, L. (2015); Can college students use emotion regulation strategies to alter intimate partner aggression-risk behaviors? An examination using I³ theory. Psychology of Violence, 5(1), 46. Manassis, K., P. Wilansky-Traynor, N. Farzan, V. Kleiman, K. Parker, & M. Sanford (2010); “The feelings club: randomized controlled evaluation of school-based CBT for anxious or 45 depressive symptoms,” Depression and anxiety, 27(10), 945–952. https://doi.org/10.1002/da.20724 Mannheim, K. (1952); “The Problem of Generations,” in Kecskemeti, P. (ed.); Essays on the Sociology of Knowledge (Routledge) Martínez-Baquero, L. C., & Vallejo-Medina, P. (2024); Evaluation of ambivalent sexism in Colombia and validation of the ASI and AMI brief scales. Plos one, 19(2), e0297981. Miller, L. D., A. Laye-Gindhu, Y. Liu, J. S. March, D. S. Thordarson, & E. J. Garland (2011); “Evaluation of a preventive intervention for child anxiety in two randomized attentioncontrol school trials,” Behavior research and therapy, 49(5), 315–323. https://doi.org/10.1016/j.brat.2011.02.006 Muñoz, R. G. (2019); “Validación psicométrica de la Escala de Valoración de Riesgo de Violencia en Adolescentes,” Archivos de criminología, seguridad privada y criminalística, (22), 107121. Murphy, C. M., C. I. Eckhardt, J. M. Clifford, A. D. LaMotte, & L. A. Meis (2020); “Individual Versus Group Cognitive-Behavioral Therapy for Partner-Violent Men: A Preliminary Randomized Trial,” Journal of Interpersonal Violence, 35(15–16), 2846–2868. Neilson, E. C., Gulati, N. K., Stappenbeck, C. A., George, W. H., & Davis, K. C. (2023); Emotion Regulation and Intimate Partner Violence Perpetration in Undergraduate Samples: A Review of the Literature. Trauma, Violence, & Abuse, 24(2), 576-596. Ng, Z. J., Huebner, E. S., Maydeu-Olivares, A., & Hills, K. J. (2019); Confirmatory factor analytic structure and measurement invariance of the Emotion Regulation Questionnaire for Children and Adolescents in a longitudinal sample of adolescents. Journal of Psychoeducational Assessment, 37(2), 139-153. Obach, A., M. Sadler, & F. Aguayo (2011); Involucrando Hombres Jóvenes en el fin de la Violencia de Género. Intervención Multipaís con Evaluación de Impacto. Caso Chileno. Santiago de Chile: CulturaSalud/EME. Olatunbosun, I. (2022); “School-Type and Gender Influence on the Efficacy of Cognitive Behavior Therapy on Reducing Bullying Among Secondary School Students in Rivers State,” EAS Journal of Psychology and Behavioral Sciences. 4. 76-82. 10.36349/easjpbs.2022.v04i02.006. 46 O’Leary, K. D., & A. M. Smith-Slep. (2003); “A dyadic longitudinal model of adolescent dating aggression,” Journal of Clinical Child and Adolescent Psychology, 32, 314-327. O’Leary, K. D., Slep, A. M., Avery-Leaf, S., and Cascardi, M. (2008); “Gender differences in dating aggressions and victimization among multiethnic high school students,” Journal of Adolescent Health, 42, 473-479. Overall, N. C., Chang, V. T., Cross, E. J., Low, R. S. T., & Henderson, A. M. E. (2021); Sexist attitudes predict family-based aggression during a COVID-19 lockdown. Journal of Family Psychology, 35(8), 1043–1052. Persson, S., & Dhingra, K. (2021); Moderating Factors in culpability ratings and rape proclivity in stranger and acquaintance rape: Validation of rape vignettes in a community sample. Journal of Interpersonal Violence, 37(13–14), NP11358–NP11385. Pulerwitz, J., G. Barker, M. Segundo, & M. Nascimento (2006); “Promoting more genderequitable norms and behaviors among young men as an HIV/AIDS prevention strategy,” Horizons Final Report. Washington, DC: Population Council. Rey-Anacona, C. A. (2013); “Prevalencia y tipos de maltrato en el noviazgo en adolescentes y adultos jóvenes,” Terapia Psicológica, 31, 143-154. Rivera, F. F., M. N. Pérez, M. D. S. Martínez, & R. A. Fernández (2021); “Diseño y validación de la Escala de Medios de Resolución de Conflictos Interpersonales (MERCI) para adolescentes,” Bordón: Revista de pedagogía, 73(4), 11-26. Rollero, C., Peter, G., & Tartaglia, S. (2014); Psychometric properties of short versions of the ambivalent sexism inventory and ambivalence toward men inventory. TPM. Testing, Psychometrics, Methodology in Applied Psychology, 21(2), 149-159. Romero-Martínez, Á., F. Santirso, M. Lila, et al. (2022); “Cognitive Flexibility and Reaction Time Improvements After Cognitive Training Designed for Men Perpetrators of Intimate Partner Violence: Results of a Pilot Randomized Controlled Trial,” J Fam Viol 37, 461473. Romero‐Martínez, Á., Lila, M., & Moya‐Albiol, L. (2019). The importance of impulsivity and attention switching deficits in perpetrators convicted for intimate partner violence. Aggressive behavior, 45(2), 129-138. Romero-Martínez, Á., M. Lila, M. Martínez, V. Pedrón-Rico, & L. Moya-Albiol (2016); “Improvements in Empathy and Cognitive Flexibility after Court-Mandated Intervention 47 Program in Intimate Partner Violence Perpetrators: The Role of Alcohol Abuse,” Int J Environ Res Public Health, Mar 31;13(4):394. doi: 10.3390/ijerph13040394. PMID: 27043602; PMCID: PMC4847056. Romero-Sánchez, M., Carretero-Dios, H., Megías, J. L., Moya, M., & Ford, T. E. (2017); Sexist humor and rape proclivity: The moderating role of joke teller gender and severity of sexual assault. Violence Against Women, 23(8), 951–972. Romero-Sánchez, M., Megías, J. L., & Carretero-Dios, H. (2021); Sexist humor and sexual aggression against women: When sexist men act according to their own values or social pressures. Journal of Interpersonal Violence, 36(21–22), NP11322–NP11348. Ron-Grajales, A., A, Sanz-Martin, R. D. Castañeda-Torres, et al. (2021); “Effect of Mindfulness Training on Inhibitory Control in Young Offenders,” Mindfulness 12, 1822–1838. Russell, B. L., & Oswald, D. L. (2016); When sexism cuts both ways: Predictors of tolerance of sexual harassment of men. Men and Masculinities, 19(5), 524–544. Sánchez-Hernández, M. D., Herrera, M. C., & Expósito, F. (2020); Controlling behaviors in couple relationships in the digital age: Acceptability of gender violence, sexism, and myths about romantic love. Intervención Psicosocial, 29(2), 67–81. Seo, E., S. J. Koo, Y. J. Kim, J. E. Min, H. Y. Park, M. Bang, ... & S. K. An (2020); “Reading the mind in the eyes test: Relationship with neurocognition and facial emotion recognition in non-clinical youths,” Psychiatry Investigation, 17(8), 835. Shorey, R. C., Brasfield, H., Febres, J., & Stuart, G. L. (2011); The association between impulsivity, trait anger, and the perpetration of intimate partner and general violence among women arrested for domestic violence. Journal of Interpersonal Violence, 26(13), 26812697. Shute, R. y Charlton, K. (2006); Anger or compromise? Adolescents' conflict resolution strategies in relation to gender and type of peer relationship. International Journal of Adolescence and Youth, 13(1-2), 55-69 Slonim, R., & P. Guillen (2010); “Gender Selection Discrimination: Evidence from a Trust Game,” Journal of Economic Behavior & Organization, 76(2): 385-405. Slotter, E. B., & Finkel, E. J. (2011); I³ theory: Instigating, impelling, and inhibiting factors in aggression. 48 Stith, S. M., D. B. Smith, C. E. Penn, D. B. Ward, & D. Tritt (2004); “Intimate Partner Physical Abuse Perpetration and Victimization Risk Factors: A Meta-analytic Review,” Aggression and Violent Behavior, 10(1): 65-98. Straus, M. (2004); “Prevalence of violence against dating partners by male and female university students worldwide,” Violence Against Women, 10, 790-811. Valencia, P. D., & A. De la Rosa-Gómez (2022); “Psychometric analysis of a simplified version of the Emotion Regulation Questionnaire in Mexican adults: The ERQ-CA-9,” Interacciones, e292-e292. Van Starrenburg, M. L., R. C. Kuijpers, M. Kleinjan, G. J. Hutschemaekers, & R. C. Engels (2017); “Effectiveness of a Cognitive Behavioral Therapy-Based Indicated Prevention Program for Children with Elevated Anxiety Levels: a Randomized Controlled Trial,” Prevention science: the official journal of the Society for Prevention Research, 18(1), 31–39. https://doi.org/10.1007/s11121-016-0725-5 Vila-Ballo, A., Cunillera, T., Rostan, C., Hdez-Lafuente, P., Fuentemilla, L., & RodriguezFornells, A. (2015); Neurophysiological correlates of cognitive flexibility and feedback processing in violent juvenile offenders. Brain Research, 1610, 98-109. Welsh, M.C. & B.F. Pennington (1988); “Assessing frontal lobe functioning in children: Views from developmental psychology,” Developmental Neuropsychology, 4:3, 199-230. 49 Annex 1. Violence-related Factors and Technical Description of Measured Variables Violence-related Factors Gender beliefs. Among the factors that are strongly associated with GBV is sexism, or gender beliefs and stereotypes (Bareket & Fiske, 2023; Madrona-Bonastre et al., 2023). This relationship has appeared with different forms of violence: physical, psychological, sexual violence, sexual harassment (Persson & Dhingra, 2021; Romero-Sánchez et al., 2017, 2021; Sánchez-Hernández et al., 2020; Overall et al., 2021; Russell & Oswald, 2016). Moreover, sexism is related to the difficulty of recognition of violence both for victims and perpetrators. Sexism can be related to GBV also indirectly, such as the influence on stereotypes that affect roles that place men in a position of power over women. Nonetheless, there are some studies that demonstrate that this factor only accounts for a small variance of violence (Ibabe et al., 2016). We have chosen the BASI questionnaire to evaluate possible changes in sexist beliefs due to our intervention, because it can be used effectively within the Mexican context (Cortés-Ayala et al., 2015; see also Rollero et al., 2014; Martínez-Baquero & Vallejo-Medina, 2024). It is a shortened version of the ASI questionnaire, which is the most widely used tool for assessing sexism (Bareket & Fiske, 2023; LeónRamírez & Ferrando, 2013). It has accepted theoretical foundation (theory of ambivalent sexism) and a broad range of empirical evidence of its strong cross-cultural validity, with a consistent factor structure, good reliability, and predictive validity (Glick et al., 2000; see also Fiske & North, 2014; Glick et al., 2004). It has also been validated in the Mexican adolescent population. Glick and Fiske’s (Glick & Fiske, 1996) theory of ambivalent sexism considers sexism as a prejudice marked by deep ambivalence and multidimensionality, denoting a mixture of hostile and benevolent attitudes. Hostile Sexism (HS) is traditional sexism oriented toward the perception of inferiority of the other sex. Benevolent Sexism (BS) presents sexist attitudes with a positive affective tone that allows men to behave pro-socially. However, it helps justify and maintain gender inequality (Glick et al., 2000). The BASI questionnaire can assess both forms of sexism. Some studies demonstrates that HS is related to a perception of the world as a competitive place and to a personality disposition high in tough-mindedness, and BS is related to a perception of the world as a dangerous and threatening place and to a personality disposition towards social conformity. 50 Impulsivity. There is a considerable literature documenting a relationship between impulsivity and violence, both for male and female (Fehon et al., 2005; Burgess, 2020; Leone et al., 2016; Shorey et al., 2011; Dodaj et al., 2020). The link between impulsivity and violent behavior may stem from individuals with high impulsivity levels engaging in impulsive actions as a means to alleviate negative emotions, often without considering the repercussions of their actions. Impulsivity is particularly concerning because individuals prone to impulsivity tend to resort to readily available coping mechanisms for immediate relief from distressing situations, disregarding potential long-term negative consequences. Additionally, various studies have indicated that impulsivity and inattentive symptoms may impair fundamental processes such as emotional decoding and setshifting abilities, which are crucial for emotional and behavioral regulation (Romero-Martínez et al., 2019). To assess the potential impact of the workshop on participants’ impulsivity, two instruments have been used: the Emotional Stop Signal Test (ESST) and the State Impulsivity Scale (SIS). The ESST is a neuropsychological task designed to assess the ability to inhibit emotional responses. The SIS is an impulsive behavior assessment instrument validated in the Spanish population. Emotional recognition/Empathy. Studies have suggested that misinterpretation of social cues can lead to inappropriate social responses, such as reacting aggressively or violently to ambiguous social situations (Akhtar & Bradley, 1991; Dodge et al., 2002). Particularly, research indicates that difficulties in interpreting facial expressions of emotion are associated with violent behavior (Hoaken et al., 2007; Babcock et al., 2008). Moreover, deficits in emotional decoding are common in a considerable number of IPV perpetrators (Babcock et al., 2008, Romero-Martinez et al., 2016). There are some examples in the literature that shows that IPV intervention programs for IPV perpetrators can improve emotional decoding abilities (Romero-Martinez et al., 2016). Our program aims to enhance these abilities, and to evaluate the potential impact, we use a specially developed Reading the Mind in the Eyes Test (RMET). Emotional regulation. Deficits in emotion regulation are also an important factor associated with GBV (Neilson et al., 2023). Emotion regulation is multidimensional and includes one’s awareness of emotion (an ability to attend to and acknowledge emotions), understanding and clarity of what emotion one is feeling, and acceptance of emotions (the ability to experience emotions without a 51 secondary negative emotional response to the emotions (Gratz & Roemer, 2004). Emotion regulation also includes one’s ability to act according to one’s goals (e.g., goal-directed behavior), rather than acting impulsively, when experiencing an emotion (Gratz & Roemer, 2004). The three facets of emotion regulation that are most consistently associated with IPV perpetration are impulse control, goal directed behavior, and access to emotion regulation strategies. The ER facet of impulse control refers to one’s ability to control one’s behavior when emotionally distressed. Some studies have shown the possibility of reducing violent behavior through training in emotional regulation techniques (Maldonado et al., 2015; Birkley & Eckhardt, 2019). To measure the potential impact of our intervention on the emotional regulation ability of our participants, we have used the Emotional Regulation Questionnaire ERQ-CA-9 and the Emotional Stop Signal Test (ESST), which is also used to measure inhibitory control capacity. Cognitive flexibility/Solving Problems. Cognitive flexibility refers to the ability to adapt and switch between different cognitive strategies or mental sets in response to changing environmental demands. While cognitive flexibility itself is not directly associated with violence, it can influence various factors that contribute to violent behavior (Vila-Ballo et al., 2015): ● Problem-solving and Conflict Resolution: Individuals with greater cognitive flexibility may be better equipped to navigate and resolve conflicts peacefully, as they can consider multiple perspectives and generate alternative solutions. Conversely, individuals with limited cognitive flexibility may struggle to adapt their thinking and may resort to aggressive or violent responses when faced with challenges or conflicts (Rivera et al., 2021). ● Emotional Regulation: Cognitive flexibility is closely linked to emotional regulation, as it allows individuals to shift attention away from distressing stimuli and adopt more adaptive coping strategies. Poor cognitive flexibility may contribute to difficulties in regulating emotions, increasing the likelihood of impulsive or aggressive behavior in response to emotional arousal (Ghosh & Halder, 2020). ● Attribution Bias: Cognitive inflexibility can lead to rigid thinking patterns and biased interpretations of social cues. For example, individuals with limited cognitive flexibility may be more prone to negative attribution biases, perceiving 52 ambiguous or neutral interactions as threatening or hostile. This can contribute to a heightened sensitivity to perceived threats and a propensity for reactive aggression (Kurt & Gündüz, 2020). ● Social Information Processing: Cognitive flexibility plays a crucial role in processing and interpreting social information, such as facial expressions, gestures, and verbal cues. Difficulties in accurately perceiving and interpreting social cues may lead to miscommunication, social misunderstandings, and interpersonal conflicts, which can escalate into violent behavior in some cases (Kazemi et al., 2023). There are some experiences that have shown an improvement in cognitive flexibility in perpetrators of intimate violence who have participated in recidivism prevention programs (Romero-Martínez et al., 2016, 2022). Our program addresses various aspects of cognitive flexibility, and to assess the potential impact on this factor, two instruments are proposed: a neuropsychological test to evaluate cognitive flexibility (WCST) and a questionnaire to assess conflict resolution style (ICRS). Assessment of risk of violence. Being able to assess the direct impact that the proposed intervention has on reducing violence in the population served is a major challenge. It is very difficult to have a reliable record of violent behaviors before and after the intervention or in control and treatment groups. Schools keep records of incidents among their students, but they lack the necessary systematicity to consider them reliable data for statistical comparison. For this reason, the Adolescent Violence Risk Assessment Scale (SVRAA) has been chosen. Technical Description of the Analytical Instruments: Self-reported Questionnaires Impulsivity (SIS) (Iribarren et al., 2011). This is a psychological assessment tool designed to measure impulsivity as a state or temporary characteristic, rather than a stable trait. It consists of 20 items (Cronbach’s α = .884) that assess impulsive behaviors and tendencies across three subscales: 53 ● Motor Impulsivity or Automatism: Which measures impulsive actions and behaviors, such as acting without thinking, being restless, or having difficulty sitting still. ● Attentional Impulsivity: Which assesses difficulties in maintaining focus and attention, including distractibility and difficulties with concentration. ● Non-planning Impulsivity or Gratification: Which evaluates impulsive decision-making and lack of future planning or consideration of consequences. Scale of Violence Risk Assessment in Adolescents (SVRAA) (Muñoz, 2019). This has proven to be an effective tool to identify subjects who are at risk of exercising violence. It is a scale composed by 25 items (Cronbach’s α = .836), which is grouped into six factors that refer to a negative self-concept, proactive or instrumental violence, lack of self-control, reactive violence, verbal violence and traits of self-harm, that intends to predict the behavior by recognizing the risk of violence (Muñoz, 2019). Interpersonal Conflict Resolution Scale (ICRS) (Rivera et al., 2021). This is a tool designed to assess conflict resolution skills among adolescents. It aims to measure how adolescents handle and resolve interpersonal conflicts in various contexts, such as peer interactions, family dynamics, and romantic relationships. The scale is based on the principles of cognitive-behavioral therapy and focuses on identifying constructive conflict resolution strategies while also recognizing maladaptive patterns. It provides a total score representing overall conflict resolution competency, as well as scores for individual subscales corresponding to three different means for conflict resolution: aggressive means, collaborative means, and passive means. The use of collaborative means is associated with good mental health, secure attachment and with the extraversion, agreeableness, and responsibility personality traits, while aggressive means are linked to socio-emotional deficiencies and low personal adjustment, and passive means are associated with clinical symptoms, avoidant attachment, and social and cognitive incompetence (Rivera et al., 2021). In practical terms, it is a scale composed of 22 items (Cronbach’s α = .810), which is structured around three factors: collaborative, passive and aggressive means for conflict resolution, factors that are related to violent behavior. 60 PAS2 17. Si creo que el punto de vista de la otra persona es muy importante para ella, trato de cumplir sus deseos PAS3 18. Cuando hay situaciones tensas, prefiero no meterme PAS4 19. Si creo que la otra persona sería infeliz si perdiese, la dejo ganar PAS5 20. Doy prioridad a lo que desean los demás PAS6 21. Siempre cedo para no herir los sentimientos de la otra persona PAS7 22. Intento que sus amigos me respalden Puntuación directa/Direct score: interpretación/interpretation Medios agresivos/Aggressive means ≥ 0 y < 7: es improbable el recurso a medios agresivos en el afrontamiento del conflicto/resorting to aggressive means in confronting the conflict is unlikely ≥ 7 y < 14: es probable el recurso a medios agresivos en el afrontamiento del conflicto/resorting to aggressive means in coping with the conflict is likely > 14 y < 21: probabilidad moderadamente alta de recurso a medios agresivos en el afrontamiento del conflicto/ moderately high probability of resorting to aggressive means in coping with the conflict ≥ 21: probabilidad muy elevada de recurso a medios agresivos en el afrontamiento del conflicto/ very high probability of resorting to aggressive means in coping with the conflict Medios colaborativos/collaborative means ≥ 0 y < 8: es improbable el recurso a medios colaborativos en el afrontamiento del conflicto/ resorting to collaborative means in confronting the conflict is unlikely ≥ 8 y < 16: es probable el recurso a medios colaborativos en el afrontamiento del conflicto/ resort to collaborative means is likely to confront the conflict > 16 y < 24: probabilidad moderadamente alta de recurso a medios colaborativos en el afrontamiento del conflicto/moderately high probability of resorting to collaborative means in coping with conflict ≥ 24: probabilidad muy elevada de recurso a medios colaborativos en el afrontamiento del conflicto/very high probability of resorting to collaborative means in coping with the conflict Medios pasivos/passive means ≥ 0 y < 7: es improbable el recurso a medios pasivos en el afrontamiento del conflicto/ resorting to passive means in coping with conflict is unlikely ≥ 7 y < 14: es probable el recurso a medios pasivos en el afrontamiento del conflicto/resorting to passive means in coping with the conflict is likely > 14 y < 21: probabilidad moderadamente alta de recurso a medios pasivos en el afrontamiento del conflicto/moderately high probability of resorting to passive means in coping with conflict ≥ 21: probabilidad muy elevada de recurso a medios pasivos en el afrontamiento del conflicto/ very high probability of resorting to passive means in coping with the conflict STATE IMPULSIVITY SCALE (SIS)/ ESCALA DE IMPULSIVIDAD DE ESTADO 1. Busco actividades en las que obtengo un placer rápido, aunque sé que me pueden ocasionar algún problema o daño físico/I look for activities in which I get quick pleasure, even though I know they may cause me some problems or physical harm 61 2. Suelo caer en tentaciones que me dificultan cumplir con un compromiso/I tend to fall into temptations that make it difficult for me to keep a commitment 3. Busco conseguir beneficios inmediatos, en vez de esperar algo mejor más tarde/ I seek immediate benefits, rather than hoping for something better later 4. Continúo haciendo determinadas actividades placenteras a pesar de que los demás me advierten que me perjudican/I continue doing certain pleasurable activities even though others warn me that they harm me 5. Cuando traigo ganas de hacer algo o se me antoja algo, surge un impulso en mí que quiere hacerlo sin poder esperar/When I feel like doing something or I crave something, an impulse arises in me that wants to do it without being able to wait 6. Obtengo más placer dejándome llevar que controlando mis acciones/I get more pleasure from letting myself go than from controlling my actions 7. Me cuesta controlar mis reacciones cuando no consigo lo que deseo/I find it difficult to control my reactions when I don't get what I want 8. Me cuesta parar de hacer algo, aunque vea que me estoy equivocando/It's hard for me to stop doing something, even if I see that I'm making a mistake 9. Tengo reacciones automáticas que no puedo evitar/I have automatic reactions that I can't avoid 10. Si hago algo y no obtengo los resultados que espero, me cuesta trabajo cambiar de estrategia/ If I do something and I don't get the results I expect, it's hard for me to change my strategy 11. Siempre actúo de la misma forma (soy como soy) aunque no sea el momento o el sitio apropiado/I always act the same way (I am who I am) even if it is not the right time or place 12. No pongo freno a mis reacciones por más que alguien me diga que pare/I don't stop my reactions no matter how much someone tells me to stop 13. Repito muchas veces la misma forma de actuar, aunque no consiga lo que busco/I repeat the same way of acting many times, even if I don't get what I'm looking for 14. Suelo equivocarme porque reacciono tan rápido que no presto suficiente atención a los detalles/I usually make mistakes because I react so quickly that I don't pay enough attention to the details 15. Ante un imprevisto, actúo sin pensar en las consecuencias/When faced with an unforeseen event, I act without thinking about the consequences 16. Me suelo equivocar por andar de apresurado (no esperar el tiempo adecuado)/I usually make mistakes by rushing (not waiting for the right time) 17. Hay ocasiones en las que dejo de prestar atención a las consecuencias de mis actos/There are times when I stop paying attention to the consequences of my actions 18. Respondo antes de que hayan terminado de formular una pregunta/I answer before they have finished asking a question 19. En algunas situaciones, no espero lo suficiente y actúo antes de tiempo/In some situations, I do not wait long enough and act prematurely 20. Actúo sin pensar que otras personas podrían enojarse por lo que hago/I act without thinking that other people might be angry about what I do SCALE: Casi nunca/Almost never = 0 Algunas veces/Sometimes = 1 Bastantes veces/Quite a few times = 2 Casi siempre/Almost always = 3 62 Los primeros 7 ítems del test corresponden a la dimensión de Gratificación Las preguntas 8 a la 13 a la dimensión Automatismo Y las preguntas 14 a 20 conforman la dimensión Atencional/ The first 7 items of the test correspond to the Gratification dimension Questions 8 to 13 to the Automation dimension And questions 14 to 20 make up the Attentional dimension El punto de corte para la puntuación total (la suma de todos los ítems) se sitúa en: > de 19. Puntuaciones iguales o menores a 19 estarían indicando niveles bajos de impulsividad. The cut-off point for the total score (the sum of all items) is located at: > 19. Scores equal to or less than 19 would indicate low levels of impulsiveness. SCALE OF RISK OF VIOLENCE ASSESSMENT SCALE (SRVAA)/ ESCALA DE VALORACIÓN DEL RIESGO DE VIOLENCIA EN ADOLESCENTES (EVRVA) 1. Me he reprochado porque no soy como los demás jóvenes/I have reproached myself because I am not like other young people 2. Me avergüenzo porque siento que mis hermanos son mejores/I am ashamed because I feel that my brothers or sisters are better 3. Sería fácil humillarme por mi baja autoestima/It would be easy to humiliate myself for my low self-esteem 4. Dudo de mis capacidades por la desconfianza de mi familia/I doubt my abilities because of my family's distrust 5. Me he reprimido por mi mal manejo de la frustración/I have repressed myself because of my poor handling of frustration 6. He despreciado los cuidados de mis padres por ser tan estrictos conmigo/I have despised my parents' care for being so strict with me 7. Quisiera irme de casa de mis padres porque no me permiten tomar mis decisiones/I would like to leave my parents' house because they don't allow me to make my own decisions 8. Realizaría bromas pesadas a los demás para pasar el rato porque la escuela me fastidia/I would play practical jokes on others to pass the time because school bothers me 9. Uso de la tecnología para molestar a mis amigos/I use technology to annoy my friends 10. Grafitearía paredes de mi colonia para expresar mi pertenencia hacia ella/I would graffiti the walls of my neighborhood to express my belonging to it 11. Podría lesionar mi cuerpo para no ir a clase/I could injure my body to not go to class 12. Amenazo a los demás cuando estoy molesto/I threaten others when I am upset 13. Destruí pertenencias ajenas por no controlar mis emociones/I destroyed other people's belongings for not controlling my emotions 14. Intimido a los otros porque cambio de humor bruscamente/I bully others because I suddenly change my mood 15. Cometí actos violentos a los que me rodean por la influencia de los medios de comunicación/I committed violent acts to those around me due to the influence of the media 16. Sería capaz de robar a alguien por el bienestar familiar/I would be capable of robbing someone for the sake of family well-being 63 17. Vendería droga ilegal a otros para solucionar los problemas económicos de mi familia/I would sell illegal drugs to others to solve my family's financial problems 18. Me gustaría tener un arma para poder defenderme de los demás/I would like to have a weapon so I can defend myself from others 19. Callo a los demás porque creo tener la razón todo el tiempo/I keep quiet about others because I think I'm right all the time 20. Hice comentarios hirientes a otros por caerme mal/I made hurtful comments to others because I didn't like them 21. Uso malas palabras contra los que me rodean porque me viene de familia/I use bad words against those around me because it runs in my family 22. Sería menos agresivo con mis palabras hacia los demás si mi familia me cuidara mejor/I would be less aggressive with my words toward others if my family took better care of me 23. He sufrido accidentes a propósito para llamar la atención de mis padres/I have had accidents on purpose to get my parents' attention 24. Podría difamarme con tal de conseguir lo que quiero/I could slander myself to get what I want 25. Me castigo físicamente cuando las cosas no suceden como quiero/I physically punish myself when things don't happen the way I want SCALE: 1 = Nunca/Never 2 = Raramente/Rarely 3 = Frecuentemente/Frequently 4 =Siempre/Always Puntuación total/Total score: se suman todos los valores/all values are added Puntuaciones parciales por Factores de Violencia (Suma parcial de cada factor)/Partial scores for Violence Factors (Partial sum of each factor) Autoconcepto negativo/Negative self-concept: Preguntas/Questions 1-7 Violencia proactiva/Proactive violence: Preguntas/Questions 8-11 Carencia de autocontrol/Lack of self-control: Preguntas/Questions 12-15 Violencia reactiva/Reactive violence: Preguntas/Questions 16-18 Violencia verbal/Verbal violence: Preguntas/Questions 19-21 Rasgos de autolesión/Self-harm traits: Preguntas/Questions 22-25 Puntuación mínima/Minimum score: 25 Puntuación máxima/Maximum score: 100 SOCIAL DESIRABILITY INDEX / ÍNDICE DE DESEABILIDAD SOCIAL 1. A veces se me hace difícil ponerme a trabajar sin que me pidan que lo haga (Revertido)/Sometimes it is difficult for me to get to work without being asked to do so (Reversed) ● Sí/Yes ● No/No 64 2. A veces me siento frustrado(a) o triste porque las cosas no salen como yo quiero (Revertido)/Sometimes I feel frustrated or sad because things don't go the way I want (Reversed) ● Sí ● No 3. En algunas ocasiones, he dejado de intentar hacer algo porque he pensado que soy poco capaz de hacerlo (Revertido)/On some occasions, I have stopped trying to do something because I have thought that I am not capable of doing it (Reversed) ● Sí ● No 4. En ocasiones quiero llevarle la contraria a la gente con autoridad, aunque sepa que tienen razón (Revertido)/Sometimes I want to contradict people in authority, even if I know they are right (Reversed) ● Sí ● No 5. Sin importar con quien esté hablando, siempre escucho con atención/No matter who I am talking to, I always listen carefully ● Sí ● No 6. Ha habido ocasiones en que me he aprovechado de alguna persona (Revertido)/There have been times when I have taken advantage of someone (Reversed) ● Sí ● No 7. Siempre estoy dispuesto/a a aceptar cuando cometo un error/I am always willing to accept when I make a mistake ● Sí ● No 8. En ocasiones trato de desquitarme o vengarme en lugar de perdonar u olvidar (Revertido)/Sometimes I try to get even or take revenge instead of forgiving or forgetting (Reverted) ● Sí ● No 9. Siempre soy amable, aun con la gente que no es tan agradable/I am always nice, even to people who are not so nice ● Sí ● No 10. Nunca me molesto cuando la gente tiene ideas que son muy distintas a las mías/I never get upset when people have ideas that are very different from mine ● Sí ● No 11. A veces he sentido muchos celos de la buena suerte de otras personas (Revertido)/I have sometimes felt very jealous of other people's good luck (Reversed) ● Sí ● No 12. Algunas veces me irrito porque la gente me pida favores (Revertido)/Sometimes I get irritated because people ask me for favors (Reversed) 65 ● Sí ● No 13. De forma consciente he dicho cosas que han herido los sentimientos de otra persona (Revertido)/I have knowingly said things that have hurt another person's feelings (Reversed) ● Sí ● No El resultado es la suma de todas las respuestas positivas/The result is the sum of all positive responses. 66 Annex 3. Addressing Spillovers Here we show the differential effect of the treatment between the Pure Control group and the Contaminated Control group for our 20 outcome variables, in Tables A1 and A2. Each table contains one of the two different econometric specifications. The specifications are: (1) 𝑌𝑌 𝑖𝑖=𝛽𝛽0+𝛽𝛽1𝐶𝐶𝐶𝐶𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶𝑇𝑇𝑇𝑇𝑇𝑇𝐶𝐶𝐹𝐹 +𝑢𝑢𝑖𝑖 (2) 𝑌𝑌 𝑖𝑖=𝛿𝛿0+𝛿𝛿1𝐶𝐶𝐶𝐶𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶𝑇𝑇𝑇𝑇𝑇𝑇𝐶𝐶𝐹𝐹 + 𝛿𝛿2𝐶𝐶𝐶𝐶𝑇𝑇𝑇𝑇𝐶𝐶𝐶𝐶𝑇𝑇𝑇𝑇𝑇𝑇𝐶𝐶𝐹𝐹 ∗𝐹𝐹𝑇𝑇𝑇𝑇𝑇𝑇𝐹𝐹𝑇𝑇 +𝛿𝛿3𝐹𝐹𝑇𝑇𝑇𝑇𝑇𝑇𝐹𝐹𝑇𝑇 +𝑠𝑠𝑖𝑖 We see no significant differential effects for 18 of our variables. A priori, we cannot be sure if these results are driven by the lack of power to detect spillovers or the inexistence of spillovers. 67 Table A3.1. Model: Differential Effect of the Treatment between the Pure Control Group and the Contaminated Control (main effect) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) Questionnaires results Neuropsychological tests results Sexist Beliefs (BASI) Impulsivity (SIS - All) Impulsivity (SIS - Attentional) Impulsivity (SIS - Automatism) Impulsivity (SIS - Gratification) Emotional Regulation (ERQ -CA-9 - Cognitive Reappraisal) Emotional Regulation (ERQ -CA-9 - Expressive Suppression) Cognitive flexibility (ICRS - Aggressive means) Cognitive flexibility (ICRS - Collaborative means) Cognitive flexibility (ICRS - Passive means) Assessment of risk of violence (SVRAA - All) Assessment of risk of violence (SVRAA - Negative self-assessment) Assessment of risk of violence (SVRAA - Lack of self-control) Assessment of risk of violence (SVRAA - Selfharm) Assessment of risk of violence (SVRAA - Proactive Violence) Assessment of risk of violence (SVRAA - Reactive Violence) Assessment of risk of violence (SVRAA - Verbal Violence) Impulsivity (ESST) Emotional recognition (RMET) Cognitive flexibility (WCST) Contaminated Control 1.628** -0.197 0.031 3 -0.137 - 0.091 4 -0.132 -0.206 0.107 -1.006* -0.0934 0.0591 0.102 - 0.0678 0.0145 0.112 -0.0691 -0.0318 - 0.00514 -0.0121 -0.0140 (0.702) (0.912 ) (0.376 ) (0.252 ) (0.441 ) (0.326) (0.293) (0.549 ) (0.553) (0.301) (0.812) (0.193) (0.189) (0.180) (0.165) (0.136) (0.134) (0.0092 6) (0.0241) (0.0146) Constant - 3.064** * 0.740 - 0.024 9 0.362 * 0.403 - 1.657** * - 0.966** * 0.333 - 1.504** * - 0.528** * 1.611** * - 0.0084 0 0.338* * 0.233* * 0.420** * 0.405** * 0.223** * - 0.00850 - 0.0554** * 0.587** * (0.344) (0.780 ) (0.285 ) (0.197 ) (0.360 ) (0.224) (0.191) (0.326 ) (0.304) (0.103) (0.285) (0.073 1) (0.112) (0.103) (0.0907 ) (0.0769 ) (0.0437 ) (0.0052 5) (0.00611 ) (0.0092 7) Observations 826 673 673 673 673 763 763 724 724 724 916 916 916 916 916 916 916 814 786 903 Rsquare d 0.005 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.003 0.000 0.000 0.000 0.000 0.000 0.001 0.000 0.000 0.000 0.001 0.003 SE clustered by school-shift *p<.05; **p<.01; ***p<.001 68 Table A3.2. Model 2: Differential Effect of the Treatment between the Pure Control Group and the Contaminated Control (effect by gender) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) Questionnaires results Neuropsychological tests results Sexist Beliefs (BASI) Impulsivity (SIS - All) Impulsivity (SIS - Attentional) Impulsivity (SIS - Automatism) Impulsivity (SIS - Gratification) Emotional Regulation (ERQ -CA-9 - Cognitive Reappraisal) Emotional Regulation (ERQ -CA-9 - Expressive Suppression) Cognitive flexibility (ICRS - Aggressive means) Cognitive flexibility (ICRS - Collaborative means) Cognitive flexibility (ICRS - Passive means) Assessment of risk of violence (SVRAA - All) Assessment of risk of violence (SVRAA - Negative self assessment) Assessment of risk of violence (SVRAA - Lack of self-control) Assessment of risk of violence (SVRAA - Selfharm) Assessment of risk of violence (SVRAA - Proactive Violence) Assessment of risk of violence (SVRAA - Reactive Violence) Assessment of risk of violence (SVRAA - Verbal Violence) Impulsivity (ESST) Emotional recognition (RMET) Cognitive flexibility (WCST) Contaminated Control 1.500 0.250 0.608 0.088 7 - 0.446 -0.837 -0.528 0.357 - 1.689 - 0.416 -0.501 0.289 -0.178 - 0.131 -0.137 - 0.246* - 0.097 3 0.0045 2 -0.0105 - 0.0264 -1.179 - 1.376 (0.49 6) (0.37 0) (0.68 6) (0.576 ) (0.454 ) (0.74 7) (0.96 9) (0.68 2) -1.136 (0.37 2) (0.234 ) (0.21 7) (0.139 ) (0.129 ) (0.24 2) (0.010 3) (0.0288 ) (0.019 6) ContC ontrol*Fe male 0.249 - 0.843 - 1.089 - 0.427 0.673 1.386 0.619 - 0.478 1.295 0.612 1.077 - 0.358 0.212 0.278 0.479* 0.340 0.126 - 0.0184 - 0.00312 0.0240 -1.503 - 2.067 (0.86 1) (0.65 2) (0.80 8) (0.879 ) (0.557 ) (0.81 3) - 1.179 - 1.043 -1.399 (0.46 5) (0.234 ) (0.22 6) (0.265 ) (0.255 ) (0.32 8) (0.014 5) (0.0169 ) (0.021 0) Female - 0.0768 0.841 0.663 * 0.106 0.072 8 -0.113 -0.349 0.059 8 - 0.461 - 0.178 -0.339 0.338 -0.145 - 0.228 -0.199 -0.101 - 0.003 79 - 0.0151 - 0.00112 - 0.0108 -1.194 (0.47 7) (0.32 7) (0.23 4) (0.32 8) (0.326 ) (0.204 ) (0.45 3) (0.35 2) (0.78 4) -1.068 (0.26 4) (0.160 ) (0.19 7) (0.203 ) (0.175 ) (0.27 2) (0.010 6) (0.0093 6) (0.013 9) Constant - 3.024* ** 0.294 - 0.376 * 0.306 ** 0.365 - 1.597* ** - 0.780* ** 0.301 - 1.256 ** - 0.432 1.789* ** - 0.185 0.414* ** 0.352 ** 0.524* ** 0.458* ** 0.225 - 0.0006 47 - 0.0548* ** 0.593* ** (0.921 ) (0.52 6) (0.17 2) (0.10 5) (0.41 1) (0.239 ) (0.188 ) (0.42 6) (0.46 5) (0.36 1) (0.442 ) (0.18 2) (0.105 ) (0.12 7) (0.035 8) (0.059 2) (0.13 9) (0.003 40) (0.0061 9) (0.015 1) Observations 826 673 673 673 673 763 763 724 724 724 916 916 916 916 916 916 916 814 786 903 Rsquare d 0.005 0.001 0.003 0.001 0.002 0.006 0.002 0.001 0.005 0.001 0.001 0.001 0.001 0.002 0.005 0.003 0.001 0.009 0.001 0.005 SE clustered by school-shift *p<.05; **p<.01; ***p<.001 69 Annex 4. Descriptive Analysis of Our Results In this section we compare the distribution of outcomes for the Treatment and Pure Control groups. Table A4.1. Comparing Treatment and Control 76 Table A7.2. Questionnaires Results: Gender Effects Interacted with Social Desirability Index (Model 2) (1) (2) (3) (4) (5) (6) (7) (8) Sexist Beliefs (BASI) Impulsivity (SIS - All) Emotional Regulation (ERQ -CA-9 - Cognitive Reappraisal) Emotional Regulation (ERQ -CA-9 - Expressive Suppression) Cognitive flexibility (ICRS - Aggressive mea ns) Cognitive flexibility (ICRS - Collaborative means) Cognitive flexibility (ICRS - Passive means) Assessment of risk of violence (SVRAA - All) Treatment 6.285 3.868 4.393*** 3.366*** 1.900 0.818 3.675** 5.711** (5.594) (2.572) (1.029) (1.160) (1.782) (1.890) (1.804) (2.835) Female -0.513 0.926* 0.00623 -0.337 -0.166 -0.129 0.0246 -0.609 (1.138) (0.511) (0.357) (0.233) (0.580) (0.488) (0.855) (1.255) SDI 0.381 -1.087*** -0.0373 -0.0294 -0.477*** 0.659*** 0.0955 -0.313 (0.299) (0.234) (0.0852) (0.0713) (0.0920) (0.239) (0.212) (0.245) Treatment*Female -1.103 -1.353 -3.449 -2.558** -1.683 0.394 -3.408 -3.484 (4.770) (3.186) (2.322) (1.294) (2.107) (3.444) (2.376) (3.765) SDI*Female -0.759** 0.595 0.183 0.0570 -0.148 0.234 0.262* -0.471 (0.299) (0.407) (0.160) (0.152) (0.196) (0.301) (0.144) (0.373) Treatment*SDI -0.354 -0.457 -0.449*** -0.393*** -0.213 -0.188 -0.544*** -0.622** (0.603) (0.327) (0.161) (0.142) (0.172) (0.264) (0.173) (0.267) Treatment*Female*SDI -0.0224 0.269 0.487 0.409** 0.237 0.113 0.438** 0.361 (0.540) (0.424) (0.336) (0.164) (0.184) (0.365) (0.214) (0.384) Constant -2.832*** -0.182 -1.623*** -0.778*** 0.103 -0.982* -0.392 1.775*** (0.827) (0.482) (0.277) (0.201) (0.479) (0.520) (0.460) (0.510) N 850 707 792 792 761 761 761 850 In parenthesis, clustered SE at the school-shift level, adjusted by small number of clusters using reg_sandwich command in Stata *if p-value<.1; ** if p-value<.05; *** if p-value<.01