Will I like Myself If You Hurt Me? Experiences of Violence and Adolescents’ Self-Esteem

: Self-esteem seems to be a crucial factor in adolescents’ psychological well-being. Our study aimed to identify the likelihood of high/medium/low adolescents’ self-esteem in personal experiences of abuse and/or violence (abuse by an adult in childhood, bullying and cyberbullying, and dating violence victimization), different socioeconomic characteristics, perceived social support, and ability to resolve social problems. The study participants were 1451 students from secondary schools (age 13–16) in Spain, Italy, Romania, Portugal, Poland and the UK. We calculated relative risk ratios using multinomial regression models to understand how socioeconomic characteristics, personal experiences of abuse and/or violence, perceived social support, ability to resolve social problems were associated with a self-esteem level. Having no experience of being a victim of physical and sexual abuse in childhood and not being a victim of bullying and cyberbullying is connected with the likelihood of high self-esteem regarding having low self-esteem. Taking as reference those who have never been in a dating relationship, the probability of medium and high self-esteem, decreases when teens are in a romantic or dating relationship and they are victims of intimate partner violence, but the negative effect is explained when other violent variables are included in the model. Being a boy, begin a younger teenager, mother’s paid work, high problem-solving skills and perceptions of social support is in relation with the prospect for higher self-esteem. Promoting adolescent self-esteem means preventing all forms of violence perpetrated by signiﬁcant others, as well as increasing the ability of youth to seek social support and develop problem-solving skills.


Introduction
Self-esteem is a psychological construct described and defined as a person's positive or negative attitude toward himself or herself [1], a positive or negative self-experience [2], or a response to oneself that can take a variety of forms according to different periods and circumstances [3]. Scientific research on adolescents' self-esteem has suggested its positive influence in well-being conditions, particularly in high happiness and life satisfaction and low psychopathology evidence [4]. High self-esteem translates into a greater number of as a serious form of peer violence that is regular and intentional, involves an imbalance of power, and includes the involvement of a perpetrator and a victim [36], support such hypotheses [37].
Adolescence is a period in which most youth engage, for the first time, in intimate relations [38] which means starting dating and building a reciprocal romantic relationship with a partner [39,40]. When dating violence-defined as a behavior that occurs in the context of an intimate relationship, can be manifested in various forms: physical, emotional, and sexual, and can be driven by manipulation, use of power, or control [41]-occurred the consequences are dramatic. It threatening self-perception [42,43] and disturbing selfesteem [44]. If people with more positive evaluations of their intimate relationships report higher levels of self-esteem, negative evaluations tend to correspond to lower levels of self-esteem [45].
Social support has been linked to positive self-esteem [46,47]. Supportive relationships enhance developmental strengths and provide a buffer against risk [48]. Perceived social support and self-esteem are important factors when a relation between childhood maltreatment and well-being is observed [47]. As adolescents' self-esteem is related to their personal competencies, training in social problem-solving strategies may improve self-esteem [49].
The positive youth development model was the basis of the project Lights4Violence, which gave us the opportunity to identify how socioeconomic characteristics; personal experiences of violence in childhood, among peers, and in an intimate relationship; personal and environmental resources; and competencies are associated with adolescent self-esteem. The project Lights4Violence was a longitudinal quasi-experimental educational intervention conducted with European adolescents between ages 13 and 16 and focused on promoting adolescents' capabilities to improve their intimate relationships with their peers [50].
Because self-esteem appears to be a crucial factor in adolescents' psychological wellbeing [51,52] and-according to the model of positive youth development-an important individual force [18], this study aimed: to identify the likelihood of high/medium/low self-esteem in adolescents with different sociodemographic characteristics (students' age, sex, and parents' employment); -to identify the likelihood of high/medium/low self-esteem in adolescents with personal experiences of violence (dating violence, bullying/cyberbullying, physical and sexual abuse in childhood); -to identify the likelihood of high/medium/low self-esteem in adolescents with a different perception of social support and social problem-solving skills.

Design
The study has a cross-sectional design. Data were gathered from adolescents at the baseline stage of their engagement in the Lights4Violence project [50]. The data were collected using an online questionnaire including demographic variables, socioeconomic variables, experience of violence, the Student Social Support Scale [53], the Social Problem-Solving Inventory-Revised Scale [54], the Rosenberg Self-Esteem Scale [1], and other scales defined by the project Lights4Violence. The data were gathered in 12 schools between October 2018 and February 2019. The program content was presented, and the opportunity to participate was offered to the school headmasters. Participation was offered to all the students of the classes selected. Participation was 98.78%.

Ethical Considerations
Data were collected by project partners based at universities in various countries, and all the information gathered by the project partners and beneficiaries was confidential. Participation was voluntary. Each partner was required to obtain the permission of his or her own ethics committees and a signed informed consent document from the school, headmasters, parents, and students. Participants created a unique participant code for themselves at the first data collection point. In cases in which a student reported hav-ing been abused by an adult, each country used its own protocol to inform the school. Due to the anonymity of the questionnaire, it was impossible to identify the victims. However, it was possible to inform the school about the number of student reports of abuse. Each school was responsible for following the respective protocol to intervene.
The Lights4Violence project protocol was approved by the ethics committee of the University of Alicante, University Institute of Maia, Universitatea de Medicina si Farmacie Grigore T. Popa, and Adam Mickiewicz University. Waivers were obtained from the Libera Universita Maria SS. Assunta of Rome and the Cardiff Metropolitan University. It was also registered in ClinicalTrials.gov (accessed on 13 October 2021) by the coordinator (Clinicaltrials.gov: NCT03411564. Unique protocol ID: 776905. Date registered: 18 January 2018).

Participants
We recruited 1555 participants between 13 and 16 years old from secondary schools in Alicante, Spain (130 girls and 125 boys); Rome, Italy (206 girls and 79 boys); Iasi, Romania (214 girls and 129 boys); Matosinhos, Portugal (125 girls and 134 boys); Poznan, Poland (135 girls and 55 boys); and Cardiff, UK (112 girls and 92 boys). The sample was performed within the framework of a quasi-experimental intervention: a secondary education programme to promote positive relationships among adolescents [50]. The involved schools were selected (non-probabilistic sampling) according authors' settings in order to guarantee institutional collaboration in this intervention.
Once we eliminated missing values, the analyzed sample included 1451 questionnaires (879 girls and 572 boys). A statistical power analysis was performed for sample size estimation, based on data from a previously published random-effects meta-analysis of 23 studies about school-based interventions to prevent violence and negative attitudes in teen dating relationships [55].

Dependent Variables
In this study, the predicted variable was subjective self-esteem, collected by the Rosenberg Self-Esteem Scale. The Rosenberg Self-Esteem Scale is a 10-item scale that measures global self-worth by assessing both positive and negative feelings about the self (e.g., "On the whole, I am satisfied with myself"; "I wish I could have more respect for myself"). The scale is believed to be unidimensional. All items are answered using a fourpoint Likert scale format ranging from strongly agree to strongly disagree [1]. We classify self-esteem as low (<25 points), medium (26-29 points), and high (30-40 points) [56]. In our study, The Rosenberg Self-Esteem Scale showed satisfactory internal consistency (Cronbach's alpha) ranged 0.82.

Independent Variables
In this study, we considered a predictor variable related to experiences of abuse and/or violence by an adult in childhood before 15 years old, bullying and cyberbullying, and dating violence victimization.
-Two questions with dichotomous answers (yes/no) were included: "Before you were 15 years old, did any adult-that is, someone 18 years or older-physically hurt you in any way (for example, slapped, kicked, pushed, grabbed, or shoved you)?"; "Before you were 15 years old, did someone 18 years or older force you to participate in any form of sexual activity when you did not want to?". -Bullying and cyberbullying scales were adapted from the Lodz Electronic Aggression Questionnaire (LEAQ). The tool measures bullying and cyberbullying, understood as a serious form of peer violence that is regular and intentional, involves an imbalance of power, and includes the involvement of a perpetrator and a victim, also in the context of current or former romantic partners. The four questions referred to the last three months, and the scale includes Likert answers (never-three times or more) [36].
-Those who had been in a dating relationship were asked: "Has anyone whom you have ever been on a date with physically hurt you in any way (for example, slapped, kicked, pushed, grabbed, or shoved you)?"; "Has the person whom you have been on a date with ever attempted to force or forced you to take part in any form of sexual activity when you did not want it?"; "Has the person whom you have been on a date with ever tried to control your daily activities, for example, who you could talk with, where you could go, how to dress, check your mobile phone, etc.?"; "Has the person whom you have been on a date with ever threatened you or made you feel threatened in any way?" The exposure to dating experiences was measured by a variable created for the data analysis with the following categories: has never been in a partner relationship; has been in a relationship but never experienced violence; has been in a relationship and has experienced violence.

Covariates
The following covariates were also used for this study: -Sociodemographic characteristics: students' age, sex, and parents' employment. The answers were collected through a multiple-choice format. The employment variable was classified as "paid work" and "unpaid work" (homemaker, unemployed, retired and unable to work because of a disability, student, died). - The Child and Adolescent Social Support Scale is a 60-item, multidimensional scale that measures the social support perceived by students from parents, teachers, classmates, friends, and "other people at school" (e.g., principal, counselor), with a range of 12-72 for each area. It includes 12 items in each subscale with six Likert-type response categories that range from never to always. A higher score indicates greater social support. Students rate each behavior on two dimensions: availability (six-point rating scale) and frequency (three-point rating scale) [53]. For this study, we only analyzed the results of the frequency dimension because the trend of both dimensions related to dependent variables and co-variables was similar. In our study, The Child and Adolescent Social Support Scale showed satisfactory internal consistency (Cronbach's alpha) ranged 0.96. - The Social Problem-Solving Inventory-Revised Scale abbreviated version is a brief scale of 25 items that indicates the ability to resolve social problems. Items are answered on a five-point Likert-type scale, from "this is not true" to "extremely true". Items are distributed in five sub-scales (five items in each subscale) that evaluate functional and dysfunctional aspects of the ability to problem solve. The functional dimension is evaluated through two sub-scales: positive problem orientation and rational problem resolution. The dysfunctional dimension is evaluated through the subscales negative problem orientation, avoidance style, and impulsivity/carelessness style. These five dimensions allow to obtain a total score that corresponds to a general estimation of the ability to solve problems, in addition to the average scores in each dimension [54]. Only the summary result was used in this work. In our study The Social Problem-Solving Inventory-Revised Scale showed satisfactory internal consistency (Cronbach's alpha) ranged 0.84.
All the models were adjusted by country (Poland, Portugal, Spain, Italy, Romania, and the UK).

Statistical Analyses
A description of the total sample, and in the sample stratified by sex, was carried out for each of the variables included in the study. For quantitative covariates, we described the mean and the standard distribution. In order to understand which variables were associated with low/medium/high self-esteem, we calculated relative risk ratios (RRR) using multinomial regression models. Statistical significance was considered to be a p-value <0.05. Stata 14.0 was used.

Results
The mean value for self-esteem is 27 Table 1 describes the sample characteristics stratified by low (score < 25), medium (score [26][27][28][29], or high level (score 30-40) of self-esteem. Statistical significance was calculated using chi square for categorical variables and ANOVA test for continuous variables.   Tables 3-5 show the results of the multivariable multinomial regression. In Table 3, Model 1 is adjusted by sociodemographic variables. In Table 4 Model 2 is adjusted by experience of violence. In Table 5  Including the Social Problem-Solving Inventory-Revised Scale and the Child and Adolescent Social Support Scale in Model 3 (Table 5) shows that the relation between physical and sexual abuse in childhood and bullying/cyberbullying and high self-esteem is only slightly explained. Furthermore, as problem-solving skills and perceptions of social support increase, so does the prospect for medium and high self-esteem.

Discussion
In the positive youth development model, it is assumed that self-esteem as a personal resource interacts with other factors, both individual and contextual, shaping health and resilience. In this study, we identified which factors are associated with low/medium/high self-esteem. The likelihood of high self-esteem is related more with situation, when teens have not experienced sexual or physical abuse in childhood and are not victims of bullying/cyberbullying, than those, who had it. Moreover, higher self-esteem is more likely among boys and adolescents whose parents are employed, have higher problem-solving skills, and perceive higher social support. On the other hand, the likelihood of high selfesteem declines among older teens as well as among those who have experienced dating violence.
Our study indicated that self-esteem is lower among girls. Previous studies have suggested that being a boy is associated with higher self-esteem levels [57]. Self-esteem is influenced by gender roles [58]. Research in many countries shows that boys achieve higher self-esteem scores during adolescence than girls. Many features associated with the traditional male role are consistent with high self-esteem. Boys are expected to develop high self-confidence. Showing high self-esteem by girls is assessed negatively as exceeding the gender role [59]. Earlier studies have also shown that girls tend to underestimate themselves, so it is necessary to provide girls with greater support, whereas boys overestimate themselves [60]. Furthermore, girls are more compelled than boys to pursue a certain pattern of beauty, which puts them in continuous confrontation with their self-image [61]. Although some studies found that self-esteem in not affected by the advance of age [51,62], in our research self-esteem decreased with age. It is worth mentioning that many teens in our project were in early adolescence. This stage is characterized by rapid physical changes, significant cognitive maturation, and increased sensitivity to peer approval [63]. Middle adolescence is likely to mark a point of increased depressed mood and depressive syndromes [64], with emotional stabilization taking place after the age of 16. Our research shows that experiencing social support and the ability to solve social problems may eliminate the negative effect of age in adolescents.
The results suggest a link between parental employment status and adolescent selfesteem. Mother's paid work especially increases the likelihood of higher self-esteem. Previous research also has shown that there is a link between parents' employment status and their children's health and self-esteem. Parental long-term unemployment has been negatively associated with adolescents' subjective health [65,66]. Thus, the experiences of a family member are important in shaping the concept of oneself and overall adjustment.
Our research indicates that paid maternal employment may be an important contextual factor in building adolescent self-esteem, the effects of which remain even when problemsolving skills and social support are included in the model.
As has been previously observed [19,[67][68][69][70][71], the likelihood of high self-esteem increases when teens have not experienced sexual or physical abuse in childhood and are not victims of bullying/cyberbullying. In this study, the effect of these experiences on self-esteem remains even after problem-solving skills and social support are included in the model (although the influence of these factors is slightly smaller in Model 3 than in Model 2). Although support [46,47] and social competencies [49] are of great importance for adolescents' mental health and self-esteem, our research reveals the substantial impact of being subjected to physical/sexual violence and bullying/cyberbullying. Actually, people who have not experienced such types of victimization have higher self-esteem levels, regardless of their perceived support and problem-solving skills. Though research supports the efficacy of positive youth development as an approach for changing developmental odds [18], our results show, that in order to help young people who have experienced childhood violence and bullying/cyberbullying, positive strategies based on strengthening resources (e.g., social support and social skills) may not be sufficient and risk reduction strategies will also be necessary. As Benson et al. [18] writes in relation to adolescents from higher-risk groups "It is likely that reducing risks and promoting assets can be complimentary strategies for enhancing positive developmental outcomes".
The negative impact of experiencing childhood violence on self-esteem and mental health is well described in the literature. As a result of trauma, children may lose the sense of security and integrity. Moreover, the misperception of reality, influenced by cognitive mechanisms, could reinforce the feelings of harm and incompleteness and increase re-experiencing difficult emotions. This influences the perceived self-image and self-esteem [72,73]. If the perpetrators are parents, such experiences can have a significant impact on the bond between them and the child. Attachment theory helps how to understand the role of the parent-child relationship and its consequences for the child's self-concept [74,75]. Children who have experienced parental violence are characterized by an insecure attachment style. The strategies that the physically abused and neglected children adapted for relations with parents also marked their interactions outside the family, shaped them and their way of treating themselves and others [76]. Our research confirms that experiencing sexual/physical abuse in childhood could be related with difficulties in building positive self-esteem during adolescence. Building a positive self-image seems to be linked to not having experienced peer violence. According to the positive youth development model, safe peer group interaction fosters building personal resources (such as the self-esteem) [18]. The reasons why the absence of peer abuse experience is so important to self-esteem may be explained also through the sociometer theory. According to this theory, the degree of inclusion or exclusion impacts the level of self-esteem [77]. Self-esteem reflects the state of being accepted or not (e.g., by schoolmates, a peer group, or an intimate partner). The perceived perception of being included or excluded strongly influences self-esteem, particularly during adolescence [77].
Our results also show that compared to people who have never been in a dating relationship, the likelihood of medium and high self-esteem decreases when teens are in a romantic or dating relationship and are exposed to intimate partner violence ( Table 2), but the negative effect is explained when other violent variables are included in the model (Table 4, Model 2). People who have experienced sexual abuse on dates, as well as those who were afraid of their dating partner, are less likely to achieve average and high selfesteem than non-dating teens. Young people who have experienced physical abuse and those whose daily activities were controlled by their intimate dating partner are less likely to achieve high self-esteem than teenagers who have never dated. Our research is consistent with other research indicating the negative impact of dating violence on teenagers' selfesteem [42,43,68,78]. According to the positive development model, the sense of supportive relationships, characterizing themselves by warmth, closeness, good communication, se-cure attachment, and responsiveness support the possibility of growth and happiness [18]. Such conditions are not met by a violent relationship. It is not a supportive relationship, on the contrary, it has a negative impact on the development of a young person, including self-esteem level.
Self-esteem acts as a monitor, searching for cues to determine whether the individual is being accepted or rejected in a relationship [77]. Experiencing abusive behaviors in dating relationships is used to evaluate yourself as a partner [78]. Moreover, when referring to the attachment theory, it is worth emphasizing that the quality of the romantic relationship in adolescence may also be related to the quality of the bond with parents in childhood.
As the results of studies show, aggression in adolescent romantic relationships may have roots in non-secure attachment [79], so it is worth assessing in the context of broader patterns of behavior via the attachment system. Therefore, it can be assumed that in our study, when other variables related to violence were added to the model, e.g., experiencing violence in childhood, the negative effect of dating violence was explained.
Also, previous research has shown that individuals with high self-esteem may also seek and receive more social support, which may facilitate more adaptive coping behaviors and overall adjustment [46,57]. Individuals with low self-esteem may experience having no comfort in themselves and their capabilities and be more challenged regarding the ability to identify coping resources and to use them for intended purposes [52]. A lack of self-confidence can also make it difficult to use effective problem-solving methods [49]. According to the positive young development model, social support and social skills such as solving social problems are important developmental assets [18]. Our research results indicate that people with higher problem-solving skills and perceived social support have greater likelihood of medium and high self-esteem. The findings of this study support the validity of the use of preventive strategies aimed at shaping social skills, such as problem-solving and seeking social support to strengthen adolescents' self-esteem.
In interpreting our results, it is necessary to consider some limitations. The sampling procedure does not allow us to generalize the study results to the population of each country. It was calculated to have enough statistical power to analyze the results as a whole. It should also be mentioned that some information related to sociodemographic characteristics was missing, because the adolescents declined to provide that information. The information lost was related to parents' education level, because they did not know this information about them, so a high percentage of the answers were marked as "don't know". The perception of having been exposed to dating violence could be different depending on the cultural context of the students. To address this, our models were adjusted by country, but there may be residual confusion.

Conclusions
Adolescents who have not been exposed to sexual or physical abuse in childhood or have not been bullied or cyberbullied reveal high levels of self-esteem when compared with those who were victims. Considering this relation, the negative and pervasive effects of victimization on one's self-perception development are undoubtedly demonstrated. In addition, the likelihood of high self-esteem increases when adolescents are boys, young adolescents, perceive high social support, and have high problem-solving skills. Considering the main results of the study, it is reasonable to deduce that promoting adolescents' self-esteem implies preventing and combating all forms of violence perpetrated by significant others, along with increasing their ability to seek social support and develop problem-solving skills. Special support should be given to girls because they have lower self-esteem than boys. As sexual or physical abuse in childhood and bullying or cyberbullying reduce the chances of higher self-esteem levels in adolescence, even when social support is relatively high and problem-solving skills are achieved, all efforts to protect children from victimization are required. However for adolescent victims of childhood violence, resource-enhancing strategies must be complemented with those that reduce the risk.