Introduction

Research, for the most part, has mainly focused on a specific type of victimization, such as child maltreatment, exposure to family and community violence, bullying, and sexual violence. However, studies have also found that an experience in one form of violence is likely to contribute to other forms of violence. For instance, some studies have reported that adolescents who are maltreated or frequently exposed to violence in the home have an increased risk of bullying victimization (Nicholson et al., 2018; Zhu et al., 2018). Other studies also found that victims of bullying are likely to be victimized by their dating partners (Debnam et al., 2016; Zych et al., 2021). Some theoretical perspectives have highlighted the linkages among various forms of violence. The social learning theory purports that an individual learns violent behaviors through observation, role modeling, and reinforcement (Baldry, 2003). The violence-breeds-violence hypothesis proposes that individuals exposed to violence during childhood have a high probability of being a victim of violence (Silver et al., 1969).

All too often, children and adolescents are exposed to multiple forms of violence—this phenomenon has been coined polyvictimization (Finkelhor et al., 2011). Over the years, polyvictimization has been explored in the United States, Europe, and China (Chan, 2011, 2013, 2017, Dong et al., 2013; Hu et al., 2018). Comparatively, empirical studies on polyvictimization in South Korea are limited (Emery et al., 2019; Nam et al., 2022; Song & Cho, 2021; Song et al., 2022). However, the prevalence rates of violence against children in South Korea are serious concerns, including maltreatment (17.0% in 2016), exposure to domestic violence (9.9% to 34.4% between 2010 and 2016), verbal bullying (35.6%), ostracism (23.2%), and dating violence (from 16.2% in 2016 to 19.9% in 2017) (Han & Choi, 2021; Korean Women’s Development Institute, 2013; Lee et al., 2021a, 2021b; Ministry of Education, 2019; Nam et al., 2020; Statistics Korea, 2018). Similar to adolescents in other countries, adolescents in South Korea are likely to experience more than one form of violent victimization in various settings, which is detrimental to their health and psychosocial well-being.

As research suggests, polyvictimization is related to greater internalizing and externalizing problems, health-related problems, drug use, and suicidal thoughts and behaviors (Adams et al., 2016; de Oliveira & Jeong, 2021; Feng et al., 2019; Johns et al., 2021; Suarez-Soto et al., 2019; Xavier Hall et al., 2022). However, little is known about whether polyvictimization might be associated with sexual harassment victimization. Defined as unwanted sexual behaviors that reinforce a hostile environment and impede an individual’s academic opportunities (AAUW, 2001; Hill & Kearl, 2011), sexual harassment is a significant concern for children and adolescents in South Korea. Despite this concern, a bulk of research studies have examined adults’ experiences with sexual harassment (Choi et al., 2019; Kim et al., 2017; Kim et al., 2018; Lee et al., 2011; Park et al., 2013; Yoo et al., 2019). Research on sexual harassment experienced by minors in South Korea is scarce (Choi et al., 2017; Jang & Lee, 2013).

As research shows, prior victimization, such as a history of family violence (e.g., parental maltreatment) and bullying victimization, is positively associated with sexual harassment victimization (Fineran & Bolen, 2006; Wei & Chen, 2012). Children who are maltreated at home are likely to develop a maladaptive internal working model of close relations and are likely to be engaged in social contexts that are well-matched with their representation of close relationships (Ellis & Wolfe, 2009). Not surprisingly, studies have shown that maltreated adolescents are prone to affiliate with peers who engage in deviant and delinquent behaviors (Yoon, 2020; Yoon et al., 2020), which would likely expose them to violent victimization (Goemans et al., 2021). Hence, it is likely that adolescents experiencing various forms of victimization have an even higher risk of sexual harassment victimization. However, it is also essential to understand that the mechanisms underlying the linkage between polyvictimization and sexual harassment victimization might be influenced by potential mediating variables. As studies demonstrated, for instance, victims of multiple forms of violence are at an elevated risk of mental health problems and school-related problems (Haahr-Pedersen et al., 2020; Mrug & Windle, 2010; Wright & Wachs, 2021), which might increase their odds of experiencing sexual harassment victimization. Adolescents showing signs of internalizing problems tend to be in a vulnerable state and exhibit social-behavioral deficits (Kochel & Rafferty, 2020), which could predispose them to negative interactions with their peers, including sexual harassment. Research, to date, has not explored whether school-related problems, such as low school satisfaction and dropping out of school, predict sexual harassment in adolescents, even though they are shown to be significantly associated with bullying victimization (Chen et al., 2021; Fu et al., 2013; Lee et al., 2021a, 2021b; Lee et al., 2022). Adolescents who feel disconnected from their school or intend to drop out of school are likely to report feeling lonely and socially isolated (Frostad et al., 2015), putting them in a vulnerable state. These youth have higher odds of being harassed by their peers, who may perceive them as an “easy target.”

The Current Study

There are multiple pathways by which polyvictimization might be positively related to sexual harassment victimization. As previously stated, victims of violence are at risk of developing internalizing problems (Alvarez–Lister et al., 2014; Dierkhising et al., 2019), school disconnect (Chen et al., 2021), and dropping out of school (Lee et al., 2022, 2021a, 2021b). Internalizing problems are also positively related to feeling disconnected from school (Klinck et al., 2020; Loukas et al., 2016), which is a significant predictor of dropping out of school (Melkevik et al., 2016; Orpinas & Raczynski, 2015). Consequently, these pathways likely increase adolescents’ risk of sexual harassment victimization. Moreover, sexual harassment tends to occur offline or online. Although the internet and smartphone are used excessively among South Koreans, including adolescents (Choi et al., 2015), past studies have not separated online forms of sexual harassment from offline forms.

The present study proposes and empirically tests these pathways from a nationally representative sample of adolescents in South Korea. Specifically, we test the following hypotheses: H1: Adolescents who experience polyvictimization have an increased risk of offline and online sexual harassment. H2: Adolescents who are polyvictimized have an increased risk of internalizing problems, decreased school satisfaction, and a high likelihood of dropping out of school. H3: Internalizing problems, decreased school satisfaction, and dropping out of school are significantly and positively associated with offline and online sexual harassment.

Method

Participants

The analytic sample of this study was derived from the Children and Youth Rights Survey (아동 청소년 인권 실태 조사) 2019 in South Korea. Children's and Youth Rights Survey is a periodic survey based on human rights indicators. The National Youth Policy Institute of South Korea developed a human rights indicator system to understand human rights conditions comprehensively. The sample design of the survey is a stratified multi-level colony sampling extraction method considering areal and school classification as stratified factors. The sample consisted of 9,000 students enrolled in elementary school (years 4–6), middle school, and high school across South Korea. The survey used a two-type questionnaire (elementary school, middle school, and high school), a self-administered survey in the presence of a researcher in the classroom. Children and Youth Rights Survey data have been published every year since 2006, approved by the National Statistical Office (No. 402001).

The analytic sample of this study was 6353 adolescents who responded to the sexual harassment questions in middle and high school questionnaires. The mean age for the study sample was 15.6 years (SD = 1.70). Among the study sample, 47.7% were female, and 52.3% were male. The average family economic situation was 4.58 points (SD = 1.05), which means slightly above the average level, taking into account the perfect score on the scale (see Table 1).

Table 1 Descriptive statistics of the study sample (N = 6,353)
Full size table

Measures

Offline and online forms of sexual harassment were each measured with one item. The item for offline sexual harassment was, “(Have you) experienced sexual harassment (teasing) or molestation from your seniors, juniors or friends?” Online sexual harassment was measured as “(Have you) been sexually teased in cyberspace?” Response options for each item are no (0) and yes (1).

Polyvictimization was measured with six items, such as “(Have you) experienced physical punishment from a parent (or guardian)?”, “(Have you) experienced physical punishment from a teacher?”, “(Have you) experienced violence from your seniors or juniors or friends (assault or beating)?”, “(Have you) experienced bullying from your seniors or juniors or friends (be treated as an outcast)?”, “(Have you) experienced abusive language or insults in cyberspace?”, and “(Have you) experienced harm in cyberspace (Do not want to be revealed)?” Response options are Never (0), once or twice a year (1), once or twice in 2 ~ 3 months (2), 1 or 2 times a month (3), and 1 ~ 2 or more times a week (4). Each item was dichotomized based on “once or twice a year” Thus, response options were recoded as never (0) and more than once (1) (Cronbach’s α = 0.54).

Internalizing problems were measured with three items, which included, “(Thinking about your self) have you ever felt anxious for no reason?”, “(Thinking about your self) have you ever been sad or depressed for no reason?”, and “(Thinking about yourself) you don’t have much to be proud of.” Response options ranged from not at all (0) to strongly agree (4) (Cronbach’s α = 0.79).

Low school satisfaction was measured with two items, which were “(Thinking about yourself) you enjoy going to school” and “Overall satisfaction with school life.” Response options were not at all (1) to strongly agree (4). The measurements of items were reverse coded (Spearman–Brown reliability = 0.78).

Dropping out of school was measured with one item: “Have you ever wanted to quit school in the last year?” Response options are no (0) and yes (1).

Covariates for the present study included age (birth year), sex (female [0] and male [1]), and family economic situation (“What do you think the family’s circumstances [economic levels] are?”; very poor [1] to very well [7]).

Analytic Techniques

The analyses for this study included descriptive statistics, correlation analysis, and structural equation modeling (SEM) (N = 6,353). Statistical software programs used were SPSS 26 and Mplus7. SEM, including covariates (age, sex, and family economic situation), was used to examine pathways from polyvictimization to offline and online forms of sexual harassment. The pathways of SEM involve (a) the direct effect of polyvictimization on offline and online forms of sexual harassment (involving the direct effect of the mediators on offline and online sexual harassment) and (b) the indirect effects via internalizing problems, low school satisfaction, and dropping out of school. As the outcome variable is binary, estimation methods for SEM were adopted as robust weighted least squares estimation (WLSM). WLSM algorithm is designed for a categorical observed dependent variable (e.g., binary outcome). Overall goodness-of-fit for SEM was identified by comparative fit index (CFI), root mean square error of approximation (RMSEA), and standardized root means square residual (SRMR). The bootstrapping method was used to examine the statistical significance of the specific indirect effects among variables.

Results

Table 2 displays bivariate correlations among the study variables. Regarding offline sexual harassment, polyvictimization, low school satisfaction, internalizing problems, and dropping out of school were positively correlated. The bivariate correlations between online sexual harassment and each of the major variables in this study were also positive, as in offline sexual harassment. Regarding polyvictimization, low school satisfaction, internalizing problems, and dropping out of school were positively correlated. Except for control variables (e.g., sex, age, and family economic situation), bivariate correlations between all major variables were significantly positive (see Table 2). The correlation coefficients of all variables did not exceed 0.44.

Table 2 Correlations among the variables
Full size table

As a result of estimating SEM for this study, concerning goodness-of-fit for the path model, the estimated fit indices were CFI = 0.936, RMSEA = 0.050 [0.048–0.053], and SRMR = 0.037, which means the study model is acceptable. Regarding the direct effects of offline sexual harassment, polyvictimization (B = 4.220, p < 0.001) and dropping out of school (B = 0.124, p < 0.05) were positively significant. Adolescents who were polyvictimized or dropped out of school were more likely to experience offline sexual harassment. However, the direct effects of low school satisfaction and internalizing problems on offline sexual harassment were not significant in the study model. Regarding the direct effects of online sexual harassment, polyvictimization was positively significant (B = 5.084, p < 0.001). However, the direct effects of low school satisfaction, internalizing problems, and dropping out of school on online sexual harassment were not significant. In terms of dropping out of school, low school satisfaction (B = 0.289, p < 0.001), internalizing problems (B = 0.084, p < 0.001), and polyvictimization (B = 0.330, p < 0.001) were positively significant. Concerning low school satisfaction, internalizing problems (B = 0.410, p < 0.001) and polyvictimization (B = 0.702, p < 0.001) were significant, meaning that higher levels of internalizing problems and polyvictimization were associated with lower school satisfaction. The effect of polyvictimization on internalizing problems in the SEM analysis was significant (B = 2.106, p < 0.001) (see Table 3).

Table 3 Estimates of the path analysis
Full size table

The specific indirect effects of polyvictimization on offline sexual harassment were significantly associated with paths of mediators: (1) polyvictimization was associated with dropping out of school, which in turn, was related to offline sexual harassment (B = 0.041, 95% CI = 0.008–0.091); (2) polyvictimization was associated with internalizing problems, which in turn, was related to dropping out of school, and subsequently offline sexual harassment (B = 0.022, 95% CI = 0.003–0.045); (3) polyvictimization was associated with low school satisfaction, which, in turn, was related to dropping out of school, and subsequently offline sexual harassment (B = 0.025, 95% CI = 0.006–0.053); and (4) polyvictimization was related to offline sexual harassment through the mediating roles of internalizing problems, low school satisfaction, and dropping out of school (B = 0.031, 95% CI = 0.005–0.062). The total indirect effect of polyvictimizations on offline sexual harassment was 0.328 (95 CI = 0.099–0.517) (see Table 4).

Table 4 Direct and indirect effects of polyvictimization on offline sexual harassment
Full size table

However, all specific indirect effects of polyvictimization on online sexual harassment through internalizing problems, low school satisfaction, and dropping out of school were not significant. As shown in Table 3 and Fig. 1, only the direct effect of polyvictimizations on online sexual harassment was significant (B = 5.084, p < 0.001), and the effects of mediators on online sexual harassment were not significant.

Fig. 1
figure 1

Estimates of the path model. Model Fit: CFI = .936, SRMR = .037, RMSEA = .050 [.048–053] The effects of covariates are not displayed in the figure. PV = polyvictimization, Off_SH = offline sexual harassment, On_SH = online sexual harassment, IP = internalizing problems, LSS = low school satisfaction, DOS = dropping out of school.* p < .05, *** p < .001

Full size image

Discussion

Using a nationally representative sample of South Korean adolescents, the present study aimed to investigate the associations between polyvictimization to offline and online sexual harassment and whether the associations might be mediated by internalizing problems, low school satisfaction, and dropping out of school. Our findings suggest that adolescents who were polyvictimized are at an increased risk of offline and online sexual harassment victimization, consistent with prior research findings (e.g., Fineran & Bolen, 2006; Wei & Chen, 2012) and our proposed hypothesis. This finding also supports the violence-breeds-violence hypothesis in which adolescents who experience violence during childhood are likely to be involved in other forms of violence (Silver et al., 1969), such as sexual harassment victimization. Adolescents who have been abused by parents and are targets of peer violence may have an increased tendency to become a victim of sexual harassment. These adolescents normally lack social skills and adaptive coping strategies and are less likely to defend themselves (e.g., Fox and Boulton, 2003; Hansen et al., 2012), making them easy targets.

As hypothesized, our findings indicate that adolescents who experience polyvictimization have an increased risk of internalizing problems, low school satisfaction, and dropping out of school. These findings are consistent with our hypothesis and other findings (Haahr-Pedersen et al., 2020; Mrug & Windle, 2010; Wright & Wachs, 2021). These outcomes have been documented in research on bullying victimization (Cornell et al., 2013; Schoeler et al., 2018; Suresh, 2011). However, these findings are expected as adolescent victims of multiple forms of violence may have an even higher risk of adverse psychosocial outcomes than those victimized within one setting. Also, as shown in prior study findings, internalizing problems were positively associated with low school satisfaction (Klinck et al., 2020; Loukas et al., 2016) and dropping out of school (Melkevik et al., 2016). Internalizing problems were also positively related to dropping out of school. These findings are not surprising, as research literature has long documented how children’s and adolescents’ internalizing problems, such as depression and anxiety, might impact their school-related thoughts and functioning (e.g., Pedersen et al., 2019; Van der Ende et al., 2016).

In addition to polyvictimization, only dropping out of school was significantly related to offline sexual harassment, which partially supported our hypothesis. Youth at risk of dropping out of school may become vulnerable to bullying victimization (Lee et al., 2022, 2021a, 2021b) and other forms of victimization, such as sexual harassment. Given that dropping out of school might potentially mediate the association between polyvictimization and sexual harassment, future research should explore, longitudinally, under what conditions would dropping out of school increases sexual harassment risks. Adolescents exhibiting such behavior might turn to deviant peers, increasing their odds of sexual harassment. Unexpectedly, neither internalizing problems nor low school satisfaction was associated with sexual harassment victimization. It may be that adolescents with internalizing symptoms or those who show low school satisfaction might disengage from their peers, making them less prone to sexual harassment risks.

Although adolescents who are victims of multiple forms of violence are at an increased risk of online sexual harassment, the indirect pathways were not found to be significant. More specifically, internalizing problems, low school satisfaction, and dropping out of school did not mediate the association between polyvictimization and online sexual harassment in our study. Adolescents who experience these psychosocial issues might not run the risk of experiencing online sexual harassment. Perhaps these adolescents are likely to be depressed or feel less motivated, making them feel less inclined to use their smartphones or the internet as they feel disconnected from their peers. Because these adolescents have limited contact with their peers, they may also have a lower risk of online sexual harassment.

Limitations and Implications for Future Research

This is among the few empirical studies on the polyvictimization experiences of adolescents in South Korea. The results suggest that polyvictimization is significantly associated with adverse psychosocial outcomes, which warrants further investigation. However, several limitations need to be noted. First is the cross-sectional study design, which limits our understanding of the relationship between polyvictimization and sexual harassment as causal inferences cannot be established. Also, the items for the polyvictimization in the dataset did not consider the time of occurrence. A longitudinal research design is needed to understand better the causal association between polyvictimization and sexual harassment victimization.

Another limitation is related to the measures for polyvictimization in the study, which was comprised of six items, of which two items were physical punishment from a parent (or a guardian) or teacher, two items were bullying and peer violence, and two items were abusive languages or insults and harm in cyberspace. The study did not include other forms of victimization, such as psychological maltreatment inflicted by teachers and exposure to violence in the community. Furthermore, offline and online sexual harassment variables were each measured with one item.

For polyvictimization, which contained six items, Cronbach’s alpha coefficient was 0.54. The low alpha value can be considered as a limitation and a call for a need to interpret the findings with caution. However, the polyvictimization variable was built upon six yes/no questions. Using only two response options instead of a Likert scale might contribute to lowering the alpha value—leading to a latent variable that could vary from 0 to 6 depending on how many forms of victimization the participants reported. Thus, a high alpha value was not expected because the six items in the scale represented various dimensions of the polyvictimization construct. The assumption behind this was that adolescents can vary from not being victimized at all to being victimized in more than one form of victimization. The scale tapped different aspects of polyvictimizations (i.e., various forms of victimization). While this broadness lowers Cronbach’s alpha (internal consistency), this should not be seen as a weakness but a strength of the scale in content validity (cf., Crutzen & Peters, 2017).

Future studies need to include multi-informants, such as family members, peers, and teachers, which will provide more reliable information on the polyvictimization experiences of adolescents. Additionally, polyvictimization needs to be measured with various types of victimization, such as being physically victimized by siblings or neighbors, sexually victimized by family members or others, maltreatment by family members or teachers, and witnessing domestic violence and community violence.

Furthermore, this study has not considered sexual and gender identity even though female students are more likely to experience sexual harassment than males (DeKeseredy et al., 2019). Also, lesbian, gay, bisexual, transgender, and queer+ (LGBTQ+) youth tend to be at a higher risk of experiencing various types of violence and sexual harassment than heterosexual youth (Sterzing et al., 2019). This is likely the case with female and LGBTQ + youth in South Korea. And finally, although quantitative findings are essential, qualitative research can complement quantitative results by shedding new light on how experiences in various forms of victimization might increase adolescents’ risk of sexual harassment. Therefore, a mixed method research design is warranted, which would increase the validity of the study findings.

Implications for Practice

Crisis intervention has not been well established for adolescents in South Korea who are victims of polyvictimization. There is a serious dearth of evidence-based intervention programs for polyvictimized adolescents in South Korea. These adolescents are likely to develop internalizing problems (e.g., anxiety, depression, and lower self-esteem), be less connected to their school, and are more likely to think of dropping out of school. These adolescents are also at a heightened risk of sexual harassment victimization. Academic achievement and educational success are paramount in South Korea, which will likely be hampered when students’ mental health is impaired.

Early intervention and prevention in preschool, primary, and secondary education could be beneficial, along with an awareness that family violence, emotional dysregulation, unsafe neighborhoods, and mental health disorders can all be early warning signs of present or future polyvictimization (Fry et al., 2012). Practitioners who work with adolescents should consider a multiple-set prevention and intervention system to inhibit the onset and progression of polyvictimization. Practitioners who work with adolescent survivors of polyvictimization need to identify and intervene on behalf of adolescents exposed to multiple forms of violence and may need to be assessed for more victimizations, including broadening what protection from victimization looks like. Examinations of polyvictimization are not comprehensive and are subject to a complicated and emotional process (Finkelhor et al., 2007).

It may be helpful for practitioners to work with teachers and other school administrators to quickly identify adolescents who may be victimized to ensure that the focus is on prevention. In addition to the mental health services polyvictimized adolescents might receive, practitioners might assess and refer adolescents and their families to preventive interventions. Interventions that can address individual, relationship, and community factors that predict perpetration and prevent repeated or additional forms of victimization experiences from occurring are crucial (Turner et al., 2006).

School behavioral specialists, staff, and child welfare workers should pay particular attention when children report any type of sexual victimization, including sexual harassment by peers or any person close to them (Turner et al., 2013b). These events may signal a broader level of vulnerability to victimization and responding adults may need to enhance their focus beyond the typical sexual report to include an assessment of other forms of exposure to victimization. Specifically, practitioners in South Korea should a) provide a counseling program, b) carefully monitor and assess adolescents’ relationships with their peers, and c) consider ways to increase students’ bonding to school, such as increased school activities. Social workers in South Korea might refer students to the Mental Health Support Center (정신건강지원센터), Dream Start, or Youth Counseling (청소년 상담 복지 센터). These programs provide an array of mental health and other needed services.

Conclusion

The insights provided by the findings of our study attest to the importance of continued scientific research concerning how various forms of violence are interrelated and adversely affect adolescents’ psychosocial well-being. The role that specific types of psychosocial well-being play in mediating the relationship between polyvictimization and sexual harassment victimization needs to be carefully tested with a longitudinal study design with more robust measures. This is a necessary first step towards developing programs that are culturally relevant and scientifically validated for adolescents who are victims of multiple forms of violence in South Korea.