Methods: We collected data through the direct survey with runaway youth living in 53 shelters nationwide in South Korea in 2019 and 316 adolescents participated in this survey. First, we used a latent class analysis (LCA) to identify unobserved classes of poly-victimization using eight binary variables from the juvenile victimization questionnaire (JVQ). Second, we conducted a multinomial logistic regression to compare the identified latent classes of poly victimization by sociodemographic characteristics. Third, we conducted two OLS regression analyses to examine the association between the latent classes of poly-victimization (independent variable) and mental health outcomes (dependent variables: depression and social withdraw). As control variables, age, gender, grade , family structure, type of shelters, household income, and numbers of runaway were included in the OLS regression models. For missing data, the multivariate imputation by chained equation approach was employed before the main analyses conducted.
Results: Based on the evaluation of the fit indices (e.g., AIC and BIC) and substantive criteria, LCA identified three heterogeneous classes of poly-victimization: (1) Relatively Low Level of Poly-victimization (45.57%); (2) Poly-victimization except for Sexual Harassment (32.28%); and (3) Severe Level of Poly-victimization (22.15%). Multinomial logistic regression found that compared to youth in the low-income family, youth in the middle-income family was less likely to be classified as the class of “Poly-Victimization Except Sexual Harassment” relative to the class of “Relatively Low Level of Poly-Victimization” (β=-0.73, p<.05). Compare to males, females (β=1.54, p<.001) were more likely to be classified as the class of “Severe Level of Poly-victimization” relative to the class of Relatively Low Level of Poly-victimization” (β=1.54, p<.001). Compared to youth in two-parent family households, youth in single-parent households (β=-.1.51 p<.001), remarried parent households (β=-1.31, p<.01), and other households (β=-.1.40 p<.05) were less likely to be in the class of “Severe Level of Poly-victimization” relative to the class of “Relatively Low Level of Poly-victimization.” Finally, OLS regression analyses showed that youth in both classes of “Poly-victimization except Sexual Harassment” (β=.22, p<.05) and “Severe Level of Poly-victimization” (β=5.92, p<.001) were more likely to have depression than youth in the class of “Relatively Low Level of Poly-Victimization”. Also, youth in the class of “Severe Level of Poly-victimization” was more likely to report social withdraw (β=1.07, p<.05) than youth in the class of “Relatively Low Level of Poly-victimization.” .
Discussion and Implications: Using the collected data from the youth shelters in South Korea, the current study provided supporting evidence for the relationship between poly-victimization and negative mental health outcomes. Our findings of this study suggest that the need for intervention programs addressing poly-victimized youth with negative mental health outcomes and alternative coping mechanisms.