The Pattern of of Adolescent Relationship Abuse and Longitudinal Mental Health: Three-Step Latent Class Analysis

Background: Teen dating violence (TDV) is highly prevalent with an abundance of negative consequences. Few studies have considered TDV victimization and perpetration simultaneously (e.g., Haynie, et al., 2013) despite substantial evidence indicating their co-occurrence. Moreover, most research on TDV is limited by a focus on physical or psychological abuse. The current study extends previous research by examining the co-occurrence of perpetration and victimization including different types of TDV (i.e., physical, psychological, sexual, and relational abuse). We will 1) Identify homogenous classes based on youths’ TDV victimization and perpetration; 2) Investigate predictors of membership in these classes, and 3) examine mental health consequences associated with each class.

Method: 1,042 adolescents in the 9^th or 10^th grade at 7 public high schools in Texas participated in the survey. At Wave 1, the sample was 56% female with a mean age of 15.1. Participants self-identified their racial/ethnic background as Hispanic (31%), White (29%), African American (28%), and other (12%). The conflict in adolescent dating relationships inventory (Wolfe et al., 2001) was used to identify victimized and perpetrated classes together. A three-step Latent class analysis (LCA, Asparouhov & Muthen, 2013) was conducted to predict membership class and to examine health outcomes at wave 2 (one year later).

Results: Four latent TDV classes were identified: Not involved (NI, 43%); emotional abuse victimization and perpetration only (EV, 29.1%); sexual abuse victimization (SA, 15.5%); and emotional and physical abuse victimization and perpetration (EPV, 12.2%).  Demographic information was included to predict the membership of these groups. Compared to youth in the NI class, EV and EPV class were more likely to be composed of females and African American youth. Parent education, as a proxy for SES, was unrelated to class membership. LCA demonstrated that youth in the EV (M_h= 2.24, M_d = 3.22, M_a = 2.17) and SA (M_h= 2.08, M_d=3.16, M_a = 2.17) classes had greater hostility, depression, and anxiety scores than youth in the NI class (M_h= 1.61, M_d=2.63, M_a = 1.70). Interactions between gender and these classes emerged. Specifically, females in the EPV class were more likely to endorse hostility (b =0.75) and depression (b =0.38) compared with males in the EPV class. However, no significant gender differences emerged within other classes.

Conclusion: Consistent with theory and limited empirical research (e.g., Haynie, et al., 2013), adolescent girls who experienced multiple types of dating violence reported increased mental health concerns one year later. Implications will be discussed, including the benefit of identifying homogenous latent classes of TDV for the development of prevention and treatment programs.