Statement of purpose Very few studies have examined dating violence over time, with even less known about psychological abuse. We examine the development of psychological abuse perpetration through 4 years of adolescence; influences on growth factors of time-invariant predictors; and change of psychological abuse due to time-varying effects of hostility, controlling for time-invariant predictors.

Methods/approach 958 ethnically diverse adolescents from multiple schools participated (57% female, mean age = 16.1). The CADRI measured past-year psychological abuse perpetration. Gender, ethnicity, family composition, and parental education served as time-invariant variables to predict initial abuse and growth rate. Hostility (SCL-90) at each wave served as time-varying variable.

Results Unconditional liner model fit was acceptable, x2(df = 5) = 11.47, p < 0.05, RMSEA = 0.04, CFI = 0.99. The mean of intercept (est = 3.23) was positive and significant, indicating that average of psychological abuse at wave2 was significantly larger than zero. The mean slope (est = -.11) was negative and significant, indicating that perpetration of psychological abuse linearly decreases over time on average. In addition, the significant variance of the intercept (est = 4.69) and slope (est = 0.45) indicated inter-individual variability around this mean group and in growth over time. Conditional growth model including time-invariant and time-varying variables also evidenced an acceptable fit, x2(df = 29) = 54.19, p < 0.05, RMSEA = 0.03, CFI = 0.98. Adolescents with more hostility reported significantly more psychological abuse perpetration than would be expected given their trajectory alone. Female and African American adolescents reported more psychological abuse relative to their respective counterparts. None of the lifetime-invariant variables significantly predicted growth rate.

Conclusions Overall, the developmental trend of psychological abuse perpetration decreased over time. Gender and ethnicity were associated with initial psychological abuse but not significantly associated with developmental trend. Hostility at each wave was associated with time-specific increases in psychological abuse perpetration.

Significance and contributions Hostility can play a role as risk factor, actively delaying the normative decrease of psychological abuse perpetration over adolescence.