Developmental Trajectories of Online and Offline Risk Behaviors and Its Predictors: An Analysis By Gender Using Latent Class Growth Anlysis

Background/Purpose: With adolescent risk behaviors emerging as a social problem, various efforts have been made for prevention and intervention. Adolescent risk behavior is exacerbating in quality, however, as it can be characterized as high recidivism, low age, and increased violent crimes. In recent years, adolescents’ activity space has expanded resulting youth behaviors to occur in both online and offline domains. In addition, there are salient gender differences in type, cause, and frequency of adolescent risk behaviors. Risk experiences is not only a maladaptive factor in development and adaptation in adolescence, but also adversely affects transition into adulthood and subsequent adaptation, such that early intervention and follow up are very important. To do so, in-depth and long-term approaches that allow observation of diversity and variability of adolescent risk behaviors is necessary. This study derived latent trajectories of risk behaviors based on risk behavior from online and offline domains. Grounded on social bond theory, we examined whether four unique elements of social bonds predicted membership in the latent risk groups.

Methods: The current study investigated latent developmental trajectories based on gender and on/offline risk behaviors. Furthermore, the study explored whether social bond elements predict latent groups. Waves 2-6 of the Korean Child and Youth Panel Survey data collected by the National Youth Policy Institute (N = 1,483) was analyzed. Latent class growth analysis, which is appropriate for handling longitudinal data to identify latent groups based on gender and on/offline risk behavior development trajectories, was used. Multinomial logistic regression analysis was conducted to examine whether social bond elements predict latent groups, while controlling for parents’ educational background and annual family income.

Results: First, latent class of adolescents showed differences by gender and type of risk behavior. In male adolescents, five groups were derived, showing higher levels and various risk behavior patterns than female adolescents. Female adolescents, on the other hand, were classified into three groups, representing low levels of risk behavior and declining trends. Second, four elements of social bonds were examined as predictors of latent risk groups for each gender. A strong bond with higher parental attachment, commitment, and belief predicted low-risk groups. In other words, lower level of parent attachment, commitment, and belief at baseline (second year of middle school) predicted the probability of belonging to the groups experiencing various risk trajectories. On the other hand, high peer attachment and involvement levels predicted risk groups, such that students with higher levels of peer attachment and involvement at baseline are more likely to be in high-risk trajectory group than low–risk trajectory group.

 

Conclusions and Implications: These diverse latent groups emphasize the importance of an integrated approach that considers both gender and type of risk behavior rather than considering risk experience as uniform in type at a single point of time. Also, we expect information derived from this study to inform the development of a targeted program that reflects the distinct characteristics of each latent risk group and promote effectiveness of intervention.