The Measurement Of Adolescent Health Risky Behavior And Its Latent Structure

Background:During adolescence health-risk behaviors (HRBs) or/and problem behaviors increase the likelihood of negative long-term consequences (e.g., unintended pregnancy and sexually transmitted infections) and predict decreased psychosocial functioning throughout adulthood. Adolescence is one of the important stages of growth, in which adolescent will experience great changes in physical and psychological. Due to brain structure and function of the imbalance in development, adolescence is marked by greater involvement in problem behaviors than either earlier or later developmental periods. Previous studies revealed that various adolescent risky behaviors are correlated and tend to co-occur. In theory, Jessor and colleagues (Jessor&Jessor,1977) used the concept of a "Problem Behavior Syndrome" to explain intercorrelations among behaviors, including smoking, alcohol use, marijuana use, and sexual activity. It was suggested that the relations among the various problem behaviors were due to an underlying construct or latent variable of unconventionality in adolescence. So far, the previous studies did not provide strong evidence for problem behavior theory.Purposes:firstly, based on problem behavior theory, we plan to develop a comprehensive tool, which assessing multiple domains of potential problem behavior domain, as well as with good psychometric properties. Secondly, to explore the underlying structure of adolescent risky behavior, several statistical techniques (i.e., Taxometric and Bifactor model) that designed to detect latent structure were employed in an independed sample. Finally, to detect group heterogeneity of adolescent risk behavior and to provide tailored intervention plan for different potential group, latent class analysis was conducted in a sample of3226students.Research methods:Two adolescent samples were recruited from high schools, the sample size were6633and3226, respectively. The first sample was used to develop and validate the Health-Risk Behavior Inventory for Chinese Adolescents (HBICA), the second sample was used to explore the underlying structure of health-risk behaviors. Several statistical techniques (i.e., Taxometric, Bifactor model and latent class analysis) were employed.Results and conclusions:(1) the current study identified27items and5factors to assess health-risk behaviors:Suicide and Self-Injurious Behavior (SS,5items), Aggression and Violence (AV,8items), Rule Breaking (RB,6items), Substance Use (SU,4items), and Unprotected Sex (US,4items), and the scale exhibited good reliability and validity.(2) Three popular taxometric procedures (i.e., Mean Above Minus Below a Cut, Maximum Eigenvalue and Latent-Mode factor analysis) were employed to decide wheather the latent structure of risk behavior is continuous or categorical variable. According to CCFI (Comparison curve fit index) indices and the shape of fit curves, the results demonstrated that the latent structure of risk behavior is a continuous variable.(3) Various adolescent risky behaviors are correlated and tend to co-occur, convential signle factor model fit the data not well. The one-order multiple correlated factors model and second-order factor model fit the data at accepted level, however, the best fitted model was the Bifactor model.(4) The latent class analysis yielded four latent classes:The first class, high-risk group, which comprises roughly14%of the sample, shows relatively high probabilities of engaging in all type of risky behaviors. The second latent class, low-risk group, those who engage in very few risky behaviors, make up roughly45%of the overall sample. The third class captured about14%of the sample, who engage in very few substance abuse behaviors and high probability of engaging in Suicide and Self-Injurious Behavior. The fourth class, which is labeled as high substance abuse and low other risk behavior youth, makes up roughly28%of the overall sample.