Abstract:
OBJECTIVE : Methods that are more advanced than conventional mediation analysis are needed in social and behavioral health research to determine causal relationships and evaluate intervention effects and mechanisms. This paper uses longitudinal data to illustrate one such method: full longitudinal mediation model (FLMM) analysis, which is an advanced method for analyzing theory-guided social and behavioral studies examining complex causal relationships. METHOD : Analyses included three waves of data from the Focus on Youth in the Caribbean project. We applied FLMM to test the causal relationship between HIV/AIDS knowledge, condom-use self-efficacy, and condom-use intention. We used Mplus to fit FLMM to the data due to its capacity to handle missing values with a full information maximum likelihood method. RESULTS : : Data fit the FLMM satisfactorily. Results showed that HIV/AIDS knowledge and condom-use self-efficacy formed a self-enhancement system over time: HIV/AIDS knowledge was significantly and causally related to intention to use condoms through a multipath mediation process, and self-efficacy for and intention to use condoms during sex formed a positive feedback loop. CONCLUSIONS : Findings demonstrate several strengths of FLMM and underscore the need for further research of this method in educational intervention trials.
