Identifying latent classes of empowerment among early adolescent girls and the factors that cultivate them: Evidence from a community-based skills training program

Document Type

Article (peer-reviewed)

Publication Date



Background: Efforts to empower adolescent girls remain constrained by confusion around how to conceptually define empowerment and how to design interventions. More research, particularly among young adolescents, is needed that considers contextual and age-specific circumstances. This paper examines the latent classes of empowerment among early adolescent girls in Bangladesh and factors associated with membership in these classes. Using data from the Bangladeshi Association for Life Skills, Income, and Knowledge for Adolescents project, a four-arm randomized controlled trial designed to evaluate different community-based approaches for adolescents' skills development, we identify the latent classes before and after intervention and explore whether girls' membership in the study arms was associated with latent class membership at endline. Methods: To identify underlying subgroups among the sample, we conducted latent class analysis at baseline and endline using indicators of seven different constructs of empowerment based on the literature. Using multinomial logistic regression analysis, we assessed whether respondents' background characteristics and membership in the study arms were predictive of latent class membership at both timepoints. Results: A three-class model at baseline and a four-class model at endline showed optimal fit. In both models, three distinct latent classes were observed: (1) a least empowered class, (2) a mobile, socially active class, and (3) a socially progressive class. At endline, a fourth class emerged—the most empowered—that was mobile, socially active, and socially progressive. Girls' marital status, religion, and parents' level of education predicted latent class membership; household wealth was predictive at baseline, but not endline. Study arm was also associated with class membership at endline: girls in the Gender-Rights Awareness Arm had the highest probability of belonging to the fourth, most empowered class. Conclusion: This paper identifies and characterizes latent classes of empowerment that exist among the sample before and after the intervention. The apparent differential effects of interventions in predicting latent class membership have implications for community-based programming. Findings suggest that skills-building curriculum focused on promoting gender equitable norms, voice, and agency has potential to influence early adolescent girls across multiple dimensions of empowerment.