Background: Health and demographic Surveillance Sites (HDSS) are long-running geographic cohort studies where detailed data are regularly collected on residents. HDSSs are valuable resources of longitudinal data but their complexity can present barriers for fully utilising them. Objectives: This thesis aims to demonstrate the value of using complex data manipulation and analytic techniques on secondary data from the Karonga HDSS, Malawi. Two key attributes of this data resource are exploited, repeated data points and detailed household and family links, to present analyses related to adolescents: who they live with and their transitions to adulthood. Methods: Complex longitudinal datasets were generated, with variables indicating presence of family. Two data reduction techniques were utilised: latent class analysis, to generate data-driven household composition variables, to compare against ‘standard’ definitions; and sequence analysis, to explore trajectories in transitions to adulthood. These analyses inputted into two longitudinal analyses: an analysis of how presence of family is associated with migration in young people; and an analysis of divorce at a young age and its impact on transition to adulthood. Results: The data reduction techniques added value to HDSS data analyses to improve understanding of complex phenomena: latent class analysis demonstrated the breadth of household types, and sequence analysis showed the diversity in adulthood trajectories, and both techniques enabled useful classification into clusters. They were also useful as exploratory tools, for generating variables and ideas. The resulting analyses enabled detailed understanding of the lives of young Malawians, including key differences by sex, i.e. female adolescents were more likely to migrate than males and at an younger age; and women were more likely than men to have their transition to adulthood disrupted following a divorce if they had children. Conclusions: HDSSs represent rich resources with great potential for complex secondary analyses and greater flexibility for data manipulation compared to other longitudinal surveys. This complexity also means that great care is needed to ensure that appropriate manipulations and techniques are used.