Bayesian Multilevel Compositional Data Analysis For Sleep-wake Behaviours and Their Daily Associations with Affect

Multilevel compositional data, such as repeated measures of 24h sleep-wake behaviours in longitudinal studies, are common, yet analytically challenging. This thesis presents an innovative statistical framework for modelling multilevel compositional data using Bayesian statistics, and its software implementation in the R package multilevelcoda. This method was applied in two empirical studies to advance our understanding of the association between the 24h sleep-wake behaviours and affect in daily life.