Fig. 2

a–d Estimates for changes in depression symptoms scores associated with hypothetical time displacements from one movement behaviour to another. All estimates adjusted for age, sex, race, employment status, education level, marital status, smoking status, alcohol consumption, diet quality score, comorbidity burden score, baseline depression symptoms. Estimates reflect the hypothetical change in depression symptoms scores associated with reallocating time spent in each movement behaviour based on parameter estimates from compositional regression. The difference in minutes/day are modelled around the mean movement behaviour composition (reference). Time is substituted between the movement behaviour on the x-axis and the movement behaviour indicated by the line. For example, Panel B shows estimated scores associated with hypothetically changing the mean amount of time spent in LIPA. As more minutes are added to LIPA, it is estimated that scores will increase if this time is taken from MVPA but decrease if this time is added to sleep. Substitutions were not made beyond the range of ± two SD for the mean of each movement behaviour (e.g., no more than 120-min per day were added to the mean 36-min per day spent in MVPA)