Fig. 3

Incremental changes in PA composition spent in each domain of PA and predicted HRQOL outcomes. Relationship between incrementally increasing the share of PA time spent in each domain of PA at 10-11y and predicted HRQOL outcomes at: 10-11y (A–C); and 12-13y (D–F). Each data point represents the predicted HRQOL outcomes for a given permutation of activity compositions (in 5% increments). Predictions controlled for sex, age, BMI, pubertal status, socioeconomic position, TUD contextual variables (season and school attendance) and the multiplicative total of PA at 10-11y. Longitudinal models additionally controlled for HRQOL scores at 10-11y. Please carefully note the differing vertical axes between individual charts in this figure (average physical HRQOL scores were several points higher than psychosocial HRQOL scores in this sample which prevented the depiction of discernible trends on a common vertical axis). Asterisks after individual chart headings indicate that the set of ilrs collectively had a statistically significant association with HRQOL outcomes, based on the ANOVA table of model fit (* = p < 0.05)