Estimated effects and 95% bootstrap confidence intervals for boosting quantile regression models for the 35% Z-score quantile (columns in grey) and the 15% Z-score quantile; see Figures 2 and 3 for detailed results of continuous covariates. 1,2

1Significant effects are shown in bold. An effect of a categorical covariate is rated as significant if the corresponding 95% bootstrap confidence interval does not contain zero. An effect of a continuous covariate is rated as significant if the effects from all 100 bootstrap samples are estimated to be below/above zero for at least one interval within the covariate range.2The effects of categorical covariates can be interpreted as their effect on the respective Z-score quantile relative to the reference category. For example, the 35% quantile of the Z-score for girls is significantly increased by 0.166 compared to the 35% quantile of boys, given all other covariates are similar.