Skip to main content

Advertisement

Fig. 4 | Emerging Themes in Epidemiology

Fig. 4

From: Modelling subject-specific childhood growth using linear mixed-effect models with cubic regression splines

Fig. 4

Subject-specific distributions of the square root MSEs for the entire growth curve for linear splines (black) versus cubic splines (red). Dashed lines correspond to the estimated median, 25 %, and 75 % percentiles of the subject-specific MSE distribution. Cubic regression splines outperformed piecewise linear splines: the median square root subject-specific MSE for linear regression splines was 0.65 vs. 0.51 for cubic regression splines. The Kolmogorov–Smirnov test indicates that the two distributions are significantly different (D = 0.19, p = 0.001)

Back to article page