Skip to main content

Advertisement

Table 4 Comparing the Akaike information criterion and Bayesian information criterion for linear and cubic spline models using OLS (fixed effects only), LME (random slope random intercept) and LME with CAR(1) errors

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

  Linear splines Cubic splines
Akaike information criterion
3 knots (3, 10, 29) 5 knots (3, 6, 18, 24) 3 knots (3, 10, 29) 5 knots (3, 6, 18, 24, 40)
Ordinary least squares 52,495.44 52,472.74 52,399.32 52,397.75
Random effects 28,608.28 28,345.14 27,560.38 27,541.87
Random effects and CAR(1) 19,719.72 19,495.80 19,222.76 19,235.37
Bayesian information criterion
 Ordinary least squares 52,553.77 52,545.24 52,472.23 52,485.24
 Random effects 28,688.47 28,439.91 27,655.15 27,651.22
 Random effects and CAR(1) 19,807.21 19,597.86 19,329.66 19,352.00