Skip to main content
. 2014 Sep 29;1:16. doi: 10.3389/fnut.2014.00016

Figure 1.

Figure 1

Model selection simulation results for the parametric bootstrap MLE 2 df selection procedure. True log odds model (Ktrue = 2) plotted (in red) by BMI with results from 20 replicate datasets (in black). Each (A–H) depicts results from data simulated under various conditions including the proportion of events, po (0.10 or 0.33) and sample size, n (500 or 5,000), and evaluated by our methods with two selection criterion settings, α (0.10 or 0.25).