Skip to main content

Table 4 Preferences of Model Selection Criteria in Simulation Studies

From: Thinking outside the curve, part I: modeling birthweight distribution

True Model Sample size FLIC preferences BIC preferences AIC preferences
   2 3 4 5 6 7 2 3 4 5 6 7 2 3 4 5 6 7
2 5000 25 0 0 0 0 0 25 0 0 0 0 0 7 9 1 3 4 1
  10000 25 0 0 0 0 0 25 0 0 0 0 0 2 12 7 2 2 0
  25000 20 5 0 0 0 0 21 4 0 0 0 0 0 5 4 8 4 4
  50000 5 19 1 0 0 0 5 19 1 0 0 0 0 0 2 6 10 7
  100000 0 20 3 1 1 0 0 20 3 1 1 0 0 0 0 0 6 19
3 5000 19 6 0 0 0 0 19 6 0 0 0 0 1 18 5 1 0 0
  10000 1 24 0 0 0 0 1 24 0 0 0 0 0 16 6 1 2 0
  25000 0 25 0 0 0 0 0 25 0 0 0 0 0 9 4 3 6 3
  50000 0 24 1 0 0 0 0 24 1 0 0 0 0 0 2 1 15 7
  100000 0 17 3 4 1 0 0 17 3 4 1 0 0 0 0 3 7 15
4 5000 18 2 5 0 0 0 20 1 4 0 0 0 0 1 20 1 2 1
  10000 10 1 14 0 0 0 10 1 14 0 0 0 0 1 24 0 0 0
  25000 0 0 25 0 0 0 0 0 25 0 0 0 0 0 24 0 1 0
  50000 0 0 25 0 0 0 0 0 25 0 0 0 0 0 6 14 3 2
  100000 0 0 25 0 0 0 0 0 25 0 0 0 0 0 0 8 10 7
5 5000 21 2 2 0 0 0 22 1 2 0 0 0 0 2 12 9 2 0
  10000 9 3 12 1 0 0 9 3 12 1 0 0 0 0 15 9 1 0
  25000 0 0 25 0 0 0 0 0 25 0 0 0 0 0 8 10 6 1
  50000 0 0 25 0 0 0 0 0 25 0 0 0 0 0 0 10 13 2
  100000 0 0 15 10 0 0 0 0 15 10 0 0 0 0 0 6 10 9
6 5000 24 1 0 0 0 0 24 1 0 0 0 0 0 3 7 10 2 3
  10000 9 2 13 1 0 0 10 2 13 0 0 0 0 0 12 4 8 1
  25000 0 0 23 2 0 0 0 0 23 2 0 0 0 0 0 2 19 4
  50000 0 0 24 0 1 0 0 0 25 0 0 0 0 0 0 0 15 10
  100000 0 0 10 7 8 0 0 0 11 6 8 0 0 0 0 0 10 15
  1. The row with "True Model" = 2 and "Sample size" = 5000 contains the preferences of the three model selection criteria for the number of components, based on 25 samples of size 5000 simulated from a 2-component normal mixture model. Other rows correspond to a different sample size and/or underlying number of components.