Skip to main content

Table 4 AIC/BIC performance when the correct model when the true model effect is in the uterus sub-model

From: Using Bonferroni, BIC and AIC to assess evidence for alternative biological pathways: covariate selection for the multilevel Embryo-Uterus model

 

Sample size

Effect size (βup)

400

800

1600

3200

6400

0

50.0%

49.9%

50.0%

48.9%

48.9%

0.1

52.0%

55.5%

59.9%

66.3%

74.4%

0.2

58.0%

65.3%

73.7%

83.1%

91.3%

0.3

64.7%

73.4%

83.5%

91.2%

97.6%

0.4

70.9%

80.7%

89.8%

96.2%

99.3%

0.5

74.8%

85.2%

94.1%

98.4%

99.9%

0.6

78.5%

89.0%

96.4%

99.4%

100.0%

0.7

81.4%

91.5%

97.6%

99.7%

100.0%

0.8

83.9%

93.5%

98.5%

99.9%

100.0%

0.9

86.1%

94.8%

99.2%

100.0%

100.0%

1

88.1%

95.9%

99.4%

100.0%

100.0%

  1. Proportion of the simulation samples in which use of the AIC/BIC criteria would select the correct model when the true model effect is in the uterus sub-model.