Table 2.

Logistic regression results (N = 56).

Category		Significance of model		Explained distribution	Model’s foodness of fit		Prediction results
		χ2	P	Cox & Snell R Square	−2 Log likelihood	Hosmer and Lemeshow Test (Sig.)	Correctly predicted %
Input all measurements	FFR	77.632	0.000***	0.750	0.000a	1.000	100
	ISO + key lit	32.041	0.031*	0.436	45.591a	0.551	78.6
	ISO	29.147	0.047*	0.406	48.485a	0.400	82.1
Input all PCs	FFR	12.672	0.049*	0.203	64.960a	0.810	67.9
	ISO + key lit	2.767	0.736	0.048	74.866a	0.276	60.7
	ISO	2.792	0.732	0.049	74.841a	0.449	57.1
Input top 5 measurements from PCA†	FFR	14.762	0.011*	0.232	62.870a	0.617	71.4
	ISO + key lit	2.852	0.723	0.050	74.781a	0.323	53.6
	ISO	2.852	0.723	0.050	74.781a	0.066	53.6

^* P < 0.05 ***P < 0.00

^†To make the condition equivalent (5 measurements for each FFR-specific and traditional anthropometry), the FFR-specific face anthropometry’s PC 6 was not included in the analysis.