Table 3 .

Overall measures of inter-judge reliability

Session	Apparatus	ICC single	ICC single *	ICC average	λ 1	Armor’s theta	Kendall’s W	p(W)	Kendall’s W *	p(W) *
1	VT	0.83	0.79	0.97	5.21	0.97	0.04	<0.01	0.10	<0.01
	UB	0.84	0.91	0.97	5.23	0.97	0.01	0.72	0.04	0.16
	BB	0.92	0.88	0.99	5.63	0.99	0.03	0.06	0.07	0.02
	FX	0.84	0.83	0.97	5.23	0.97	0.05	<0.01	0.01	0.65
2	VT	0.57	0.91	0.89	4.29	0.92	0.06	0.19	0.08	0.11
	UB	0.84	0.97	0.97	5.35	0.98	0.11	0.03	0.07	0.17
	BB	0.93	0.92	0.99	5.70	0.99	0.11	0.03	0.01	0.78
	FX	0.70	0.89	0.93	4.80	0.95	0.17	<0.01	0.04	0.43
3	VT	0.90	0.30	0.98	5.56	0.98	0.08	0.28	0.05	0.59
	UB	0.87	0.88	0.98	5.43	0.98	0.15	0.32	0.06	0.81
	BB	0.95	0.87	0.99	5.84	0.99	0.24	0.09	0.12	0.43
	FX	0.75	0.47	0.95	4.76	0.95	0.10	0.55	0.10	0.57

^* Presented results are from Universiade 2009 in Belgrade ( Bučar et al., 2012 ).

For ICC single the correlation coefficients below 0.8 are put in bold.

For Kendall’s coefficient of concordance the significant values (expressing bias in the judge panel) are put in bold.

Session 1,2,3: qualifications, all around finals, apparatus finals;

VT: vault; UB: uneven bars; BB: balance beam; FX: floor;

ICC single (average): intra-class correlation for single (average) scores;

λ ₁ : first eigenvalue from the principal component analysis;

p(W): p value of Kendall’s W.