Table 5. Between group PCA classification test to assess whether one scan device dataset performs better at identifying sexes based on shape.

This analysis averages shape among replicates, computes a between-group PCA separately for 3D and µCT datasets, and runs a cross-validation classification test. The results indicate whether one type of scan dataset is more successful at classifying males versus females based on the shape variation present in the dataset. It also returns a kappa statistic; a kappa value over 0.20 indicates “fair” agreement between the two datasets. Shape variation visualized by sex can be seen in Fig. 7.

Cross-validated classification results in frequencies
3D	f	m
f (n = 7)	5	2
m (n = 4)	2	2
CT	f	m
f (n = 7)	5	2
m (n = 4)	2	2
Cross-validated classification results in %
3D	f	m
f	71.4	28.6
m	50.0	50.0
CT	f	m
f	71.4	28.6
m	50.0	50.0
Overall classification accuracy (%)
3D	63.6
CT	63.6
Kappa statistic
3D	0.214
CT	0.214