Skip to main content
. 2021 Aug 20;9:e11804. doi: 10.7717/peerj.11804

Table 2. Nested Procrustes ANOVA on shape for device comparison.

We applied the following nested hierarchical structure: Specimen > Device > Operator > Landmark replica. (A) All datasets. (B) Only scanner-based datasets. (C) Only digitizer-based datasets. The R-squared values (Rsq) give an estimate of the relative contribution of each factor to the total shape variation. Repeatabilities are for landmark replica only.

Variables Df SS MS Rsq F Z Pr(>F) R
A. Device Comparison Dataset
Specimen 21 0.602 0.030 0.924 272.705 21.092 0.001
Specimen:Device 22 0.016 0.001 0.025 6.916 26.363 0.001
Specimen:Device:Operator 44 0.025 0.001 0.037 5.260 27.812 0.001
Residuals (Landmark replica) 88 0.010 0.000 0.014
Total 175 0.671
B. Device Comparison Dataset: Scanner
Specimen 21 0.316 0.015 0.954 114.059 17.562 0.001 0.983
Specimen:Operator 22 0.009 0.000 0.028 3.230 24.287 0.001
Residuals (Landmark replica) 44 0.006 0.000 0.018
Total 87 0.332
C. Device Comparison Dataset: Digitizer
Specimen 21 0.315 0.015 0.942 177.721 19.191 0.001 0.989
Specimen:Operator 22 0.016 0.001 0.047 8.441 25.747 0.001
Residuals (Landmark replica) 44 0.004 0.000 0.011
Total 87 0.334