Skip to main content
. 2014 Jul 7;5:116. doi: 10.3389/fneur.2014.00116

Figure 12.

Figure 12

Construct validity of the IBB Score. Principal component analysis (PCA) extracted three orthogonal multivariable principal component (PC) clusters that together accounted for 81.4% of the variance in outcome after SCI. (A) PC1, the largest cluster of variance (51.6%) reflects the relationship between forelimb function and histological outcome. Note the IBB score is the highest loading variable on PC1, providing evidence of construct validity. (B) PC2 (18.3% variance) reflected the relationship of forelimb weight support and gait. (C) PC3 (11.5% variance) reflects forelimb stride length. (D) PCA extracts the PCs through eigenvalue decomposition of the bivariate correlation matrix of all outcomes, here represented as a heat map of Pearson values. PCs are reflected as the Venn intersection (gray) across outcome domains and the PC loading values (correlation between each variable and the PC cluster) are represented as arrows where gage represents loading magnitude and heat reflects direction (red positive relationship, blue inverse relationship).