This is a default analysis, i.e. only the variables were added, no other options were selected.

MTB > PCA 'LUNGES'-'BOUT'.

Principal Component Analysis: LUNGES, BITES, ZIGZAGS, NEST, SPINES, DNEST, BOUT 

Eigenanalysis of the Correlation Matrix

Eigenvalue  2.2881  1.4542  0.9791  0.8861  0.7532  0.4048  0.2344
Proportion   0.327   0.208   0.140   0.127   0.108   0.058   0.033
Cumulative   0.327   0.535   0.674   0.801   0.909   0.967   1.000

Two components have eigen values > 1. Applying this strict rule (only keep components with an eigen value > 1) would mean than only 53.5% of the variability would be retained in two components. The thrid component has an eigen value close to 1 (0.98). Retaining three components would mean that 67.4% of the variation was retained. There is also a problem with the loadings for the components (see below).

Variable     PC1     PC2     PC3     PC4     PC5     PC6     PC7
LUNGES     0.470  -0.310  -0.482  -0.076   0.181   0.055   0.639
BITES      0.459  -0.507  -0.130   0.050   0.084   0.080  -0.706
ZIGZAGS   -0.248  -0.379   0.244  -0.783   0.022   0.346   0.053
NEST      -0.435  -0.400  -0.327  -0.095  -0.114  -0.721  -0.037
SPINES     0.164  -0.465   0.622   0.324  -0.412  -0.116   0.287
DNEST     -0.422  -0.224  -0.399   0.361  -0.391   0.574   0.005
BOUT       0.335   0.275  -0.197  -0.367  -0.790  -0.089  -0.073

None of the loadings of the seven variables on the three main components are small or large, most are in the range 0.3 - 0.6 and hence they do little to explain the data structure. Looking at the loadings for PC1-PC3 is not easy to say which variables are particularly associated with each component.

PCA with 4 components extracted

There is an argument that the analysis should consider foru components. The fourth PC has an eigen value of 0.89, and retaiing it would mean that 80% of the variation could be retained in 4 dimensions. Therefore, the PCA was repeated but with an option to extract only 4 components.

MTB > PCA 'LUNGES'-'BOUT';
SUBC>   NComponents 4.

Eigenanalysis of the Correlation Matrix

Eigenvalue  2.2881  1.4542  0.9791  0.8861  0.7532  0.4048  0.2344
Proportion   0.327   0.208   0.140   0.127   0.108   0.058   0.033
Cumulative   0.327   0.535   0.674   0.801   0.909   0.967   1.000

Variable     PC1     PC2     PC3     PC4
LUNGES     0.470  -0.310  -0.482  -0.076
BITES      0.459  -0.507  -0.130   0.050
ZIGZAGS   -0.248  -0.379   0.244  -0.783
NEST      -0.435  -0.400  -0.327  -0.095
SPINES     0.164  -0.465   0.622   0.324
DNEST     -0.422  -0.224  -0.399   0.361
BOUT       0.335   0.275  -0.197  -0.367

In fact these are the same loadings as the first analysis. Therefore there is little structure. Only the least important component, PC4, has any obvious variable loading patterns. Therefore, the analysis has demonstrated that we can reduce the dimensionality, from seven tofour, but with little insight into the structure of the data. However, this may be because the current axis orientation is sub-optimal for interpretative purposes. We will investigate this in the next analyses when we use factor analysis with rotation.