V1 3-FAC | V2 3-FAC | V12 average 3-FAC | V12 average 4-FAC |
---|
Rotation SS Loadings | Rotation SS Loadings | Rotation SS Loadings | Rotation SS Loadings |
EV | % variance | Cum %var | EV | % variance | Cum %var | EV | % variance | Cum %var | EV | % variance | Cum %var |
2.818 | 23.484 | 23.484 | 3.152 | 26.267 | 26.267 | 3.008 | 25.065 | 25.065 | 2.626 | 21.884 | 21.884 |
2.466 | 20.550 | 44.033 | 2.436 | 20.297 | 46.563 | 2.520 | 20.998 | 46.064 | 2.167 | 18.060 | 39.943 |
2.282 | 19.021 | 63.054 | 2.055 | 17.126 | 63.689 | 2.182 | 18.186 | 64.249 | 2.083 | 17.362 | 57.305 |
| | | | | | | | | 1.744 | 14.531 | 71.836 |
- EV = post-rotation eigenvalues; these are typically more evenly spread in their values than before rotation, Cum %var. = cumulative percentage of the variance explained by factor scores, ie by adding the current absolute percent variance explained to the previous
- The first two fields show straightforward 3-factor solutions on the data from each visit separately but with scale values based on Visit 1 and 2 averaged data combined, making the scaling identical across all 4 fields although the data sources differ. In the last two fields, the item data themselves are averaged for the two visits, so both the 3- and 4-factor solutions proceed on this same averaged Visit 1 & 2 data. The main text, supported by Additional file 2 b explains why visit-averaged data and then the 3-factor solution are preferred