A Data Science Central Community
I've been trying to use factor analysis. I started with binary variables hence computed tetrachoric correlations - used this matrix as input for factor analysis - used varimax rotation. I find that some of my rotated factor loadings > 1! How do I interpret this? Thanks!
Factor loadings are a representation of the variation that is explained in your data. Your initial data has patterns and noise. When you create factor loadings the noise is still there but its explained now by the factor loadings. The first factor or principal component explains most of the noise. The next explains the 2nd most and so on. The neat thing about factor loadings is that each factor loading or principal component has zero correlation. So you can use them as a "ranking" of explained variation in your data.
Now the values of factor loadings can have different meanings that are much harder to interpret. It might be best to compare the new factor scores with the raw data.