Explore the chart
Where every adapter sits, what defines each factor, and how far the factors match the Big Five.
what is this?
Two related objects appear on this page and are easy to confuse. A coordinate is where an adapter sits in the five-factor space, an inner product with an orthonormal basis of the factors' span. A loading is how strongly a trait belongs to a factor in the factor solution. The scatter uses coordinates; the bar charts use loadings.
The 141 adapters
Pick any three factors, or principal components, as axes. Colour is the factor a trait loads on most; switch to Big Five keying to compare.
The rotating chart needs JavaScript. Every number behind it is in the downloads.
What defines each factor
The traits that load most strongly on each factor, at both poles.
The elbow plots
Eigenvalues of the similarity matrix (left) and of its factor-analytic version, communalities on the diagonal (right).
notes
How many factors to keep. Each panel shows the eigenvalues of the adapter similarity matrix in decreasing order (log scale); a sharp bend, the elbow, suggests where structure gives way to noise. The dashed lines are what random data of the same shape would give at two assumed sample sizes. Parallel analysis keeps a factor only while the real eigenvalue beats the random one.
Panel a is the principal-components view (total variance). Panel b is the factor-analytic view, with each trait's shared variance on the diagonal. The answer depends on the sample size you assume, which is why five is a choice; the post explains why we made it.
Neighbourhoods
A trait's nearest and farthest adapters at stage one, stage two and in the merged persona.