Horn & Cattell (1966) was a landmark study in the advancement of intelligence theory. It prompted Cattell to refine his original gf-gc theory, bringing it closer to Thurstone’s (1938) Primary Mental Abilities. The paper’s EFA methods are a little out of date, so I wanted to see what would happen if modern EFA methods were applied. I also removed the personality variables from the analysis, as they seemed to clutter the results.
A heatmap of the correlation matrix from their Table 3 gives a nice preview of the factor analytic results:
I see 3 big clusters, 2 of which split into 2 smaller clusters.
Parallel analysis
A parallel analysis with n = 297 participants suggests 4 factors, perhaps 5:
Let’s extract up to 5 factors, starting with just a single factor.
1 Factor
Nothing to see here except the familiar finding that the reasoning tests have high g loadings and the speeded tests have lower g loadings.
2 Factors
Does this look like gf-gc theory 1.0? It does to me!
3 Factors
This looks like the higher-order gf-gc-gs factors from Schneider & McGrew (2018).
4 Factors
Here we see that gf splits into Gf and Gv:
5 factors
gs (General Speediness) splits into Gs and Gr
These results are in line with CHC Theory (McGrew, 2005, 2009; Schneider & McGrew, 2012; Schneider & McGrew, 2018), as they should be. Not much has changed! This is a good thing. Consistent results engender trust.
References
Citation
@misc{schneider,
author = {Schneider, W. Joel and Joel Schneider, W.},
title = {Horn \& {Cattell} (1966)},
url = {https://wjschne.github.io/AssessingPsyche/2025-05-18-horn-cattell-1966/horncattell1966.html},
langid = {en}
}