Wednesday, November 09, 2005
CHC characteristics of neuropsychological tests
An interesting article by Schaie et al. (2005) in Aging, Neuropsychology & Cognition (click here to view and/or download...warning...it is close to 3 MB and might open/download slow if you are on a dial-up).
Of interest to quantoids who read this blog is the use of extension analysis to "project"the neuropsych measures into the factor space of well-known CHC factors (viz., factors defined by Thurstone's Primary Ability Tests). The figure posted in this note (double clicking on the figure will enlarge it for easier viewing) displays the factor loadings of the neuropsych measures (not the PAM/CHC measures used to define the factors) on the factors, which I have taken the liberty to label as per CHC terminology. Individuals using any of these neuropsych measures might find this information useful as traditionally neuropsych measures are typically not developed to measure pure latent constructs.
As would be expected, the authors concluded (and this is only one part of the purpose of the study) that most neuropsych measures are factorially complex...they measure more than one PAM/CHC ability. Of course, in their defense, the purpose of most neuropsych measures is not to measure pure latent constructs, but instead, is to help detect and describe brain-behavior relations.