I don't use the DKEFS, mostly because it is clunky to administer, but I like some of the subtests. Norms aren't a huge concern to me. I mean, obviously, it is important to keep track of how they are derived and holes in them. But, I think functional data is more important. For example, I'll include tests like inhibition of saccadic eye movement or testing for myerson's sign. I use clock drawing, but I don't score it. I just look for disorganization or neglect and describe it qualitatively, similar approach with the rey complex figure. I generally strive to have a psychometrically solid battery, but I supplement it with functional exploration. You'll find people who try to make psychometric arguments about comparing normative score differences in say BVMT (visual memory) versus HVLT (verbal) as being advisable because of the sampling issues, but they aren't really functionally equivalent. I think we have to be careful with more subtle interpretations of functional differences and it must be done so in the context of the the patient's history. It's all about pattern clusters to me. How do things hold together? E.g., if I have someone making source memory errors, doing terribly on list b (CVLT-2), perseverating on the WCST, and with report of abulia/personality changes, I'm thinking strongly about fronto-subcortical dysfunction (provided they do okay on other tests). An isolated impaired performance statistically isn't going to get me to worked up unless it is really loud and then I am going to try to triangulate and replicate it somehow with a different instrument.
