The WJs are designed on a different model of intelligence than the Wechslers (which, of course, were originally designed on no model of intelligence at all, just a pragmatic need of Dr. Wechsler's to distinguish intellectually-disabled patients in his psychiatric population from those who were not). Over the years, the tests have influenced each other, with the WJ exercising the greater influence over the Wechslers, likely due to the advantages of having a theoretical model over not.
The WJs have all been based on various iterations of the CHC (Cattell-Horn-Carroll, named, obviously, after the respective psychologists) model, which de-emphasizes (not removes, as the GIA is still important) the concept of a unitary "g" general intelligence construct, in favor of a number (seven, if I recall, on the WJIII) of more-or-less equally-important clusters. (The equivalent on the WISC is the four factors.)
For those who enjoy additional technical info, this is a brief document comparing the about-to-be-obsolete WJIII with the already-obsolete previous round of Wechslers. It does give you a little bit of an idea of the difference between the models, though.
http://www.riverpub.com/clinical/pdf/WJIII_ASB1.pdfYou'll notice that one of the most handy aspects of the WJ compared to the Wechslers for the gifted population is that it is continuously-scored from age 4 to 90+ (from age 2-4, not all of the subtests are available).
Also, because it has many more cognitive factors on it than the Wechslers do, it has greater potential for identifying internal variability, such as in some of the subtler LDs. And, for the same reason, the factors tend to be separated better on the actual tasks (e.g., not confounding perceptual motor and perceptual reasoning in the same task, as happens often on the WISC block design subtest).
...Just realized that I am about to launch off into even more eye-glazing psychometric minutiae, so I'd better stop!
