So this has been puzzling me since I saw the first thread posted about it. I've been digging around in technical reports, and this is what I've come up with, in addition to (expanding on) previously-mentioned issues having to do with the minuscule number of members of the norm group at this normative extreme and age bracket:
The internal consistency reliability of the WJIII is sketchy for four-year-olds, and especially for the fluency measures, which are considered to have an adequate floor only beginning at age 7 (reading & math) and 8 (writing). While we're usually more concerned with ceilings for gifted kids, the floor does tell us that the spread at that age level is quite poor. Where the majority of kids score zero, or just above, there is not a good statistical basis for distinguishing the remaining small number of children, beyond "well above average". Consequently, the authors do not report reliability for 4 yo at all, for 9 out of the possible 22 subtests, including for half of the standard battery. The only subtests with good reliability (>.90) at this age are letter-word ID, spelling, passage comprehension, applied problems, word attack, and academic knowledge. (The remaining subtests are fair to poor.) This explains why a good Basic Reading and Brief Reading composite can be obtained, but not Broad Reading, and Math Reasoning, but not Broad Math.
Alfonso, V. C., & Flanagan, D. P. (2002). Comparative features of comprehensive achievement batteries
(Woodcock-Johnson III Assessment Service Bulletin No. 5). Itasca, IL: Riverside Publishing.
from:
http://www.riverpub.com/clinical/pdf/WJIII_ASB5.pdf
