Welcome!

Some context: different tests are not expected to generate identical scores, both because they are different tests with different specific items on them, and because of the concept of regression to the mean, which finds that if the first score is far off the mean, the second measure is likely to be closer to the population mean (in somewhat oversimplified terms).

Also, score stability in very young children, such as your DC, is not particularly high--not nearly as high as a few years later, closer to eight or nine years old. This is for many reasons, including the wide range of normal variation in when young children make certain developmental leaps, and the challenges (which you witnessed) of obtaining optimal performance from them, for perfectly normal developmental reasons.

To your specific questions:
1. Your expectation from the WPPSI is not exactly incorrect. 91 is not really that far away from 99, if you include confidence intervals. This is where regression to the mean and differences in tests come in.

2. The FSIQ is not derived from a simple average of the subtest scores. It is much more unusual to have multiple areas quite high than to have only one. Also, not all of the subtests are included in the FSIQ. Only the two VC, two FR, and one each from VS, WM and PS are included. As it happens, the two scores not used are his two lowest subtests (including ZL, which you questioned the validity of).

3. The SAGES-3 NVR is probably most like the WPPSI-IV FRI--but not exactly.

4. Neither is more valid, when used as they are designed to be. Children test differently on different occasions, and with different measures, even when the tests are intended to access the same underlying constructs of cognitive ability.

SAGES is also more of a screener than a comprehensive measure. If I were going to pick one to lean on more heavily as a straight-up cognitive measure, it would be the WPPSI, as it's designed as a comprehensive measure, and also had almost twice as many individuals in its standardization sample, but that doesn't mean the SAGES-3 result is not real as far as it goes, for the purpose (in the context of the original complete measure) of identifying appropriate placements to certain types of programming for advanced learners. That original context consisted of two cognitive measures (verbal and nonverbal reasoning), and two academic measures (language/social studies and math/science). I should also note that your DC was comfortably in the middle of the age range on the WPPSI, but at the very bottom of the age range on SAGES, which may or may not have affected the SAGES score more than the WPPSI scores.


...pronounced like the long vowel and first letter of the alphabet...