I see. That helps a bit. So a few other notes that one should keep in mind: administrations less than 24 months apart of the exact same instrument technically result in question marks on the second administration (all other things being equal). So I don't really encourage cherry-picking the higher scaled scores from each subtest for this kind of estimated collection of index scores, especially if any of the higher scores are in the second administration. Of course, it sounds like you already had some doubts about his testing, especially because of the ADHD aspect.
A bit more general context: The district likely only administered seven subtests because that is the standard core battery on the WISC-V necessary to obtain an FSIQ. Nothing notably suspicious about that. His suspected disability was ADHD, not an intellectual or learning disability, so there was no compelling reason for them to do supplementary or ancillary subtests when they were just documenting that he was not intellectually impaired. The university evaluators had different circumstances, and made different decisions. (Other evaluators, including myself, might have made other choices.) The ExNorms probably were not applied because he does not appear to have obtained two 19s in any given index, which is generally considered best practice.
(Incidentally, these preceding two paragraphs taken together mean that if you were really going to create a composite data set, the least concerning way would be to take the scaled scores from the seven subtests administered by the school and combine them with the additional subtests from the university to generate hybrid index scores. I still don't really recommend it, as the reliability of scores is highest for the FSIQ, lower for the index scores, and even lower for individual subtest scores.)
With regard to the variation, one should always start by repeating the caution that subtest scaled scores are substantially lower in test-retest reliability than index or IQ scores. But if there is real variation, you may very well be correct in your speculation that inconsistent attention affected performance. That wouldn't be surprising. I have already mentioned the possible artificial score-raising effects of practice, since the second administration was so close in time. We're not in a position to judge the skill of either set of evaluators, nor do I wish to cast aspersions on anyone, but that's always a possibility. Depending on his age at administration, it's also possible that some of the differences are related to differences in the items presented to him on each occasion, even for subtests with the same name. For example, the two PSI subtests are not actually the same respective tasks at age seven and eight. Even on subtests with more continuity across the ages, differences related to starting at a lower/higher level (due to age), or with different teaching items may affect performance for certain specific examinees. This is not, of course, a comprehensive list of explanations.
And on a side note, I understand why Linda is a bit skeptical about the evaluators you had, based on her experience, since the population of examinees she sees is heavily weighted toward families who were not satisfied with their experiences elsewhere (and certainly, I have seen my share of less-than-optimal evaluations from all types of sources), but I do want to point out that many of the evaluators working in K-12 public schools, clinics and hospitals came out of the same training programs. (As it happens the current presidents of the National Association of School Psychologists (NASP) and the American Psychological Association (APA) were classmates and are close friends.)
I understand that you are trying to get a better sense of what his real profile is (preferably without the $x000 expense of a fresh neuropsych!), but I think this might be a more nuanced and effective discussion starting from the two real datasets, if you are comfortable with that. (Or if not, feel free to add to our existing pm thread.)
