All right, I can see that it's at least not impossible for the standard scores to be truly standard across grades. The lower grades on non-adaptive tests just have to have different, appropriately aligned ceilings. This is still counterintuitive for me in some ways, but it helps to remember that there's probably a fair amount of one-grade-higher content on each test, with the upper-level content dwindling according to the grade differential. So, for example, if average eighth graders have only mastered fourth grade math, maybe highly capable third graders still do well on roughly fourth-grade math questions presented on the third-grade test, in addition to getting a relative boost from better general test-taking skills and higher accuracy on what they know.
I wonder what percentile actually starts to show some expertise with that grade level's material. 60th? 70th? Whatever it is, that's the bottom percentile that should be discussed in terms of placement, along with whatever percentile is chosen to indicate mastery. Out-of-level 50th percentile scores just aren't very meaningful, and are misleading because many people make the incorrect assumption that a child scoring at that percentile could function like an average child in that higher-level class. Sure, the student could at least for a while get the same standardized test scores, but at some point would be missing important foundational knowledge.
