Comment by xyzzyz

Did the test collect any data on the demographics of the takers, so that we can check whether the drop is within-group, or whether it’s just artifact of changing composition? For all we know, this might be just a case of Simpson paradox, where each subgroup actually improved, but the overall score distribution shifted downwards due to school children population growing more and more immigrant, after years of mass immigration. While US has excellent education, and immigrant children in America do better than their counterparts in their countries of origin, they don’t do as well as the modal group of white Americans, so as the composition changes, and the whites become less of a statistically dominating factor, the scores are expected to go down, even as the education quality improves.