The core problem is actually very simple. Education studies do not measure what they claim to measure. When they say, "education outcomes improve when..." they usually mean the pass rate, i.e. they only measured a signal among the bottom 20% of students. When they say, "test scores improve when..." they are, at best, measuring up to the 90th percentile. When they say, "the white/black attainment gap," or "socioeconomic disadvantages," they're usually just fishing for funding money, and their study will not actually attempt to measure either of those things. From a review of the literature on No Child Left Behind (NCLB) in 2015:

> Only one study specifically examined the achievement gap for students from low socioeconomic backgrounds (Hampton & Gruenert, 2008) despite NCLB’s stated commitment to improving education for children from low-income families. African American students were often mentioned in studies of general student achievement but none of the reviewed studies focused specifically on the effects of NCLB for this subgroup. Again, this is a curious gap in the research considering the law’s emphasis on narrowing the Black-White achievement gap. Other groups of students underrepresented in the research on NCLB include gifted students, students with vision impairments, and English proficient minority students.

("A Review of the Empirical Literature on No Child Left Behind From 2001 to 2010", Husband & Hunt, 2015)

Everything you see going wrong is downstream of this. Yes, harmful ideologies have done a lot of damage to the education system, but it could easily survive this we had actual signifiers of success.