Richard Phelps, via a kind email:

The National Assessment of Educational Progress (NAEP) boasts a long and proud history. First administered in 1969 to national samples, NAEP typically tests students in 4th, 8th, and an upper high school grade in a variety of subjects. Mathematics and reading are tested most often, but other periodically tested topics include science, history, civics, geography, technology literacy, arts, writing, and economics.

The NAEP is “low stakes,” given that neither students, nor teachers, nor schools face any consequences for test performance. Indeed, given the NAEP’s “matrix” sampling design, where only some individual classrooms within schools are selected to participate and, even then, only to administer one test section—just part of a larger whole—test scores cannot be calculated for individual students, classrooms, or schools.

NAEP had long reported test scores for some of the larger states—where sample sizes were large enough naturally to produce statistically representative results—and for smaller states willing to subsidize sample size increases. Around 1990, however, the federal government instituted a bi-annual “State NAEP” in mathematics and reading (and occasionally science or writing), covering the costs for any state wishing to participate. After the passage of the No Child Left Behind Act in 2002, all states would participate.