What Do Test Scores in Texas Tell Us?

Texas state TAAS and NAEP trends 1992-1998 for grades 4 and 8, plus a supplemental study administering three tests to about 2,000 fifth graders in 20 Texas schools

US public school students in Texas, grades 4, 5 and 8

State-level trend comparison between a high-stakes state test and an independent audit test (NAEP), plus a small primary administration study. There is no student-level causal design and no identification strategy for the policy itself, which is why it sits at C rather than B - it establishes a divergence, not its cause. Its strength is that the audit measure is genuinely independent and nationally administered.

The Texas "miracle" is largely a TAAS phenomenon. Fourth-grade reading gains on NAEP were 0.13 SD for white students and 0.15 SD for students of colour - barely above the national trend - while TAAS reading gains over the same period were 0.31 to 0.49 SD, several times larger and LARGEST for Black and Hispanic students. Fourth-grade maths on NAEP did rise genuinely above the nation (0.25 to 0.43 SD) but TAAS gains were still much larger. Eighth-grade NAEP results were undistinguished while TAAS gains were large (eighth-grade reading 0.28 SD white, 0.45 Black, 0.37 Hispanic). The racial-gap result is the sharpest: the gap is very large and slightly INCREASING on NAEP while it is much smaller and shrinking fast on TAAS. The supplemental study found school-level TAAS scores correlated only weakly and non-linearly with the Stanford-9. Two tests, same children, opposite stories about the thing policy most cared about.

Low relevance. The comparison is between two tests taken by the same populations, so composition cannot explain a divergence between them.

The same divergence pattern was found in Chicago (Jacob 2005), Kentucky (Koretz & Barron 1998) and in Koretz's re-administration designs; the authors also published a formal reply to a 2005 critique in the same journal.