All sizzle and no steak

Abstract

Houston’s experience with the Educational Value-Added Assessment System (R) (EVAAS) raises questions that other districts should consider before buying the software and using it for high-stakes decisions. Researchers found that teachers in Houston, all of whom were under the EVAAS gun, but who taught relatively more racial minority students, higher proportions of English language learners, higher proportions of students eligible for free or reduced-price lunches, and higher proportions of special education students, had significantly lower EVAAS scores than colleagues teaching elsewhere in the Houston district. Hence, results suggest that the EVAAS does not, at least in Houston and perhaps elsewhere, offer states, districts, and schools the precise, reliable, and unbiased results that go far beyond what other simplistic [value-added] models found in the market today can provide, as the software owner, SAS Institute Inc., claims. Rather, evidence shows that EVAAS estimates in Houston, and likely elsewhere, may be biased against teachers who teach disproportionate percentages of certain type of students in their classrooms.