Value-Added Measures for Schools
Accountability is a hotly debated topic in education. The emphasis on school accountability and its “high stakes” measures of student achievement puts schools and their staff under considerable pressure to perform at high levels. However, some critics believe that the measures used for accountability under No Child Left Behind are not real measures of success and are biased towards already high-performing schools. Several states and school districts have therefore implemented, or are planning implementation of, a new accountability system based on value-added modeling, which is also supported by Race to the Top. This methodology tries to attribute changes in students’ achievement to their school or classroom, taking into account several factors that vary between students and schools. Value-added scores can be used to evaluate schools or teachers and whether their average student performance exceeds or falls behind a similar classroom or school. Proponents of this methodology believe that value-added models are a fairer way to assess performance compared with the more traditional policies that only look at a “snapshot” of current student achievement which is often related to student background. In a recent study, Loris Fagioli compared value-added measures to the traditional measures of school accountability used in California, using data from 29 elementary schools in a large California district. The focus was, specifically, on how school accountability measures such as Adequate Yearly Progress (AYP) and Academic Performance Index (API) relate to value-added measures of those same schools.
The results of this study support the critique voiced against current school accountability measures. API and AYP are very highly correlated with student background. Schools with a majority of low-SES students lag behind schools from more affluent backgrounds and seem to struggle to reach the mandated goal of an API of 800. However, the relationship to background variables was far less under a value-added approach. With a value-added approach, several low-SES schools showed remarkable performance, raising students ’ scores more than 200 points above expected levels. High-SES schools still performed well, with very few schools performing below average. In addition, this research did not find any evidence that a school’s value-added score is negatively impacted by high proportions of advanced students.
Several schools in the district observed in this study did not reach the mandated levels of API. However, this study indicated that not all schools below an API of 800 are necessarily “true” underperformers. In fact, some schools ranking at the bottom of the API scale showed remarkably high scores and raised student achievement more than expected. The results clearly exemplified the advantages of value-added models compared with the currently used metrics. With the high stakes associated with API and AYP and more importantly their high correlation with SES and race, these instruments could be equaled to what Peter Sacks called the “Volvo Effect”: One can guess a school’s performance by looking at the type of cars in the school’s parking lot. The value-added scores for schools however do not seem to commit this fallacy of strong dependency on student background.
More research is certainly needed to investigate the questions addressed in this paper in more detail and with data from more than one district. However, based on the results of this study, a discussion on incorporating a value-added element in current school accountability measures seems relevant and necessary.
The full study can be found in Fagioli, Loris, A comparison between value-added school estimates and currently used metrics of school accountability in California, Educational Assessment, Evaluation and Accountability, January 2014.