Develop a Model for Predicting Student Proficiency on Statewide Tests based on Local Benchmark Results



The Challenge: 

School administrators and classroom teachers are increasingly interested in predicting student performance on statewide high-stakes examinations. Obtaining information about students that are at risk as early as possible in the school year is critical to ensure that instruction in properly targeted. Moreover, with the consequences associated with failure to achieve annual yearly progress targets under NCLB, school administrators need accurate information for decision making. While schools often have local Benchmark test results available, they typically do not have an effective way to use this information. This provider of school management systems sought out SEG to help develop a model for using local Benchmark results to predict student proficiency on statewide examinations

The Solution: 

SEG worked with this client to develop a methodology for accurately predicting whether students were at risk for failing to achieve proficiency on statewide high stakes examinations. Using a combination of test linking/equating, Item Response Theory and Multiple Discriminant Analysis, SEG assisted the client in developing a predictive model.