How to use State Test Data in Education Experiments

When the determination is made to use state assessment data in an RCT—that is, once researchers have determined that gaining access to the state data is feasible and that state test scores can be expected to produce reliable and valid information about the impacts of the intervention being evaluated—additional issues related to analyzing the data must be considered. Our discussion does not cover all potential designs or analyses for studies that rely on state assessment data. Instead, we aim to highlight important considerations for researchers formulating their analysis plans and to provide recommendations for deciding among alternative methods.

This part is organized into four sections. The first two sections focus on the costs of obtaining baseline data and the methods for and benefits of using baseline data to increase statistical power. The third section discusses analysis of different types of test scores for estimating program impacts in an RCT. The final section examines issues and methods for combining impact estimates across multiple grades and multiple states, including the assumptions underlying different approaches and recommendations for choosing appropriate methods.