Co-Principal Investigators: Kuhfeld, Megan; McEachin, Andrew; Mariano, Louis

Purpose: The RAND Corporation and the Northwest Evaluation Association (NWEA) will identify promising analytic methods to address the lack of student standardized testing results caused by extended school closures due to the COVID-19 pandemic. These methods are to help with decisions often based on student test scores regarding:

• Day-to-day instructional and other student related decisions
• School and teacher accountability systems
• Applied education research and program evaluation

Project Activities: The project team will identify the key decisions that need to be made and promising analytic methods to help make them when there is a lack of state assessment data for students through a review of the literature, interviews of researchers and practitioners, a survey of SEAs and education grantmaking organizations, and a technical advisory group.

Researchers will examine the performance of the identified promising analytic methods using both simulation and application to NWEA MAP Growth test scores in reading and math. The team has 5 years of MAP Growth data (2015 -2021) for seven million K-8th grade students in over 17,000 public schools who normally take the MAP Growth three times a year (fall, winter, and spring). The simulations and applications will test each method and compare the methods against one another and against the methods being used by SEAs under a variety of conditions introduced by COVID-19 disruptions. In addition, the project team will use the promising methods with state administrative data from before the pandemic to determine if the results from the MAP Growth are similar to those from state standardized tests.

In addition, the project team will examine the validity of remote testing data for use in both school and district decision-making and research purposes using MAP Growth test data collected through remote testing during the 2020-21 school year.

Products: For each of the three types of decisions, the team will publish a separate practice guide on the most promising methods to support them. The practice guides will include recommendations on how to use each method, the evidence behind them, their strengths and weakness, what conditions they are best suited for, and the R code for implementing them.