Panel Paper: The Effects of Formative Assessment In Early Elementary School: Evidence From a Random Assignment Study

Friday, November 9, 2012 : 8:20 AM
Salon B (Radisson Plaza Lord Baltimore Hotel)

*Names in bold indicate Presenter

Susanna Loeb and Ben York, Stanford University

Thirty seven percent of 4th graders in the United States do not meet basic reading achievement levels (based on 2007 NAEP results). The percentages are even higher for low-income, black, and students with limited English proficiency (Shanahan et al., 2008). Formative assessment programs that combine information about students’ needs with professional development and collaborative time for teachers are a popular strategy for improving instruction. This study tracks the implementation of a comprehensive, early-literacy formative assessment program in the San Francisco Unified School District (SFUSD), examining the effects of this formative assessment on teaching practices, student outcomes, and the allocation of school resources. SFUSD has implemented the Phonological Awareness Literacy Screening (PALS) in a random sample of prekindergarten, kindergarten and first grade classrooms. PALS is a screening, diagnostic, and progress monitoring system designed to measure the fundamental components of literacy.
Formative assessment programs, such as PALS, may help teachers to improve instruction by providing information on student needs, and providing instructional strategies for addressing those needs (Black & Wiliam, 1998). While they have been popular in upper elementary grades, their use is also increasing in the early elementary years, providing teachers with a systemic look at children’s early literacy skills. Yet little information has been collection on the usefulness of these programs. In fact, we could identify only one study of a randomized experiment of an early literacy assessment program; and, this one study is dated, it only looks at Kindergartners, and it is unclear whether or not the study’s treatment is scalable given its extensive nature.

This study employs a multi-site, cluster-randomized trials design (Spybrook, Raudenbush, Congdon, and Martinez, 2009), whereby we matched almost all of the elementary schools in the District on initial test performance and racial/ethnic/poverty representation. We selected one of the two matched schools randomly to participate in the early-literacy assessment program in the 2011-2012 academic year. The sample includes approximately 25 elementary schools and 14 preschools in the treatment group and an equal number of control schools. All pre-Kindergarten, Kindergarten and 1st grade classes in the treated schools received the treatment which included administration of three early-literacy formative assessments, two program-related professional development sessions given to teachers and school-level instructional coaches; and grade-level team meetings designed to facilitate teacher conversations about the assessment data. These meetings have been lead by the school’s instructional coach, and occur after each test administration.

At the end of the school year (May, 2012), students in both the treatment and control groups will be assessed on the literacy component of the Stanford 10, as well as on executive functioning skills as measured by the Head-Toes-Knees-Shoulders assessment. In addition, teachers from the treatment and control schools will participate in a survey measuring teaching practices and opinions about data and students.