Co-Principal Investigator: Biancarosa, Gina; Gearin, Brian

Purpose: The purpose of this project is to evaluate the validity of 8th Edition of Dynamic Indicators of Basic Early Literacy Skills (DIBELS 8) as a screener for dyslexia. As of 2020, 47 states require that students be screened for dyslexia in early elementary school, and state education agencies recognize DIBELS for this purpose more than any other test. However, evidence for the use of DIBELS for this purpose is limited. In this project, researchers will investigate the associations between DIBELS 8 subtests and a range of measures used in dyslexia classification, evaluate a series of classification models for establishing risk-for-dyslexia, and investigate the potential differences in the validity of these approaches to dyslexia identification for students who do and do not receive evidence-based reading instruction.

Project Activities: Project staff will recruit 48 elementary schools and train school-based data collectors to administer DIBELS 8 and standardized word reading, spelling, vocabulary, and reading achievement assessments for 4,800 students in grades K–3 across four cohorts. Researchers will analyze these data to identify the extent to which DIBLES 8 is appropriate for identifying individuals with or at risk for dyslexia.

Products: At the conclusion of the project, the research team will be able to draw conclusions about the validity of the DIBELS 8 for dyslexia screening. These conclusions will be disseminated widely to state and local education agencies, policymakers, advocacy groups, and other stakeholders through peer-reviewed publications and presentations, webinars, and technical documentation.

Structured Abstract

Setting: This project will involve a sample of students recruited from 48 schools across the United States. The schools will comprise a large, nationally representative sample that includes participants from each of the four US Census regions. Schools will include those from urban, suburban and rural settings.

Sample: Participants will include 4,800 students in grades K–3. The sample will be diverse with respect to race, ethnicity, and socioeconomic status, and, in aggregate, will closely match the demographic profile of U.S. schools. English learners and students receiving special education services will be excluded from the sample, consistent with current best practices for dyslexia identification.

Assessment: Based on more than two decades of formative assessments research, DIBELS 8 was developed and validated as a screener for early reading difficulties and for benchmarking and monitoring reading development in grades K–8. DIBELS 8 measures letter naming fluency, phonemic segmentation fluency, nonsense word fluency, word reading fluency, oral reading fluency, and reading comprehension. Subtests are administered individually for one minute, except for the comprehension measure, which is administered to a group for three minutes.

Research Design and Methods: The project will use a multi-cohort, longitudinal design to collect and analyze student reading, response to intervention, and school-level instructional practice data in grades K–3 to investigate the validity, reliability, and generalizability of multiple longitudinal dyslexia classification schemes. In the first year of the project, researchers will identify a sample of K–3 students. Students will be followed in each successive year of the project until they finish 3rd grade. Each year they participate, students will be assessed with the DIBELS 8 subtests three times (in fall, winter and spring) and once with the comparison measures in the early spring.

Key Measures: In addition to the DIBLES 8 subtests, there will be a range of standardized measures typically used in the classification of dyslexia. These include the Test of Word Reading Efficiency-2, Test of Written Spelling-5, Receptive One-Word Picture Vocabulary Test-4, and Woodcock Reading Mastery Tests-III.

Data Analytic Strategy: The research team will examine the validity of DIBELS 8 as a screener for dyslexia using four types of analyses: (a) bivariate correlations with dyslexia measures to assess concurrent and predictive validity, (b) receiver operator characteristic curve analyses to estimate concurrent and predictive screening accuracy, (c) hybrid methods to estimate longitudinal classification schemes, and (d) machine learning approaches that use computer algorithms to predict and classify scores in relation to a pre-defined target. Where appropriate, models will incorporate estimates of student response to intervention. To investigate the effect of instructional context, school-level use of an evidence-based reading program will be included in models as a covariate as appropriate.

Cost Analysis: The proposed project will apply the ingredients method of cost analysis to estimate training, implementation, data entry, and instructional decision-making costs and contrast the use of DIBELS 8 to other recommended approaches to screening for dyslexia.