Response-to-Intervention as an Approach to Preventing and Identifying Learning Disabilities in Reading
Our research addresses the key measurement issues associated with the response-to-intervention (RTI) process:
- Who should enter the RTI process?
- How does one determine whether effective Tier 2 intervention has been conducted?
- What is a valid and practical method of monitoring responsiveness to tier 2 instruction?
- What is a defensible definition of "nonresponsiveness" (i.e., reading disabled) in terms of key reading skills?
The project design incorporates two studies. Study 1 is the screening study designed to examine how three approaches to assessing beginning-of-1st-grade risk, in conjunction with computer-generated decision supports, may be considered separately and in combination to predict reading disabled (RD) risk on decoding, word identification, fluency, vocabulary, comprehension, and spelling at grades 1-4. Study 2 is an intervention study, where we assess methods for measuring intervention implementation and assess procedures for monitoring response and designating RD.
We work in six to eight schools (i.e., 30 classrooms) annually, selected to represent the demographics (i.e., 37% Caucasian, 42% African American, 18% Hispanic, 3% other) of the Metropolitan-Nashville Public Schools, an urban district of 69,000 students. Study 1 involves 452 1st-graders recruited in Year 1 and followed through Year 4. Study 2 is conducted across Years 2-4, with 100 1st-graders recruited and randomly assigned to intervention and control in Year 2 and followed through grade 3; with an additional 80 1st-graders recruited and randomly assigned to intervention and control in Year 3 and followed through grade 2; and with an additional 60 1st-graders recruited and randomly assigned to intervention and control in Year 4 and assessed through end of grade 1.
We conduct tutoring with 1st-grade children selected to be at risk for developing RD in triads for 17 weeks (December-April), 3 sessions per week, 60 min per session. Each session comprises 30 min of decoding, 10 min of fluency, and 20 min of comprehension.
We will develop: (1) a set of procedures for identifying 1st-grade children at risk of developing RD in terms of (a) screening measures, (b) measurement schedules, (c) scales, and (d) cut-points; (2) recommendations as to instruction-specific or instruction-nonspecific fidelity measures are superior for determining whether children received effective Tier 2 instruction within an RTI system; and (3) a set of procedures for classifying "nonresponsiveness" (i.e., RD) in terms of (a) measures, (b) measurement schedules, (c) scales, and (d) cut-points.
Control student will receive the standard school reading curriculum. Study 1 will involve a longitudinal correlational design (grade 1-4) to examine beginning-of-1st-grade risk as a predictor of future reading disabled (RD). Study 2 will involve a randomized controlled trial to examine methods for measuring intervention implementation and assess procedures for monitoring response and designating RD. Key outcomes include decoding, word identification, fluency, vocabulary, comprehension, and spelling growth across grades 1-4. Study 1 will rely on a two-level individual growth curve analysis (HLM) to develop the basic prediction models of academic growth across grades 1-4. Additionally, three techniques (logistic regression, classification and regression tree analysis, and artificial neural nets) will be used to develop, refine, and evaluate our decisions rules for identifying RD at each follow-up point. Study 2 will use a two-level individual growth curve analysis (HLM) to examine the predictive value of the measures derived from the two fidelity assessment approaches, along with possible interactions between fidelity and child-level variables. Finally, descriptive and inferential techniques will be used to examine RD prevalence, severity, predictive utility, and stability within RTI definitions.