Responsiveness-to-Intervention Symposium

Results of a Three Tiered Model Implementation:
The Heartland Early Literacy Project (HELP)

The remainder of this paper documents an ongoing evaluation study of the effectiveness of Heartland's initial implementation of a Three-Tiered model of service delivery. Our first efforts focused on implementing a comprehensive Early Literacy development program in 53 Heartland Districts and 121 Heartland schools across the past 4 years. The effort is ongoing and has had far-reaching effects on how early literacy skills are measured and taught across our service-delivery area.

Method

Initial HELP schools were recruited in 1999 to participate in an early literacy project designed to improve students pre-reading and early reading skills in grades K-3. The number of schools participating in the project has gradually grown from 36 buildings the first year to 121 buildings during the 03-04 school year.

Independent Variables: Participating schools all received the same general process of professional development. The general framework articulated by Kame'enui and Simmons (1998) was adapted for use with Heartland schools. The process contained a consistent set of steps which include:

1. Participating schools were trained to administer DIBELS measures, initially focusing on the K-1 measures.
2. Large-scale screening data were then collected for each participating site.
3. School and AEA personnel then analyzed and interpreted the school wide DIBELS data at a class, school, grade- and district-level.
4. Based on results, professional development was provided by Heartland staff on validated instructional strategies and programs that focused on the "Big Ideas" in beginning reading. Data based decisions were made to match instructional strategy professional development to student needs.
5. Additionally, data-based decisions were made regarding intensifying support for students who needed strategic and intensive instruction.
6. Ongoing progress monitoring and formative evaluation was conducted and strategies modified based on the data.

Throughout the first years of the project, a majority of professional development has been targeted to kindergarten and first grade teachers. Heartland's general problem-solving logic was applied in each school at a system level to identify whether and where problems were occurring, then appropriate systems-level and individual interventions were selected by each participating site. Differentiated instruction (i.e., more intensive and/or differential instruction for students who needed it) was provided to students at each grade level who needed either supplemental or intensive levels of reading instruction. This instruction was provided in different ways by different constellations of teachers in each building. The common component is that each school examined their student performance data and differentiated instruction for students based on findings. Heartland staff served as trainers and coaches to assist schools in analyzing their school-wide data, examining appropriate group and individual solutions based on the data analysis, and in teaching and coaching school staff during implementation of their selected strategies.

Dependent measures: All of the DIBELS measures were used during the school analysis of student skills and instructional needs (with the exception of Retell Fluency and Word Use Fluency). Measures included those described in Table 2.

Table 2. DIBELS Measures used in Heartland's HELP implementation.

Data from all of these measures were used to monitor progress of students at each grade level and to make instructional decisions regarding grouping of students for instruction. They were used also to identify possible focus of instruction, since the measures themselves are related directly to the "Big Ideas" of early literacy that were identified by the National Reading Panel Report (2002). Data on each of the "Big Ideas" of early literacy were used in each of the participating schools to guide selection of general targets for instruction.

This report include four sets of success indicator data. First, phonemic segmentation fluency (PSF) data from end of kindergarten are reported. PSF was selected as an outcome measure because it is predictive of future reading growth at the end of kindergarten (Kaminski & Good, 1996). Second, nonsense word fluency at the end of kindergarten is reported. Third, oral reading fluency (ORF) data from the end of grades one, two and three are reported. Oral reading fluency measures are reported for grades one through three because ORF has been demonstrated to be a sensitive measure of general reading proficiency, including comprehension (Shinn et al., 1992). Finally, data on special education placement rates in participating schools was collected, by grade level across time. Special education placement rate data were collected to document systematic reductions of special education placements due to increased reading skills in participating schools.

Specific procedures used for collecting these data are presented below.

1. Phonemic segmentation fluency data were collected at the end of Kindergarten for all kindergarten students within each participating school. The phonemic segmentation probes created at the University of Oregon were used in all cases and administration and scoring procedures describe by Good and Kaminski (2002) were used. The number of phonemes that each student correctly said within a one-minute period was counted. Then, each school's data were summarized and graphed against recommended ranges of student performance. Data from each school was summarized and used two times yearly (middle of the year and end of the year) by each school to make both group-level and individual-level instructional grouping decisions. In addition to raw score data, results were converted into z-score statistics to describe on a common scale the effect sizes associated with student growth across multiple school years for each building.
2. Nonsense word fluency data were collected in the middle of the school year and at the end of the year for all kindergarten students within each participating school. The nonsense word fluency probes created at the University of Oregon were used in all cases and administration and scoring procedures describe by Good & Kaminski (2002) were used. The number of sounds from the nonsense words that each student correctly said within a one-minute period was counted. Then, each school's data were summarized and graphed against recommended ranges of student performance. Data from each school was summarized and used two times yearly (middle of the year and end of the year) by each school to make both group-level and individual-level instructional grouping decisions
3. Oral reading fluency data were collected at the end of first, second and third grades for all students participating in the project. Passages were drawn from the DIBELS oral reading fluency passages and administration and scoring procedures described in Shinn (1989) were followed. In addition to raw score data, results were converted into z-score statistics to describe on a common scale, the effect sizes associated with student growth across multiple school years for each building.
4. Finally, to examine the effects of the HELP project on the overall system, the number of new special education identifications in each participating building across the three years prior to the project and during the first four years of project implementation was examined.

Previous Page | Next Page
Summary | Phonemic Segmentation