Here you will find the beginning of the paper. For the full paper in pdf file, go to
http://www.mireadingfirst.org/resources/research/downloads/tr1.pdf .
Evaluation of Reading First in Michigan
Technical Report #1
Do Fluency Measures Predict Reading Achievement?
Results from the 2002-2003 School Year in Michigan’s Reading First Schools
Joanne F. Carlisle
Stephen G. Schilling
Sarah E. Scott
Ji Zeng
University of Michigan
Using Classroom Reading Measures to Estimate Progress
The enactment of the No Child Left Behind Act of 2001 and its constituent
element, Reading First, present profound challenges to states, districts, schools,
administrators, and teachers by emphasizing accountability of schools for the reading
performance of their students from year to year. Reading First, Part B of NCLB, is a
focused initiative to provide high poverty, low achieving schools with the resources to
enable their students to become successful readers. A key component of Reading First is
the use of classroom-based assessment to monitor children’s progress in reading.
To be trustworthy, classroom-based measures of reading must significantly
predict performance on the year-end reading achievement test. If this is the case, teachers
and school administrators can use the measures to assess children’s acquisition of
foundational reading skills and to evaluate the effectiveness of reading instruction for
children at risk for reading failure. The goal is to prevent poor initial performance in
reading from resulting in enduring reading failure. In Michigan, schools that are part of
the Reading First (RF) initiative use a system of fluency-based assessments, the Dynamic
Indicators of Basic Early Literacy Skills(DIBELS), developed by researchers at the
University of Oregon (www.dibels.uoregon.edu) to monitor the progress of early
elementary students in reading. A key component of DIBELS is the use of a series of
benchmarks to index student performance and identify students at risk.
The purpose of this paper is to examine the nature of the relationship between
these fluency-based assessments and the year-end measures of reading achievement
(reading subtests of the Iowa Test of Basic Skills). A number of research reports have
been published in recent years that show concurrent relations of one particular measure of
progress monitoring, oral reading fluency (ORF), and year-end reading achievement
(Barger, 2003, Bucks & Torgesen, 2003; Fuchs, Fuchs, Hosp, & Jenkins, 2001; Good, et
al., 2001). These have shown that students’ performance on measures of oral reading
fluency are significantly related to reading comprehension. Similarly, Good et al. (2001)
has shown that almost all (96%) of third graders scoring above the DIBELS benchmark
on ORF met expectations on a statewide comprehensive assessment of reading.
However, a number of questions need to be addressed, First, few studies have
examined the predictive utility of fluency-based measures other than ORF, such as
phonemic awareness. Second, research on the predictive utility of fluency-based
measures has focused primarily on grades three and above. Third, the research to date
has often been small scale, on the district or school level using demonstration programs
initiated by academic researchers. Moreover, much of this research has typically focused
on normative sample rather than a population with a large number of students at risk for
poor performance in reading. In contrast, this study will examine a full gamut of fluency
measures in a large-scale implementation of RF in Michigan.
Other important questions concern the validity of ORF and other DIBELS
subtests. First, we would benefit from knowing the extent to which performance on
specific DIBELS subtests is related to similar measures of reading on the ITBS. Research
findings suggest that children’s learning of the alphabetic code (e.g., knowledge of letter
names and letter-sound correspondences) is a component of progress in reading in
kindergarten and first grade (Baker & Smith, 2001). We might expect, then, that DIBELS
subtests that assess letter knowledge, phonemic awareness, and decoding of first graders
would be related to performance on the ITBS Word Analysis subtest at year-end. Further,
by second and third grades, children generally recognize letters and letter sounds
automatically and have acquired skill at sounding out simple unfamiliar words. At this
point they face the challenge of acquiring fluency in reading natural texts. It is important,
therefore, to determine whether by the fall of second grade, performance on ORF
measures are more predictive of reading achievement than such foundational skills as
phoneme segmentation and nonsense word reading.
Second, we need to understand the extent to which performance on ORF and
other DIBELS subtests predict performance in reading at the end of the year. Although
researchers have shown that ORF is significantly related to reading achievement, it is
possible that combinations of DIBELS measures might do a better job of predicting
reading achievement in first through third grades than ORF alone. This question might be
particularly important to answer for first grade, where several DIBELS measures other
than ORF are used to determine students’ instructional needs (Good et al, 2001) and
where lower level skills might not have achieved automaticity.
The research literature
suggests that performance in phonemic segmentation might be a particularly strong
predictor of year-end performance on reading (e.g., Wagner & Torgesen, 1987). If
DIBELS subtests administered in fall and winter are found to significantly predict
performance on the ITBS reading measures, teachers would have a basis for using
performance on these subtests to determine whether students are making satisfactory
progress in reading, as is the intent of the researchers who designed DIBELS (Good,
Simmons, Kame’enui, Kaminski, & Wallin, 2002).
Finally, of crucial importance, given the “benchmark