You are on page 1of 76

Guided Independent Reading

An Examination of the Reading Practice Database and the


Scientific Research Supporting Guided Independent Reading
As Implemented in Reading Renaissance

By Terrance D. Paul
Copyright ©2003 Renaissance Learning, Inc.

All rights reserved. No portion of this material may be reproduced, by any process or technique,
without the express written consent of the publisher.

AR, Accelerated Reader, Fluent Reader, Reading Renaissance, Renaissance, StandardsMaster, STAR Early Literacy,
and STAR Reading are registered trademarks of Renaissance Learning, Inc. “Better Data, Better Learning,” Renaissance
Learning, and TOPS are common-law trademarks of Renaissance Learning, Inc.

Renaissance Learning, Inc.


PO Box 8036
Wisconsin Rapids, WI 54495-8036
(866) 846-7323
www.renlearn.com
Guided Independent Reading
An Examination of the Reading Practice Database and the
Scientific Research Supporting Guided Independent Reading
As Implemented in Reading Renaissance

1
Executive Summary

Independent literature-based reading, also called independent silent reading or just independent reading,
has been an important educational practice in building reading fluency, vocabulary, and comprehension
for more than 100 years. However, the 2000 National Reading Panel Report (NRP) called into question
the practice of “students reading individually on their own with little or no specific feedback.” The
National Reading Panel raised the question of whether there is a difference between guided independent
reading with feedback and unguided independent reading with no feedback. This study answers that
question and much more.

It turns out there is overwhelming scientific evidence supporting guided independent reading with
feedback, particularly the specific kind of guided independent reading promoted by the Reading
Renaissance program. Guided independent reading achieves high effect sizes and large reading gains
in all grades, 1–12. On the other hand, there is inconsistent evidence for unguided independent reading
with little or no feedback.

This study reviews the scientific research both for independent reading in general and for Reading
Renaissance in particular. However, the main insights of this study are gained from the statistical
examination of the Reading Practice Database (RPD). The RPD is, we believe, the largest database
ever assembled on student independent literature-based reading. The RPD contains the reading
records of 50,823 students who read more than three million books in the 2001–2002 school year.
By data mining the RPD, we confirmed most but not all formerly recommended practices of Reading
Renaissance. We also discovered some new and even startling things about independent reading.
Some of our principal findings are:

• Letting students “just read” on their own can, in some situations, actually lower student
reading achievement relative to their peers. Unguided independent reading is not supported,
particularly for low-achieving students.

• Students need to be highly successful in their independent reading to improve their reading
ability. In the context of Reading Renaissance, this means averaging above 85%, and in some
cases above 90%, correct on Accelerated Reader (AR) Reading Practice Quizzes. It’s not
practice that makes perfect, but highly successful reading practice that leads to reading gains.

• As long as students are highly successful, the more they read, the more they improve their
reading ability. This is true for all ability levels in all grades. However, the law of diminishing
returns applies. Ninety minutes of in-school engaged reading time is only slightly more
beneficial than 60 minutes.

• What determines how much time students spend reading is not their initial reading ability,
but the teacher to whom they are assigned.

2
• Boys not only read less than girls, they read differently. Boys tend to read more nonfiction
and score lower on their AR quizzes. This partially explains the reading gender gap and
also provides a clue as to how to close it.

• Guided independent reading in the primary grades improves the reading ability of
low-achieving students more than high-achieving students and helps close the gap.
However, after fourth grade, the rate of gain for equal amounts of independent reading
practice is about the same for both high- and low-achieving students.

• Nonfiction reading is negatively correlated to reading gains and students are less successful
in their nonfiction reading, scoring lower on their AR quizzes. Many teachers, however, are
able to help students achieve both high levels of success and improved reading ability with
nonfiction reading.

In addition to the many findings, this study announces a new era in educational research. While
the RPD is likely the largest database ever assembled on student independent literature-based
reading, within a few years, fully integrated web-based information systems will enable building
databases that dwarf the size of the RPD. Once this happens, educational research, including
large-scale longitudinal randomized experimental research, will be conducted on a scale never
before thought possible.

3
GUIDED INDEPENDENT READING
An Examination of the Reading Practice Database and the Scientific Research Supporting
Guided Independent Reading As Implemented in Reading Renaissance
Contents
I. Introduction ................................................................................................................................5
II. Overview of Accelerated Reader ...............................................................................................6
III. Overview of Reading Renaissance ...........................................................................................8
IV. Research Supporting Reading Renaissance and Accelerated Reader .....................................11
A. Research Supporting the Components of the Reading Renaissance Design .............11
1. High Levels of Success and Average Percent Correct....................................11
2. Reading at the Appropriate Level (ZPD) ......................................................13
3. Information Feedback.....................................................................................14
4. Goal Setting ...................................................................................................15
5. Personalized Instruction .................................................................................15
6. Time Engaged in Guided Independent Literature-Based Reading ...............15
B. Evidence-Based Research for Reading Renaissance ...................................................20
V. Methodology and the Reading Practice Database...................................................................22
VI. Results ....................................................................................................................................29
A. Engaged Reading Time..............................................................................................29
B. Average Percent Correct .............................................................................................32
C. Optimum Student Goals for Percent Correct and ZPD............................................38
1. The 92% Recommendation ............................................................................38
2. The 85% Rule .................................................................................................38
3. Appropriate Book Level .................................................................................41
D. Effect Size and Expected Improvement for High-Fidelity Implementation .............44
E. Differential Effects .....................................................................................................48
1. High- vs. Low-Achieving Students................................................................48
2. Gender............................................................................................................49
3. Race or Ethnicity............................................................................................50
4. Nonfiction Reading ........................................................................................50
VII. Discussion ............................................................................................................................53
A. Causation....................................................................................................................53
B. Time............................................................................................................................53
C. Fourth-Grade Slump..................................................................................................54
D. Average Percent Correct.............................................................................................55
E. ZPD and Book Level .................................................................................................55
F. Gender and Race .........................................................................................................56
G. High- vs. Low-Achieving Students ...........................................................................56
H. Successful Reading Practice .......................................................................................57
I. Nonfiction Reading......................................................................................................58
J. Effect Size and Reading Improvement ........................................................................58
K. Transferability .............................................................................................................59
VIII. Conclusion ..........................................................................................................................60
IX. Bibliography ...........................................................................................................................61
X. Appendices ............................................................................................................................67
XI. Acknowledgements ................................................................................................................72
4
Introduction

This is a study of guided independent literature-based reading as implemented in the Reading


Renaissance program developed by Renaissance Learning. The study provides strong empirical
support for allocating a significant amount of in-school time for guided independent reading in
all grades, 1–12.1

A primary source of information for this study comes from the statistical examination of the
Reading Practice Database (RPD) assembled by the Research and Evaluation Department of
Renaissance Learning. The RPD is, we believe, the largest database ever assembled on student
independent literature-based reading. The RPD includes the reading records of 50,823 students
in 139 schools in grades 1–12 who read more than three million books in the 2001–2002 school
year. Included for each student are such data as number of books read, readability level of each book,
number of words read, average percent correct on Accelerated Reader (AR) Reading Practice
Quizzes2, pre- and post-test reading scores, gender, and race.

We created the RPD and undertook this study for the following four reasons:

1. To provide evidence for the direction of causation between reading practice and reading
achievement. (Do better readers just choose to read more or does reading more cause
students to become better readers?)

2. To further investigate the characteristics of effective reading practice including (a) average
percent correct on AR quizzes, (b) amount of time allocated for guided independent
reading, and (c) appropriate book levels (zone of proximal development, or ZPD, ranges).

3. To estimate effect size at the classroom level and predict the increase in reading
achievement that can be expected from high-fidelity implementation of guided
independent reading as prescribed in Reading Renaissance.

4. To investigate the differential effects of guided independent reading by gender, race,


reading ability, and amount of nonfiction read.

This study provides the strongest evidence to date supporting the recommended practices of the
Reading Renaissance program. However, this study does not confirm all previous recommendations.
Not confirmed, for example, is the recommendation that teachers encourage students to read more
challenging books when their average percent correct on AR quizzes exceeds 92%. Instead, by exam-
ining the RPD, we found in many situations teachers should help students achieve success levels
well above 92%.

Before discussing our findings, we begin with an overview of AR and Reading Renaissance,
the research supporting both, and a more complete description of the RPD.

1
As used in this paper, guided independent reading has a different meaning than guided reading. Guided independent reading is student
silent reading of self-selected trade books (literature-based reading) in which the teacher is actively involved by guiding choices, estab-
lishing goals, intervening, and instructing when appropriate. Guided reading usually refers to student silent reading of textbook or basal
reader passages in which the teacher asks guided questions to students individually or in small groups to ensure comprehension.
2
In the RPD, only data from AR Reading Practice Quizzes were collected. No data from Literacy Skills Quizzes, textbook quizzes,
or magazine quizzes were collected.
5
Overview of Accelerated Reader

The Accelerated Reader (AR) computerized information system was developed by Judi and Terry
Paul, co-founders of Renaissance Learning. AR was first made available to schools in 1986 and is
currently used in more than half the schools in the U.S. Judi Paul, who has a degree in elementary
education, found through her work in schools and from trying to get her own four children to read
more books, that traditional independent literature-based reading programs did not have a mecha-
nism to provide feedback to students and teachers about student reading practice. Examples of the
traditional independent reading programs then prevalent in schools include: sustained silent reading
(SSR), drop everything and read (DEAR), individualized reading (IR), and free uninterrupted
reading (FUR). Each of these programs required schools to set aside 10–15 minutes per day for
literature-based reading. Teachers monitored student reading practice through book reports,
self-reporting mechanisms including reading journals and logs, and other means.

Judi found these self-reporting mechanisms and book reports to be unreliable indicators of engaged
reading time and successful reading practice. Students often bragged to classmates that they had
fooled the teacher by recording books they had not read in their reading logs. These reading pro-
grams often focused on the number of books read, encouraging students to choose the shortest,
easiest books. Most importantly, reading logs, reading journals, and book reports burdened teachers
by increasing their workload without providing systematic and reliable information feedback on
the quantity and quality of reading practice. Without this information, teachers were unable to guide
reading practice, intervene when necessary, and provide instruction when students were not successful.

AR, a learning information system, was developed to address these specific problems. AR systemati-
cally collects information on student reading practice through short, computer-based quizzes that
assess reading comprehension. With more than 65,000 AR Reading Practice Quizzes, there are
quizzes on practically all books in a school library. AR provides immediate feedback of results to
both the student and the teacher. The teacher uses information from AR to guide further reading,
helping students select books appropriate to their ability levels and interests. The information feed-
back generated by AR also identifies students who are not reading successfully through a quality
measure (average percent correct on AR quizzes) and a quantity measure (AR points, which are
based on the number of words in the book and its readability level).

In short, the purpose of AR is to provide teachers with the information necessary to turn unguided
independent reading into guided independent reading, and thus increase engaged reading time,
ensure more successful reading, and ultimately help students develop into successful readers who
read well and are well read.

AR assigns a point value to each book based on the number of words in the book and its readability
level. The readability level is determined by the ATOS readability formula and is reported in grade
levels from 1–13. (For more information on ATOS, see Appendix A on page 67.) Below is how the
point value of a book is calculated:

AR Points = (10 + Readability Level) x (Words in Book)


100,000

6
Using AR to guide independent reading requires students to read a book and then take an AR
Reading Practice Quiz on the computer. A Reading Practice Quiz consists of 5, 10, or 20 multiple-
choice questions on important facts or events in the book. These quizzes are designed to be relatively
easy, so if students read the book carefully they should be able to get 90 to 100% of the questions
correct, on average. Quality-control procedures assure consistency and reliability of each quiz.

AR points are therefore a fairly accurate measure of the quantity of reading. For a book worth 10
AR points, a student would receive 10 points for a score of 100%, nine points for 90%, etc. The
student must score at least 60% on the quiz to earn any points. This makes it unlikely a student
can earn points without actually reading the book. Security features in the software, as well as
monitoring by the teacher, reduce the possibility of cheating.

Since 1986, AR has become increasingly sophisticated. The software now tracks three types of
reading practice (read aloud, paired reading, and independent reading). There are also several kinds
of quizzes available, including textbook quizzes aligned to all five major reading series (Harcourt,
Houghton Mifflin, Macmillan/McGraw-Hill, Scott Foresman, and SRA McGraw-Hill); Literacy
Skill Quizzes to assess 24 higher-order thinking skills such as inference, main idea, and analysis;
Recorded Voice Quizzes for pre-K to first-grade emergent readers who have participated in a read
aloud; Spanish quizzes for students who are English Language Learners (ELL); and magazine
quizzes (TIME For Kids, Cobblestone, Kids Discover, and Weekly Reader).

In addition to tracking different kinds of reading practice on different kinds of material, AR tracks
percent correct on each quiz taken, average percent correct, points, words read, book titles, book
readability levels, and whether the book is fiction or nonfiction. Through several studies done in
1992 and 1993 (Paul, 1992, 1993), a formula was developed to convert points to an estimate for
minutes of engaged reading time. Detailed reports can be generated at the student, class, grade, and
school level. Educators can also create reports by gender, race, and program type, such as ELL, Title
I, or gifted and talented for any period of time, including day, week, month, marking period, or year.

AR also allows tracking of individual student goals including percent correct, point, and book level
goals. The software then tracks how well students are doing with respect to their goals. In total,
there are more than 50 kinds of reports available in AR. Shown in Appendix B (page 69) are two
of the more commonly used reports: the TOPS Report is printed immediately after a student
finishes a quiz to provide the student and teacher with immediate feedback; the Diagnostic Report
is printed weekly to identify students needing intervention and further instruction.

In summary, AR is a sophisticated computerized information system that provides daily, even hourly,
feedback information on three types of reading practice on a variety of reading materials. Because
it is computerized, it reduces teachers’ paperwork while providing the information needed to guide
and monitor student reading practice. While AR provides information to help teachers manage dif-
ferent kinds of reading practice with different kinds of materials, this study focuses only on student
independent reading of trade books.

7
Overview of Reading Renaissance

Accelerated Reader (AR) is a powerful tool that must be used properly to be effective. Since 1990,
internal and independent researchers have investigated AR to empirically determine best practices
related to the software. These techniques form the foundation of Reading Renaissance, which is the
set of teaching practices proven beneficial in helping students become successful readers who read
well and are well read. Since 1993 when Reading Renaissance was first introduced, more than
460,000 educators nationwide have attended Reading Renaissance professional development and
have been trained on Reading Renaissance techniques.

In 1990, Renaissance Learning began collecting AR and standardized test data in order to identify
teachers nationwide achieving exceptional results with the software. Many teachers were interviewed
by phone and, in some cases, observed in their classrooms. Several of the exceptional teachers and
implementers of AR were recruited as consultants.

In 1992 and 1993, two large-scale studies of 4,498 and 10,124 students, respectively, were completed
(Paul, 1992, 1993). Student AR records for the 1990–91 and 1991–92 school years, together with
pre- and post-test reading scores, showed that AR points were highly correlated to reading growth.
Based on these two studies, it was estimated that an hour a day of guided independent in-school
reading practice could cause an additional full year of growth in reading ability.

These studies also showed that low-achieving students benefited from additional amounts of reading
practice more than high-achieving students, and that the average book level of high-achieving
students was not significantly higher than the book level of average-achieving students.

From these studies, as well as the research literature, it was reasoned that many high-achieving
students were not being sufficiently challenged. To ensure students were sufficiently challenged,
92% average percent correct on AR Reading Practice Quizzes was estimated as the level at which
teachers should encourage students to select more challenging books. At the same time, several
smaller-scale studies showed that unless students were reading successfully (85% or above average
percent correct), growth would be negligible regardless of time spent reading.

In subsequent studies, data on average book level were collected for each grade. This information
was used to create recommended book level ranges according to a student’s grade-equivalent score.
These book level ranges were named zone of proximal development (ZPD) ranges, a term inspired
by the Russian psychologist Lev Vygotsky (Vygotsky, 1973). Each ZPD includes a range of book
levels at which students should be able to read independently while being sufficiently challenged but
not frustrated.

Reading Renaissance recommends that teachers and students focus on three goals for independent
reading practice: quality, quantity, and challenge. Of the three, quality (measured by average percent
correct on AR quizzes) has consistently been found to be most critical in ensuring that reading
practice will lead to improvements in reading ability. Consequently, quality is what we train teachers
to work on first. The next most important factor is quantity, or amount of engaged reading time
(measured by AR points). The third metric is challenge, which is measured by book level.

8
In 1993, the best practices identified through classroom observations, research studies, expert teachers,
and the reading research literature were compiled into a set of recommendations called Reading Renaissance.

Reading Renaissance, like AR, has been continuously improved over the years as new research has
become available. It has also been improved as additional information systems have been developed,
including: STAR Reading computer-adaptive test, a 10-minute reading test for grades 1–12; STAR
Early Literacy diagnostic reading assessment for kindergarten through third-grade students that
provides information feedback in seven domains including phonics, phonemic awareness, vocabulary,
comprehension, etc.; and Fluent Reader, a modeled and repeated oral reading information system.

Reading Renaissance recommendations vary at the different reading levels. For instance, for students
reading at the kindergarten through third-grade levels, guided independent reading is designed to
supplement the direct instruction of phonics skills and phonemic awareness. In high school, Reading
Renaissance is used to manage independent reading practice. Table 1 contains the reading practice
time recommendations by reading level. As stated previously, while Reading Renaissance promotes
three types of reading practice—read aloud, paired reading, and independent reading (also
called read to, read with, read independently, abbreviated to TWI)—this study focuses only
on independent reading.

Table 1: Reading Renaissance Reading Time Recommendations

RECOMMENDED AVERAGE MINUTES OF TWI


READ
READING LEVEL READ TO READ WITH INDEPENDENTLY

K, semester 1 30 0 0
K, semester 2 30 5 5

1, semester 1 30 10 10

1, semester 2 15 15 30

2, semester 1 15 10 35

2, semester 2 15 5 40

3–5 0–15 0–15 60

6–8 0–15 0–15 45

9–12 0–15 0–15 30

Note: Sixty minutes of allocated daily reading practice time had been the standard recommendation for elementary, middle,
and high school. However, because allocating 60 minutes is logistically difficult in the upper grades, the recommendations
were revised in 2003 to 45 minutes a day for middle school and 30 minutes for high school. The recommendation remains
60 minutes for middle and high school students reading below grade level.

9
Although recommendations vary at the different grade levels, Reading Renaissance’s model for
guided independent reading includes six common elements that remain consistent at all grade levels.
These are:

1. Time. Extended in-school time for guided independent literature-based reading of self-
selected books.

2. Success. High levels of successful reading, with an average percent correct of 85% or above
on AR quizzes.

3. Appropriate Level. Students read books matched to their abilities within their recommended
ZPD range for book readability.

4. Information Feedback. Information feedback from three primary sources: (1) daily and weekly
feedback from AR; (2) periodic progress monitoring 3 to 10 times per year using STAR Reading
and STAR Early Literacy computer-adaptive assessments; (3) daily information feedback through
“status of the class,” a technique that involves teachers monitoring and conferencing with each
student on a daily basis.

5. Personalized Goals. Teachers work with each student to establish personalized goals in
three areas: engaged reading practice time, average percent correct on AR quizzes, and average
book level.

6. Personalized Instruction. Teachers combine the information generated by AR with Reading


Renaissance techniques to closely monitor all reading practice—intervening and providing
additional appropriate instruction as needed.

One of the biggest challenges facing all educational programs involves implementation integrity.
Educational programs must be implemented properly to produce the expected results. One of the
greatest advantages of computerized information systems is that an objective measure of implemen-
tation can be calculated using AR data without as much need for actual classroom observation.
In fact, Renaissance Learning uses the information generated from AR to verify that classroom
teachers are properly implementing the program. This process certifies teachers as either Model
or Master Educators. We have found engaged reading time (as measured by AR points) and average
percent correct to be the two most important factors for measuring implementation (Brophy &
Good, 1986; Anderson et al., 1985; Ames, 1990; Walberg, 1996). The importance of these two
measures is consistent with what one would expect from the Academic Learning Time (ALT)
research (Berliner, 1990). Indeed, Reading Renaissance could be called a program for managing
the academic learning time associated with reading practice.

10
Research Supporting Reading Renaissance
and Accelerated Reader

The No Child Left Behind (NCLB) Act of 2001 mandates that schools adopt programs supported
by “scientifically based research.” The quality of this scientifically based evidence is then character-
ized by degree. Under NCLB, there is good, better, and best scientifically based research. A program
has “good” support if its theoretical foundation is based on scientific research. A program has
“better” support if its design is not only based on research but also has actually been proven to
work in the field through “evidence-based” research. Finally, “best” evidence involves research using a
quasi-experimental or experimental design including comparable treatment and control groups
to demonstrate causation. There is also a preference for longitudinal, large-scale, and randomized
control studies as well as multiple studies with replicated results. In short, there is no black-
and-white rule that determines whether an educational product meets the federal definition
of “scientifically based research,” only degrees of evidence.

Under NCLB, demonstrating that a program is proven by scientifically based research is a two-step
process. First, one must show the program components are supported by research (i.e., “research-
based”). Second, one must show the total program, when implemented in real classrooms, works
using quasi-experimental and experimental studies (i.e., “evidence-based”). Consistent with this
two-step approach, we next discuss the scientific research base that supports the design of Reading
Renaissance’s components.

Research Supporting the Components of the Reading Renaissance Design

The six primary components of the Reading Renaissance design are high levels of success,
appropriate level (ZPD), information feedback, goal setting, personalized instruction, and time spent
reading. The research supporting each element is discussed separately.

High Levels of Success and Average Percent Correct

Accelerated Reader (AR) measures successful reading practice according to average percent correct
on AR Reading Practice Quizzes. High levels of success in academic tasks have been shown to be
important to improved educational outcomes (Berliner, 1981; Block, 1971; Bloom, 1976; Denham
& Lieberman, 1980; Fisher, et al., 1978; Skinner, 1954, 1968) and in particular with large gains in
reading (Allington, 1984; Betts, 1946; Rosenshine & Stevens, 1984). An often-mentioned rule is
that for independent reading, students should be able to read 90 to 95% of the words correctly
(Barr, Sadow, & Blachowicz, 1989; Betts, 1946; Fry, 2000; Laufer, 1989; Nation, 1990; Shany &
Biemiller, 1995; Snow, Burns, & Griffin, 1998).

There are both cognitive and motivational reasons high levels of success are important. Success rate
on tasks, usually measured by percent correct, is a more powerful metric than it might seem at first.
It does not just measure comprehension, but also engagement and indirectly even motivation and
interest (Cunningham & Cunningham, 2002). From a cognitive standpoint, high success indicates
there is sufficient help, scaffolding, and context clues for students to learn new material (Beck &
McKeown, 1991). From a motivational perspective, high success assures students are not frustrated,
which is a demotivator (Gage & Berliner, 1992).
11
Several other important outcomes of high rates of success on educational tasks were identified by
Marliave and Filby (1985). Their findings indicate that high rates of success are more critical for
low-achieving students than for high-achieving students; high rates of success have more significant
effects than engaged time on task; and high rates of success are an ongoing learning behavior which
is transferable to other learning tasks.

In applying the research on success to literature-based reading, the descriptive work of Chall and
others on the stages of reading development (Chall, 1996; Ehri, 1995) and in particular the distinc-
tion between “learning to read” and “reading to learn” (Chall, 1996; Chall, Jacobs, & Baldwin, 1990;
Juel, 1991) is relevant. In first through third grades, most students are learning to read, learning
how to decode words, connecting those words up to their listening vocabulary of familiar words,
and gaining reading fluency (Biemiller, 2003). Starting in approximately fourth grade, students
begin “reading to learn.” They are increasingly exposed to unfamiliar words not within their
listening vocabulary. Therefore, they must either learn these new words from the context of the
reading material, from the dictionary, or from direct instruction on the word (Biemiller, 1999;
Chall, Jacobs, & Baldwin, 1990).3

The stages “learning to read” and “reading to learn” help explain what is called the “fourth-grade
slump.” (Snow, Burns, & Griffin, 1998). In essence, the rate of reading growth slows down consider-
ably in the fourth grade because reading becomes more difficult. Connecting written text to familiar
words (as you do when you “learn to read”) is easier than learning new words from context (as you
do when you “read to learn”).

The phenomenon of the fourth-grade slump, the significantly reduced rate of reading growth after
third grade, is likely important with respect to optimal success rates for literature-based reading. We
hypothesize that students need higher success rates (higher percent correct on AR quizzes) after
third grade to reduce the cognitive load and assure there are sufficient context clues to figure out the
meaning of new words (Stahl, 1998).

3
“Reading to learn” has both a narrow and broad meaning. Narrowly it means reading in content area textbooks and expository trade
books. Broadly it means reading any textual material, including narrative fiction, nonfiction, or trade books in which new vocabulary
or knowledge is introduced. We use the broader meaning of “reading to learn” in this paper. In the context of Chall’s stages of reading
development theory, “learning to read” takes place during stages 0–2 and “reading to learn” happens during stages 3–5. According to
Chall, students who are reading to learn, “use reading as a tool for acquiring new knowledge.” With regard to fiction reading, the
“new knowledge” is in many cases new words not in the student’s listening vocabulary.
12
Reading at the Appropriate Level (ZPD)

In order to ensure successful reading practice, students must read books at the appropriate level for
their reading abilities. Indeed, success rate as measured by average percent correct on AR quizzes is
one way to determine if material is at the appropriate level for a child. Matching students to appro-
priate book levels has a long history and is proven in research (Din, 1998; Gersten & Carnine, 1981;
Gickling & Armstrong, 1978; Slavin, Karweit, & Madden, 1989).

The graph below shows the expected relationship between learning rates and the readability level
of a text (either textbook or tradebook).

Figure 1: Relationship Between Text Readability and Learning Rate

In Reading Renaissance, the STAR Reading computer-adaptive assessment is usually used to


provide a recommended book level (or ZPD range) for each student.4 A paper-based nationally
normed reading test can also be used. In either case, the student’s grade equivalent (GE) score is
used to determine a student’s ZPD range.

4
Reading Renaissance’s Goal-Setting Chart (see Appendix C) provides guidance to teachers on the book level range students should
initially be placed in based on their grade equivalent (GE) score. This ZPD chart is only a guideline. Teachers must use their
professional judgment to adjust the ZPD ranges to meet the needs of each student and to move students up or down based on
their success on AR quizzes.
13
Information Feedback

A broad research base supports the importance of information feedback. A number of meta-analysis
studies show strong effect sizes for information feedback ranging from 0.54 to 1.35 (Bloom, 1976;
Haller, Child, & Walberg, 1988; Kumar, 1991; Scheerens & Bosker, 1997; Walberg, 1999). Without
information feedback, teachers cannot monitor student goals, intervene to make adjustments, or
provide appropriate instruction. Feedback also helps students develop meta-cognition skills and
self-control while serving a motivational function (Butler & Nisan, 1986; Palincsar & Brown, 1984).
Furthermore, feedback allows administrators and parents to work from the same objective informa-
tion, fostering better communication and teamwork for joint action (Sadusky & Brem, 2002).

Quizzing students to collect feedback information upon completion of a task helps to ensure that
students will be more attentive and careful when completing the task at hand, and will consequently
gain more from the experience (Black & Wiliam, 1998). Additional validation of this point can be
seen in the meta-analysis study by Walberg (1984) which showed assigned (but not corrected)
homework had an effect size of 0.28 whereas corrected homework had an effect size of 0.79.

Information feedback systems in schools can be broadly classified into three types, or tiers.
This is shown in Figure 2 below.

Figure 2: Information Pyramid

Tier 1 information systems for daily progress monitoring, also called task-level systems, provide
hourly, daily, and weekly formative assessment information linked to the curriculum. AR is a
Tier 1 computerized information system. Tier 1 systems are sometimes referred to as curriculum-
based assessments, curriculum-based instructional management systems, and curriculum-based
monitoring systems (Deno, 1985; Fuchs & Fuchs, 1988). Tier 1 information systems collect most
of the information teachers need to manage important tasks and guide instruction. These systems
do not have to be computerized. In fact, most Tier 1 systems are paper based or even oral.
However, computerized systems help make information collection more systematic, less expensive,
and less time consuming.
14
Tier 2 information systems, also called periodic progress-monitoring systems, are used to
determine if student achievement is increasing as a result of the instruction and accompanying
practice activities. Tier 2 information systems can be either norm- or criterion-referenced. In either
case, they help place students in appropriate material, assess whether Tier 1 learning is effective,
and predict performance on Tier 3 assessments.

Tier 3 information systems today involve high-stakes tests as part of state accountability systems.
Tier 3 information systems are primarily used to measure the effectiveness of a school or district.
The information from Tier 3 systems is typically too little, too late to be useful in the classroom.
Therefore, Tier 3 information systems are summative rather than formative.

In summary, research indicates that information feedback is fundamental to effective learning,


teaching, and the operation of schools. Hattie (1992), after reviewing almost 8,000 feedback
studies wrote, “The most powerful single modification that enhances achievement is feedback.”

The recognition that teachers, librarians, administrators, and students lacked frequent, systematic,
reliable, and objective information feedback at both the Tier 1 and 2 level is what led to the creation
of AR and all the other computerized learning information systems developed by Renaissance Learning.

Goal Setting

Personalized student goals establish clear expectations and provide a checkpoint to measure
progress and trigger intervention. Goals are also motivational for students and teachers. A number
of meta-analyses suggest relatively strong effect sizes for goal setting, from 0.46 to 0.55 (Lipsey &
Wilson, 1993; Marzano, 2003; Walberg, 1999; Wise & Okey, 1983). Goal setting not only includes
the actual negotiation between the student and teacher to establish attainable goals, but also assumes
an information system is in place so progress can be periodically measured. There needs to be a
feedback system in place.

Personalized Instruction

Adjusting instruction to address the individual needs of students is supported in the research
and commonly accepted as good practice. In a 1984 meta-analysis, Walberg found effect sizes for
personalized instruction ranging from 0.32 to 0.57. Bloom (1984) estimated an effect size of 2.0
for one-on-one tutoring, the ultimate in individualization. One-on-one tutoring, of course, is not
possible in a classroom of 20 to 30 students and one teacher. One of the difficulties with personaliz-
ing instruction is the extra work required to assess and track individual progress. This workload can
be reduced with computerized learning information systems.

Time Engaged in Guided Independent Literature-Based Reading

All six elements of the Reading Renaissance design are highly supported by the research literature.
However, one of these elements, time engaged in literature-based reading, is surrounded in contro-
versy. Because it is intuitively logical that the more students read, the better readers they become
(how do you get better at anything without practice?), many are surprised to discover how
controversial this topic really is. The story behind this controversy is central to the current study.

15
Tier 1 information systems are designed to help teachers manage time on task and ensure students
are academically engaged. The controversy surrounding time engaged in literature-based reading,
however, relates to the question of whether the task being managed is worthwhile. Specifically, does
time spent reading improve student achievement? The answer is most definitely yes—if it’s the right
kind of independent reading. To understand the controversy about this topic and our conclusions,
some background information is helpful.

In 1985, the Commission on Reading published Becoming a Nation of Readers. Funded by the U.S.
Department of Education, the Commission brought together the top authorities in the country to
produce a concise, comprehensive summary of the research in reading. Becoming a Nation of Readers
supported direct systematic phonics and phonemic awareness instruction over the look-say or whole
word approach. It also supported independent reading and recommended schools allocate more
in-school time to independent reading with statements such as: “Children of every age and ability
ought to be doing more extended silent reading,” (p. 54) and “priority should be given to independent
reading.” (Anderson, et al., 1985, p. 82.)

Fifteen years after Becoming a Nation of Readers was published, a different national commission
on reading, the National Reading Panel (NRP), published another review of the reading research.
Whereas Becoming a Nation of Readers was funded by the Department of Education, the NRP report
was funded by the National Institute of Child Health and Human Development (NICHD). The NRP
report largely supported the recommendations made in Becoming a Nation of Readers—with one major
exception: The NRP found no evidence to support the value of independent reading. On this topic,
the NRP concluded:

“There has been widespread agreement in the literature that encouraging students to engage
in wide, independent, silent reading increases reading achievement. Literally hundreds of
correlational studies find that the best readers read the most and that poor readers read the
least. These correlational studies suggest that the more that children read, the better their
fluency, vocabulary, and comprehension. However, these findings are correlational in nature,
and correlation does not imply causation. No doubt, it could be that the more that children
read, the more their reading skills improve, but it is also possible that better readers simply
choose to read more.” (NICHD, 2000a, p. 12).

In other words, in the 15 years between the publication of Becoming a Nation of Readers and the
NRP report, independent literature-based reading went from being highly recommended and
supported in research to questionable in value. What happened?

The answer is whole language. The 15-year period between 1985 and 2000 came to be known as
the “reading wars.” Proponents of direct systematic instruction of phonics skills and phonemic
awareness were pitted against those who believed reading and writing developed naturally. Luckily,
no shots were fired, and the war was carried out largely in journals and verbal harangues in schools
of education. By 1990, whole language had become the dominant method prescribed by schools of
education for teaching children to read.

16
Whole language advocates downplay direct phonics instruction and recommend that students learn
to read naturally, similar to how children learn to talk, by first learning a basic set of sight words and
then by figuring out new words from the context (Goodman, 1986; Johnson & Stone, 1991; Rich,
1985). Whole language also endorses independent reading with self-reporting methods for gathering
feedback information (Goodman, Goodman, & Hood, 1989). In other words, this was the kind of
unguided independent reading Judi Paul encountered in schools in the 1980s that inspired her to
develop AR.

When it comes to feedback, the whole language movement prefers qualitative to quantitative
methods of information gathering and is generally opposed to standardized assessments and
multiple-choice tests (Goodman et al., 1987; Kohn, 2000). This conflicts with the quantitative
and experimental methods of scientific inquiry used in many fields, especially the natural sciences.
This is one of the reasons many of whole language’s harshest critics come from the fields of physics,
chemistry, biology, and medicine.

However, whole language appeals to many educators. By the mid-1990s, proponents of direct
instruction of decoding skills were in the minority in schools of education and even within the U.S.
Department of Education—in spite of the recommendations of Becoming a Nation of Readers.

The tide began to turn against whole language in 1995 when a letter was written to the Governor
of Massachusetts by Professor David Pesetsky of MIT and signed by 40 other nationally and world-
renowned linguists, psycholinguists, cognitive psychologists, pediatricians, and child development
experts—protesting the statewide adoption of whole language as the approved method of teaching
reading. These 40 eminent scientists wrote that whole language is based on “. . . an erroneous view
of how human language works, a view that runs counter to most of the major scientific results of
more than 100 years of linguistics and psycholinguistics.” Interestingly, the revolt against whole
language came from outside the educational community, not from within.

Pesetsky’s letter made newspaper headlines and stopped the whole language adoption in
Massachusetts. But these 40 scientists had power beyond Massachusetts, including the halls of
Congress and the NICHD. In 1997, Congress instructed the NICHD “in consultation with the
Department of Education” to assemble a national panel of experts (the NRP) to assess and report on
the quantitative scientific research on teaching students to read. Fourteen experts were selected. Not
surprisingly, given how the revolt against whole language arose, the panel chairman was an eminent
physicist, Donald N. Langenberg.

17
The NRP was directed to base reading instruction on rigorous, quantitative, experimental
evidence. The NRP report was intended to end the reading wars. The NRP confirmed what
Becoming a Nation of Readers had said 15 years earlier: scientific evidence showed direct instruction
in the areas of phonics, phonemic awareness, vocabulary, comprehension, and fluency was effective
in improving reading achievement. Students who received such instruction perform better on
standardized reading tests. However, where the two federal commissions differ is on the question
of independent literature-based reading practice. The NRP report directly contradicts Becoming
a Nation of Readers on this point.5

Given the history of the reading wars, criticism of the NRP report has been surprisingly muted.
One of the major criticisms involves its narrow focus. The Panel focused on only five topics in read-
ing and reviewed only experimental research published in peer-reviewed journals in the U.S. from
1966 to 1999. The other major criticism of the report involves the conclusion that independent
reading is “unproven” because it is only supported by correlational studies and not the experimental
studies required by the NRP methodology. In addition, of the studies actually reviewed by the NRP,
most focused on “silent reading procedures, that is, students reading individually on their own with
little or no specific feedback” (NICHD, 2000b, p. 21). In other words, they examined unguided
independent reading advocated by the whole language movement.

It is not surprising that the NRP found only correlational studies, rather than experimental studies,
of independent reading. Experimental studies are difficult to conduct and independent reading has
been an accepted classroom practice for more than 100 years. It only became controversial recently
when the whole language movement advocated unguided independent reading to the exclusion of
direct instruction of decoding skills—what many would call an “unbalanced” approach.

Dr. S.J. Samuels, a member of the NRP and co-author of the section on fluency, was concerned
that the report, because of its narrow scope (in part caused by resource and time constraints),
did not do justice to the topic of independent reading. Consequently, after the report was published,
he embarked on three research projects: (1) a broader review and meta-analysis of the research
literature on independent reading; (2) a randomized experimental study investigating whether
higher amounts of classroom time spent reading independently caused higher levels of reading
achievement; and (3) a randomized experimental study examining whether independent reading
with information feedback (using AR) caused higher achievement than independent reading
without systematic feedback.

5
On the other hand, the 2000 Performance and Engagement Across Countries study, conducted by the Programme for International
Student Assessment (PISA), supports Becoming a Nation of Readers and contradicts the National Reading Panel on this point. The PISA
study examined the reading behaviors of 15-year-olds in more than 30 countries and found that “all students who are highly engaged in
reading achieve reading literacy scores that are significantly above the international mean, whatever their family background. Conversely,
students who are poorly engaged in reading achieve scores below the international mean, regardless of their parents’ occupational status”
(Kirsch et al., 2002).
18
The following are Samuels’ findings from these studies:

1. The extensive meta analysis and literature review on independent reading found:

a. Forty-nine studies on the effects of independent reading have sufficient quantitative


data to compute effect sizes. Of these 49 studies, 8 are experimental, 24 are quasi-
experimental, and 17 are correlational. Forty-five of the studies reported positive
results of improved reading scores, while four reported negative results that were not
statistically significant (Lewis & Samuels, 2003).

b. The effect size calculated for the eight randomized experimental control studies
on independent reading practice is 0.42 (Lewis & Samuels, 2003).

2. The randomized experiment of third- and fifth-grade students showed that high- and average-
achieving students who read 40 minutes per day had significantly greater improvement on
standardized measures of reading than students who read 15 minutes per day. Low-achieving
students benefited from smaller amounts of reading practice at a time (Samuels, et al., 2003b).

3. The randomized experiment showed third- and fifth-grade students of teachers who used AR to
help guide reading and provide feedback significantly outpaced students not using AR who read
for the same amount of time (Samuels et al., 2003a).

One of the leading authorities on the impact of independent literature-based reading is Richard C.
Anderson, who heads the Center for the Study of Reading at the University of Illinois. Anderson
also chaired the Commission on Reading which authored Becoming a Nation of Readers. In a 1996
review of both domestic and international research on independent reading, Anderson considered
both correlational and experimental studies (causation studies) and found remarkable consistency
supporting the strong and positive relationship between time spent reading and reading growth
(Anderson, 1996). In addition, there are a number of high-quality international studies on
independent reading not reviewed by the NRP which also provide evidence for the effectiveness
of independent reading practice. (See, for example, the well-designed Elley and Mangubhai (1983)
experimental study of 614 Fijian students.)

Anderson’s 1996 review also includes some experimental studies of the whole language approach to
independent reading. These studies provide little evidence for the effectiveness of unguided reading
practice (Stahl, McKenna, & Pagnucco, 1994; Stahl & Miller, 1989). With regard to these studies,
Anderson states:

“The apparently weak and inconsistent results may mean that whole language has
dysfunctional aspects or that the philosophy is being imperfectly realized in practice.
Immersing children in literature is one of the axioms of whole language. The research
I have reviewed elsewhere in this chapter shows fairly strong and rather consistent benefits
to basic literacy from immersing children in literature. So, unless there are problems in the
ways whole language is being realized in practice, it is otherwise curious why research has
not been able to show consistent benefits in whole language classrooms.” (p. 72).

19
Whole language recommends that students should be free to choose books and read without
interference from the teacher. Consequently, this method supports student self-reporting of books
read, rather than systematic feedback such as that generated by AR. However, the Lewis and
Samuels study (2003) clearly shows that systematic, quantitative feedback improves the effectiveness
of independent reading practice. In fact, Lewis and Samuels concluded that “the evidence from
this review strongly supports the belief that students who read more are better readers; there is
also evidence that reading contributes to higher achievement and does not merely correlate with it.”
(p. 27). These findings are corroborated by additional evidence showing that actively monitoring
reading practice leads to better gains for students (Anderson, Wilson, & Fielding, 1988; Yu, 1993).

The current study provides further investigation into why guided independent reading is significantly
more effective than unguided independent reading.

Evidence-Based Research for Reading Renaissance

The previous section reviewed the scientifically based research supporting the six major components
of Reading Renaissance’s approach to guided independent reading as required by the No Child Left
Behind Act. The other type of evidence needed to demonstrate the program’s effectiveness is called
evidence-based research. In other words, once we’ve shown it works in theory, we must also show it
works in practice.

There is a large and growing number of independent quasi-experimental and experimental studies
demonstrating the effectiveness of the Reading Renaissance approach to guided independent read-
ing. The matrix below illustrates the evidence base of Reading Renaissance as of May 2003.

Table 2: Reading Renaissance Research Matrix


Reading Renaissance is Proven by
109 Scientific Research Studies
TYPE OF RENAISSANCE INDEPENDENT RESEARCH
LEARNING TOTAL
RESEARCH FOR
STUDY INTERNAL
RESEARCH SCHOOL UNIVERSITY GROUP

Scientifically Based
Research 3 2 16 21
(NCLB definition)

Correlational and
Qualitative Research 7 50 7 64
(no controls)

Product Foundation
Research Papers
14 n/a n/a 14

Reliability and Validity


Assessment Research
3 1 6 10

TOTAL 27 53 29 109

Total Independent Research Studies = 82

20
Summaries of all the research supporting Reading Renaissance and AR are available online
at www.renlearn.com.

Recent studies done by independent university researchers report the following effect sizes for
implementation of Reading Renaissance in whole schools or districts: 0.57 (Sadusky & Brem,
2002), 0.50 (Holmes & Brown, 2003), and 0.22 (Nunnery, Ross & Goldfeder, 2003). While an
effect size of 0.22 would not necessarily be considered significant, Nunnery, Ross, and Goldfeder
studied a districtwide implementation of Reading Renaissance and reported that an effect size of
0.22 compares favorably with other schoolwide models such as Success for All, which is considered
to be proven by scientific research.

Topping and Sanders (2000) published an important study on Reading Renaissance (too late to
be considered by the NRP) that has particular relevance to the current study. This study merged
a database of AR records for nearly 63,000 students in grades 2 through 8 in Tennessee with the
Tennessee Value-Added Assessment System (TVAAS) database. Results of this sophisticated
statistical analysis provide evidence of causation: the more students read the better they read.
Consistent with the research reported above regarding the importance of high levels of success,
Topping and Sanders demonstrated that while both time spent reading (measured by AR points)
and average percent correct (on AR Reading Practice Quizzes) produced statistically significant
effects, average percent correct was more important. Specifically, results indicate that unless students
average 85% or above on AR quizzes, little reading growth occurs. This confirmed one of the
primary recommendations in Reading Renaissance that all students achieve 85% correct or higher
on AR quizzes. The study also showed that teachers completing Reading Renaissance training were
significantly more effective than control teachers who also used AR but had not been trained. In
addition, teachers certified as Model Classroom Educators were more effective in helping students’
reading improve than were non-Model-Certified Educators (Sanders & Topping, 1999). These
findings, consistent with the findings of the current study, demonstrate that Reading Renaissance-
trained teachers are the key to implementing guided independent reading and to helping students
become successful readers.

One phenomenon observed by Topping and Sanders that was inconsistent with Reading Renaissance
recommendations involved book level. Specifically, book level was negatively correlated with
improved teacher effectiveness and reading growth. Instead, students who read easier books
did better than students who read harder books.

All of these topics, including percent correct, time spent reading, and book level (with a possible
explanation as to why Topping and Sanders found a negative correlation) are explored extensively in
the current study.

21
Methodology and the Reading Practice Database
Approach. This study was conducted by collecting and analyzing extensive student data from U.S.
schools using Accelerated Reader (AR) in the 2001–2002 school year. The sections below describe
the sample, the measurement instruments used, the variables of interest, types of analyses performed,
and methodological issues specific to the study.

The Sample

Sample selection. Given that the purpose of this study is to examine the effects of guided independ-
ent reading as prescribed in Reading Renaissance, the sampling was focused on getting data from
schools we believed were likely to be implementing Reading Renaissance. It is therefore not a
nationally representative sample but rather a representative sample of Reading Renaissance schools.

AR reading practice data and STAR Reading pre- and post-test scores for the 2001–2002 school
year were collected from 139 schools in 24 states (Ala., Ark., Ariz., Calif., Colo., Del., Ga., Idaho,
Ill., Ky., La., Mich., Mo., Miss., N.C., Neb., N.M., Okla., Pa., S.C., Texas, Utah, Va., Wis.).

Within the sample of schools, 72 (51.8%) were certified as Model Schools. Additionally, 1,171
(43%) classroom teachers were certified as Model or Master Educators in reading (including those
Model or Master Classrooms within Model Schools). From the 139 schools included in the RPD,
there were 2,365 classrooms with between 12 and 30 students per class.6 Because there were a large
number of classrooms implementing Reading Renaissance along with many who were not, there is
significant variability in the RPD. Conclusions for grades 8–12 are likely to be more tentative as
the number of students and Model Classrooms drops off considerably after the seventh grade.
Eighteen-and-one-half percent of the students and less than 1% of the Model Classrooms were
from the eighth to twelfth grades.

Method of data collection. Schools were contacted in spring 2002 via a recruitment letter with
follow-up telephone calls to non-respondents. Schools with district consulting contracts had a
response rate of 78.8%; schools without district consulting contracts responded at a 9.6% rate.
Overall, the total response rate was 20%. Schools were provided instructions for uploading or
emailing AR and STAR Reading data directly to a third-party research company. The company
then forwarded the data to Renaissance Learning, while maintaining a copy of the raw files.
This allows independent researchers to verify the data and published results.

6
Some teachers use AR with only a small number of students within their class. Because of the probable differences in Reading
Renaissance implementation with a very small number of students as compared to a full-class implementation, classroom-level
analyses in this study will include only those classrooms with between 12 and 30 students.

22
Measurement Instruments

STAR Reading computer-adaptive test and database. STAR Reading is a nationally normed
standardized reading test designed to provide teachers grade equivalent, percentile, and normal
curve equivalent scores in 10 minutes. STAR Reading can be used repeatedly during the year to
measure student, class, or school progress. It is a Tier 2 information system. STAR Reading uses
adaptive-branching procedures to tailor each student’s test to his or her ability level based on his
or her responses to previous test items. Test-retest reliabilities, by grade, range from 0.79 to 0.91,
with an overall reliability of 0.94. Generic reliabilities, as derived from individual standard errors
of measurement, range from 0.89 to 0.92, and vary little from grade to grade. Split-half reliabilities
range from 0.89 to 0.93.

The validity of STAR Reading is demonstrated by high correlations between STAR Reading scores
and many other norm-referenced standardized test instruments, including the Iowa Test of Basic
Skills (average r = 0.74 across Forms K, L, M for grades 1–6) and the Stanford Achievement Test,
Ninth Edition (average r = 0.69 for grades 1–6).

Although STAR Reading’s reliability and validity is comparable to nationally normed 50-minute,
paper-based Tier 2 assessments, its primary purpose is to provide formative information and intra-
year progress monitoring. As such, while STAR Reading itself is standardized, the actual test
administration is not as controlled as is typical of high-stakes (Tier 3) assessments.

Accelerated Reader (AR). The vast majority of the data available in the RPD is from AR. Details
of the type of data collected by AR were described earlier in this report. However, it is important
to reiterate that given the focus of this study, only data from Reading Practice Quizzes were includ-
ed in these analyses. No data from Literacy Skill Quizzes, textbook quizzes, or magazine quizzes
were included in these analyses.

Principal Variables. AR and STAR Reading collect and store information on student independent
reading and reading achievement progress throughout the year. Of the many variables available in the
data from the two programs, those listed in Table 3 are the ones used in the analyses presented here.

23
Table 3: Principal Variables
Variable Name Source/Formula
Normal Curve STAR Reading (available for pre- and post-test)
Equivalent (NCE)
Percentile STAR Reading (available for pre- and post-test)
Scale Score STAR Reading (available for pre- and post-test)
NCE Gain Computed per formula on page 25;
see Regression to the mean section (page 25) for additional information
Average Book Level AR/averaged over all AR books read between pre- and post-test
Average Percent Correct AR/averaged over all AR books read between pre- and post-test
Total Points Earned AR/sum of all points earned on AR quizzes taken between pre- and post-test
Percent Fiction/Nonfiction AR/percent of AR books categorized as fiction or nonfiction
Engaged Reading Time Computed from AR/equal to [(actual points earned) ÷ (points expected
based on the Goal-Setting Chart)] x 60; see “Calculation of Estimated
Engaged Reading Time from AR Points” (on page 28) for additional
information
Challenge Computed from AR/equal to (actual average book level) ÷ (predicted
average book level from regression by pretest NCE level)
Implementation Index Class or grade level variable computed from AR/equal to (proportion of
students in the class with average percent correct ≥ 85%) x (proportion of
60 minutes engaged reading time for class)
Race AR or STAR Reading, where available/see “Demographic Statistics”
(page 26) for additional information
Gender AR or STAR Reading, where available/see “Demographic Statistics”
(page 26) for additional information
Renaissance Renaissance Learning customer records
Certification Status

First-grade issues. Due to the significant differences between data from first-grade classrooms and
data from higher-grade classrooms, combined analysis in this study generally exclude data from first-
grade classrooms. The reasons are as follows:

STAR Reading was normed in the spring and normed scores for first-grade students are projected
backward to provide scores for students tested in the fall. However, because there is not norming
data for first-grade students in the fall or kindergarten students in the spring, the actual beginning
point for first-grade students remains in some doubt. STAR Reading is designed to be given to stu-
dents with a sight vocabulary of at least 100 words. Given the vast variability in students’ beginning
reading level in the fall of first grade, some of the growth measured by STAR Reading is likely due
to pretesting of students who are not yet able to read, resulting in abnormally low pretest scores and
likely overstated estimates of reading improvement.

24
In addition, a significant percentage of AR data collected on first-grade students reflects read-aloud
and paired-reading experiences rather than independent reading. Based on survey data, we estimate
that as much as 50% of reading in first grade was read-aloud. While AR has the capability of track-
ing independent reading vs. read-aloud activity, teachers may not have enabled this feature or may
have had an older version of AR without the tracking feature. Second-grade survey data indicate
that 80% or more of the reading is independent. By the third grade, it is more than 90%.

Use of NCE scores. The most stable of the reading achievement scores reported by STAR Reading
are scaled scores. However, the use of scaled scores in analyses would have resulted in imprecise
means and large standard deviations when groups of students in multiple grades were combined for
analysis. Therefore, it was most desirable to use a norm-referenced score that would allow for easy
interpretation of score changes across grades. Of the norm-referenced scores available—grade equiv-
alent, percentile, and normal curve equivalent (NCE)—only NCE is based on an equal interval
scale, making it appropriate for use in statistical analysis.

At times, however, some results may be reported in percentiles because of the familiarity of this score
to educators. Percentile change scores are obtained by converting the beginning and ending NCE
scores to percentiles.

Regression to the mean. Although the current vogue for statistical analysis in educational research
focuses on Analysis of Covariance (ANCOVA) techniques which allow for the researcher to take
students’ prior achievements into account, results of these analyses are difficult to translate into
recommendations for classroom practice. The use of change scores, when properly adjusted,
reveal the same information as ANCOVA analyses in a more easily understood fashion (Allison,
1990). One concern with using change scores, however, is controlling for the regression artifact
known as regression to the mean.

Simply stated, regression to the mean is demonstrated by observing that those students with the
lowest pretest scores tend to do better on a post-test, and those students with the highest pretest
scores tend to do worse on a post-test—even when no intervention occurs. Failure to account for
regression to the mean can result in attributions of very high gain for low-performing students
where in actuality the gain may be simply due to the imperfect correlation between measures.

Campbell & Kenny (1999) present corrections for regression to the mean for a variety of different
study designs in their book A Primer on Regression Artifacts. The standard adjustment for regression
to the mean for change scores is the residualized change score, which seeks to determine what the
change for an individual would have been had everyone in the sample started at the same point.
The formula for residualized change score is:

Y – bYX(X – MX) – MY
where:
Y = observed post-test score
bYX = slope of the regression equation Y = a + bX
X = observed pretest score
MX = group mean of pretest scores
MY = group mean of post-test scores

25
Using residualized change scores in place of raw change scores results in more conservative and
more accurate estimates of the effects of a program. The magnitude of the effect of this adjustment
is shown in Table 4, which displays the mean raw NCE change scores and the mean adjusted NCE
change scores for students with pretest scores above and below the median pretest NCE by grade in
the RPD.

Table 4: Raw vs. Adjusted NCE Change Scores by Grade


Pretest NCE Score
Below Median Above Median
Grade Raw NCE Adjusted Raw NCE Adjusted
Change NCE Change NCE
Score Change Score Change
Score Score
First 23.99 19.87 7.09 10.46
Second 9.01 6.41 3.03 7.06
Third 8.46 5.36 –2.05 1.79
Fourth 5.85 3.09 –0.72 2.76
Fifth 3.13 0.43 –2.46 0.95
Sixth 1.71 –1.19 –4.34 –0.89
Seventh 2.62 –0.23 –2.67 0.65
Eighth 3.34 0.10 –3.85 –0.23
Ninth 2.88 –0.23 –2.84 0.64
Tenth 3.50 0.14 –2.76 0.62
Eleventh 4.01 0.88 –2.35 1.14
Twelfth 3.06 0.04 –5.78 –2.46

All analyses in this study showing reading growth use the residualized change score (Adjusted
NCE) in order to avoid overstating the effects of AR and Reading Renaissance with respect
to low-ability students or understating the effects with respect to high-ability students.

Demographic statistics. Demographic data for students in the RPD were not uniformly available.
Not all participating schools inputted even the most basic demographic information on their stu-
dents into AR or STAR Reading which resulted in a large number of students for whom race and
gender are not known. We were able to match student names to genders in most cases with the aid
of Peoplesmith™ Software’s Personator™ program, which was able to assign gender to 75.8% of the
names. Lists of popular baby names from different ethnic backgrounds compiled from the Internet,
purged of names that could be male or female, enabled gender assignment to another 4.5% of the
individuals in the original list. After the adjustments indicated by these procedures, the demographic
breakdown of the RPD appears in Table 5.

26
Table 5: Number of Students by Grade, Race, and Gender
Grade 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th 11th 12th Total
Male 5 6 13 8 8 5 4 6 12 10 5 0 82
Asian Female 12 6 8 6 9 7 3 9 13 7 4 5 89
Unknown 0 0 0 0 0 0 0 0 0 0 0 0 0
Male 113 231 285 191 155 104 87 61 67 41 26 15 1,376
African
Female 112 203 266 210 158 84 84 78 55 38 27 23 1,338
American
Unknown 1 0 4 0 1 5 1 0 1 0 0 0 13
Male 67 87 84 73 82 158 68 108 187 46 32 31 1,023
Hispanic Female 79 85 70 58 67 151 66 98 177 44 50 21 966
Unknown 0 0 2 0 1 0 0 0 0 0 0 0 3
Male 2 2 2 4 1 3 39 4 4 4 2 2 69
Native
Female 4 5 1 9 4 1 40 3 7 3 1 1 79
American
Unknown 1 0 0 0 0 0 0 0 0 0 0 0 1
Male 363 683 736 576 443 227 205 306 330 316 237 171 4,593
White Female 344 631 630 545 380 267 179 305 296 306 231 161 4,275
Unknown 0 2 4 1 0 0 1 0 0 0 1 0 9
Male 1,577 2,130 2,035 2,026 1,805 1,508 1,018 729 529 381 304 282 14,324
Unknown Female 1,403 1,864 1,728 1,726 1,570 1,379 859 705 532 345 307 300 12,718
Unknown 1,070 1,762 1,484 1,417 1,144 987 976 283 285 167 144 146 9,865
Male 2,127 3,139 3,155 2,878 2,494 2,005 1,421 1,214 1,129 798 606 501 21,467
Total Female 1,954 2,794 2,703 2,554 2,188 1,889 1,231 1,198 1,080 743 620 511 19,465
Unknown 1,072 1,764 1,494 1,418 1,146 992 978 283 286 167 145 146 9,891
Grand Total 5,153 7,697 7,352 6,850 5,828 4,886 3,630 2,695 2,495 1,708 1,371 1,158 50,823
27
Calculation of Estimated Engaged Reading Time from AR Points. In order to make Reading
Renaissance recommendations easy to translate into classroom practice, Renaissance Learning
developed the Goal-Setting Chart. The Goal-Setting Chart (see Appendix C) shows the approximate
number of points students of varying achievement levels (grade-equivalent scores) should be able
to earn with 60 minutes of daily guided independent reading practice. The conversion formula
for points to engaged reading time at various grade equivalent scores was originally based on data
from Paul’s 1992 reading study. The point-time conversion was recently validated in two ways:
(1) by comparing the predicted number of words read by students from the Goal-Setting Chart to
independent research on student reading rates, and (2) by looking at data from more than 7,000
students in classrooms known to allocate 60 (or 30) minutes of reading practice daily. Overall,
these results show the point to time conversion estimates in the Goal-Setting Chart are realistic for
students at all levels. We believe the time estimates are reasonable estimates of actual engaged time
and most importantly are especially useful for comparing relative differences of groups of students.

This method for estimating engaged reading time provides a systematic, comparable measurement
of engaged academic time, as opposed to allocated time which is less useful (Berliner, 1990).

Calculation of Effect Sizes. In some analyses, it is useful to provide effect sizes comparing the results
for two groups at different levels of implementation. The earliest and most common calculation of
effect size is Cohen’s d (Howell, 2002), which is calculated for groups of equal size by dividing the
mean difference between the two groups by their pooled variance.

M1 – M2


(n1 – 1)s12 + (n2 – 1)s22
d=
n1 + n2 – 2
where:
n = sample size of group
s2 = standard deviation of group

28
Results

Engaged Reading Time


The correlation between time spent reading and reading ability is well documented. Indeed,
the National Reading Panel (NRP) found many correlation studies. Two correlation studies are
particularly relevant to our topic and deserve mention. They not only illustrate the strong correlation
between ability and time spent reading, but also illustrate the huge variation typically seen in student
reading time between students in the top, middle, and bottom 10% of ability (Anderson, Wilson
& Fielding, 1988; Paul, 1993).

Table 6: Independent Reading Time Per Day


Variation for Fifth-Grade Students in the Anderson, et al. and Paul Studies

Standardized Anderson, et al., (n=155) Paul (n=1,112)


Test Score Minutes Ratio Minutes Ratio
Percentile Per Day (% of Bottom) Per Day (% of Bottom)
90 40.0 26.67 61.0 30.50
50 13.0 8.67 15.0 7.50
10 1.5 1.00 2.0 1.00

The Anderson, et al., and Paul studies were done at different times and with different sample sizes
but came to remarkably similar results. They show the typically huge variations in time spent reading
between the top 10% and bottom 10% of readers. In fact, students in the 90th percentile read 26–30
times more than students in the 10th percentile.

Note that the classrooms included in the Paul study were using AR without Reading Renaissance
techniques, as they had not yet been developed. Table 7 shows the difference implementation of
Reading Renaissance makes.

Table 7: Independent Reading Time Per Day


Variation for Fifth-Grade Students in the Anderson and Paul Studies
Compared to the Reading Practice Database (RPD)
Anderson, et al.,
Standardized (n=155) Paul (n=1,112) RPD (n=5,828)
Test Score Minutes Ratio (% of Minutes Ratio (% of Minutes Ratio (% of
Percentile Per Day Bottom) Per Day Bottom) Per Day Bottom)
90 40.0 26.67 61.0 30.50 46.0 1.44
50 13.0 8.67 15.0 7.50 38.0 1.19
10 1.5 1.00 2.0 1.00 32.0 1.00

With the Reading Renaissance approach, engaged time spent reading increases significantly for
average and below-average readers and the variation between students is reduced dramatically. This
is because Reading Renaissance classrooms allocate more in-school time for independent reading
and have better monitoring by teachers to assure higher academic engaged time. High-achieving
students still read the most, even in a Renaissance classroom, but low- and average-achieving
students read a lot more than they would otherwise.
29
The next table (Table 8) shows that in a Reading Renaissance classroom, it is not a student’s reading
ability that causes high amounts of reading. It also shows the significant correlation between
engaged reading time and NCE improvement in very narrow ranges of ability level. Table 8 is for
grades 2–12 combined. Grade-level analysis shows the same pattern holds for all grades. Increasing
amounts of reading time results in improvement in reading ability at all ability levels.

Table 8: NCE Change At Varying Levels of Engaged Reading Time and Beginning Ability Levels
Grades 2 through 12 (n=45,670)

Engaged Reading Time in Minutes Per Day Correlation


Percentile
Between Engaged
Range by N
10 or Reading Time and
Deciles 11–20 21–30 31–40 41–50 51–60 61–70
Fewer NCE Change1
0–10 6,114 –1.15 0.23 1.04 3.28 3.03 5.45 4.02 0.140
11–20 5,082 –0.37 0.87 2.67 3.88 5.45 5.27 5.97 0.167
21–30 4,320 –0.91 1.35 1.82 3.41 4.30 5.43 5.11 0.155
31–40 4,537 –0.74 0.30 1.81 3.20 4.05 3.46 4.56 0.144
41–50 5,049 –1.49 1.37 2.93 4.78 5.01 5.35 5.88 0.167
51–60 4,135 –0.79 1.14 1.56 3.50 3.53 4.15 5.25 0.126
61–70 3,917 –0.51 1.18 1.57 2.78 3.27 3.33 3.70 0.098
71–80 4,062 –2.71 0.12 1.47 2.67 2.51 2.82 4.82 0.156
81–90 4,453 –2.32 –0.44 0.98 2.41 3.08 2.05 4.21 0.131
91–100 4,001 –4.06 –0.19 –0.19 0.58 2.75 2.29 2.90 0.121
All Deciles 45,670 –1.20 0.63 1.59 3.06 3.71 3.90 4.59 0.137
All correlations significant at the p<0.001 level.
1

The NRP said “correlational studies find the best readers read the most and that poor readers read
the least,” and “it could be that the more that children read, the more their reading skills improve,
but it is also possible that better readers simply choose to read more” (p. 21). Table 8 shows that in
Reading Renaissance, all readers—not just the best readers—read more, and those that read more
have higher test score improvement.

Additional evidence supporting the case that it is not just better readers who read more, but that
reading more causes better readers, comes from examining the source of variation in reading time.
For the students in the RPD, we explored this question by looking at how much of the variation
in time is explained by differences in student reading ability vs. the classroom they are in. To arrive
at the answer, we first regressed engaged reading time on beginning NCE level to determine the
variance in time explained by the student’s ability or pretest score. Following are the results of
this analysis.

30
Table 9: Variance in Engaged Reading Time Explained by Beginning NCE Level
Grade N R R2 p
First 5,152 0.070 0.005 < 0.001
Second 7,696 0.089 0.008 < 0.001
Third 7,351 0.061 0.004 < 0.001
Fourth 6,849 0.162 0.026 < 0.001
Fifth 5,827 0.146 0.021 < 0.001
Sixth 4,885 0.174 0.030 < 0.001
Seventh 3,629 0.154 0.024 < 0.001
Eighth 2,694 0.215 0.046 < 0.001
Ninth 2,494 0.227 0.051 < 0.001
Tenth 1,707 0.251 0.063 < 0.001
Eleventh 1,370 0.227 0.052 < 0.001
Twelfth 1,157 0.160 0.025 < 0.001

Within each grade, the correlation between pretest score and engaged reading time is small to
moderate, ranging from 0.061 to 0.251. The R-squareds range from 0.004 to 0.063, which can
be interpreted as saying that between 0.4% and 6.3% of the variance in reading time is explained
by pretest score. The unexplained, or residual, variance ranges from 93.7% to 99.6%, depending on
the grade. The next step is to see what portion of the unexplained variance in individual student
reading time is explained by the teacher and classroom the student is assigned to. To estimate this,
we use the residuals from the first regression and create an average classroom residual for each
student excluding the residual of the student. We then regress the individual student’s residual
(unexplained variance) on the average classroom residual. The results of this analysis follows.

Table 10: Variance in Engaged Reading Time Explained by Teacher/Classroom1


Grade Student N Classroom N R R2 p
First 4,363 261 0.768 0.590 < 0.001
Second 7,389 418 0.758 0.574 < 0.001
Third 7,069 388 0.682 0.466 < 0.001
Fourth 6,505 317 0.614 0.377 < 0.001
Fifth 5,566 255 0.581 0.337 < 0.001
Sixth 4,256 200 0.517 0.267 < 0.001
Seventh 2,702 133 0.365 0.133 < 0.001
Eighth 1,978 101 0.528 0.279 < 0.001
Ninth 2,244 146 0.321 0.103 < 0.001
Tenth 1,442 76 0.422 0.178 < 0.001
Eleventh 1,161 51 0.318 0.101 < 0.001
Twelfth 931 27 0.447 0.200 < 0.001
1
Classrooms with between 12 and 30 students.

31
The R-squareds range from 0.101 in eleventh grade to 0.590 in first grade. They are substantially
higher than what is explained by beginning student ability. Clearly, to predict the amount of time
a student is engaged in independent reading, it is far more important to know something about the
student’s environment (in this case the teacher/classroom) than the student’s initial ability level. In
the case of the Renaissance approach to independent reading, students do not primarily read more
because of their reading ability. The primary cause of the time they spend reading is the teacher they
are assigned to.

Another aspect of reading time that is somewhat evident from examining Table 8 is that there seems
to be diminishing marginal returns to time spent reading. The following graph (Figure 3) shows
adjusted NCE gain at various levels of reading time and a curved line which is the best fit.

Figure 3: NCE Change by Engaged Reading Time, Grades 2 through 12


6

5
4.99
4.62
4 4.29
3.90
3.71
3 3.06
NCE Change

1.59
1

0.63
0
-1.20

-1

-2
10 or Fewer 11–20 21–30 31–40 41–50 51–60 61–70 71–80 More than 80

Engaged Reading Time in Minutes

While in general the more time spent reading the better, it is not a linear function. This pattern is
the same at all grade levels and for both above- and below-average students.

Average Percent Correct


We turn our attention next to average percent correct as a co-factor in explaining improvements
in reading achievement from independent reading. The research literature cited earlier and our
own studies indicate that average percent correct on AR Reading Practice Quizzes is a significant
predictor of test score improvement. Percent correct is also correlated to beginning NCE ability. In
this regard, potentially the same controversy surrounding time as a causative factor also applies to
percent correct. The argument would be that high-ability students also have higher percent correct
on AR quizzes, so you don’t know what caused the test score improvement, high percent correct
or high beginning test scores.

We address this question in the same manner we did for reading time. Table 11 shows reading gains
for varying levels of average percent correct among different achievement levels.
32
Table 11: NCE Change At Varying Levels of Average Percent Correct and Beginning Ability Levels
Grades 2 through 12 (n=45,670)

Average Percent Correct on Accelerated Reader Reading Practice Quizzes Correlation


Percentile
Between Percent
Range by N
Below 95– Correct and NCE
Deciles 60–69% 70–74% 75–79% 80–84% 85–89% 90–94%
60% 100% Change1
0–10 6,114 –2.43 –1.02 –1.27 0.86 1.70 3.29 5.25 0.62 0.171
11–20 5,082 –2.54 –1.57 –0.07 0.03 2.32 4.59 7.59 8.79 0.237
21–30 4,320 –3.26 –2.43 –1.28 0.18 1.72 3.69 6.53 5.76 0.236
31–40 4,537 –3.47 –2.17 –1.34 –1.19 1.09 3.02 5.84 6.89 0.232
41–50 5,049 –5.58 –2.58 –1.76 –0.24 1.47 4.01 7.15 7.58 0.263
51–60 4,135 –5.23 –1.69 –0.79 –1.73 –0.94 3.30 6.08 6.64 0.232
61–70 3,917 –1.80 –3.98 –0.25 –1.20 0.21 1.89 4.18 6.77 0.172
71–80 4,062 –6.60 –4.67 –5.23 –2.21 –0.01 1.85 3.64 6.55 0.225
81–90 4,453 –3.98 –4.63 –4.46 –3.07 –2.00 0.65 3.68 6.23 0.215
91–100 4,001 –8.27 –9.31 –4.03 –2.50 –3.21 –0.57 2.32 3.82 0.199
All Deciles 45,670 –3.37 –2.33 –1.52 –0.60 0.79 2.74 4.86 5.64 0.197
1
All correlations significant at the p<0.001 level.
33
As Table 8 demonstrates for time spent reading, Table 11 shows that percent correct is a significant
predictor of reading growth at all ability levels.
Tables 12 and 13 show that percent correct, like time, is more highly influenced by the teacher/class-
room than by beginning ability. Table 12 shows the variance in average percent correct explained by
beginning NCE level and Table 13 shows the variance in average percent correct explained by the
teacher/classroom using the same methodology as was used for time (Tables 9 and 10).
Table 12: Variance in Average Percent Correct Explained by Beginning NCE Level
Grade N R R2 p
First 5,152 0.237 0.056 < 0.001
Second 7,696 0.325 0.106 < 0.001
Third 7,351 0.339 0.115 < 0.001
Fourth 6,849 0.340 0.116 < 0.001
Fifth 5,827 0.344 0.118 < 0.001
Sixth 4,885 0.366 0.134 < 0.001
Seventh 3,629 0.317 0.100 < 0.001
Eighth 2,694 0.339 0.115 < 0.001
Ninth 2,494 0.366 0.134 < 0.001
Tenth 1,707 0.340 0.116 < 0.001
Eleventh 1,370 0.286 0.082 < 0.001
Twelfth 1,157 0.241 0.058 < 0.001

Table 13: Variance in Average Percent Correct Explained by Teacher/Classroom1 Factors


Grade Student N Classroom N R R2 p
First 4,363 261 0.620 0.384 < 0.001
Second 7,389 418 0.543 0.294 < 0.001
Third 7,069 388 0.532 0.283 < 0.001
Fourth 6,505 317 0.497 0.248 < 0.001
Fifth 5,566 255 0.551 0.303 < 0.001
Sixth 4,256 200 0.488 0.238 < 0.001
Seventh 2,702 133 0.532 0.283 < 0.001
Eighth 1,978 101 0.509 0.259 < 0.001
Ninth 2,244 146 0.304 0.092 < 0.001
Tenth 1,442 76 0.358 0.127 < 0.001
Eleventh 1,161 51 0.318 0.101 < 0.001
Twelfth 931 27 0.342 0.116 < 0.001
1
Classrooms with between 12 and 30 students.

Again, we see far more of the variance is explained by the teacher and classroom the student is
assigned to than by beginning NCE level.
The fact that the teacher/classroom determines far more gain than beginning reading ability for
both time and percent correct is very good news. If the students’ beginning ability was the key factor,
then the teacher could do very little to impact student performance. However, this analysis shows
that the teacher is key and can do a great deal to improve student performance.
When we put both time and average percent correct together, we end up with several quite remark-
able tables. These are Tables 14(a), (b), and (c). Table 14(a) is for all grades 2–12, Table 14(b), grades
2–5, and Table 14(c) grades 6–12.
34
Table 14a: Combined Impact of Engaged Reading Time and Percent Correct on Reading Improvement (NCE Change)
for Students in Grades 2 through 12 (n=45,670)
Engaged Reading Time in Minutes Correlation
Between Time
10 or 80 or and NCE
11–20 21–30 31–40 41–50 51–60 61–70 71–80
Fewer More Change
Below Mean –3.22 –4.01 –2.71 –6.92 –5.84
60% –0.024
N 1,709 273 85 23 7
Mean –2.27 –2.10 –2.33 –2.45 –3.88 –7.55 –4.97
Average Percent Correct Range

60%–69% –0.030
N 1,238 679 270 95 48 13 11
Mean –1.30 –0.79 –3.23 –1.73 –3.33 3.53 –1.22 –1.01 –6.25
70–74% –0.027
N 781 660 350 153 76 32 15 13 12
Mean 0.04 –0.03 –1.30 –1.19 –0.69 –1.88 1.61 –4.22 –3.43
75–79% –0.052**
N 749 1,107 749 437 273 149 66 35 52
Mean 0.12 0.51 0.62 1.25 1.02 1.55 0.41 1.52 0.77
80–84% 0.022
N 807 1,611 1,545 1,333 958 541 358 209 271
Mean 0.53 1.71 1.97 2.83 3.01 3.36 4.06 4.04 4.44
85–89% 0.077*
N 597 1,735 2,379 2,310 2,065 1,512 906 526 760
Mean 1.24 2.56 4.05 4.96 5.70 5.52 6.31 6.10 6.33
90–94% 0.080*
N 521 1,211 1,885 2,161 2,106 1,541 929 557 912
Mean 0.54 2.91 4.53 7.42 7.67 5.87 4.84 8.18 8.65
95–100% 0.147*
N 340 347 426 503 504 316 193 156 272

Correlation Between
Percent Correct and 0.134* 0.144* 0.168* 0.186* 0.202* 0.163* 0.161* 0.209* 0.214*
NCE Change

*Correlation significant at the p<0.001 level.


**Correlation significant at the p<0.05 level.
Cells with fewer than five students excluded.
35
36

Table 14b: Combined Impact of Engaged Reading Time and Percent Correct on Reading Improvement (NCE Change)
for Students in Grades 2 through 5 (n=27,727)
Engaged Reading Time in Minutes Correlation
Between Time
10 or 80 or and NCE
11–20 21–30 31–40 41–50 51–60 61–70 71–80
Fewer More Change
Below Mean –2.40 –2.82 –1.24 –12.76
60% –0.076
N 250 91 34 11
Mean –2.50 –2.51 0.18 –1.00 –2.83 –1.35 –0.13
Average Percent Correct Range

60%–69% 0.025
N 252 229 115 47 24 8 10
Mean –1.45 0.22 –3.31 –1.76 –0.96 2.33 –2.84 –0.25
70–74% 0.001
N 189 223 170 89 49 21 12 6
Mean –0.80 1.33 0.37 0.02 0.18 –1.29 2.33 –3.74 –1.08
75–79% –0.041
N 206 459 359 280 198 113 51 31 43
Mean –0.41 0.91 1.22 2.59 1.84 2.72 1.31 1.63 0.72
80–84% 0.021
N 260 680 886 906 768 460 307 187 231
Mean 0.58 2.75 2.61 3.77 3.61 4.27 4.67 4.84 5.19
85–89% 0.072*
N 234 837 1,485 1,682 1,708 1,274 803 452 671
Mean 2.30 3.32 5.05 6.02 6.86 6.41 6.84 6.71 6.96
90–94% 0.056
N 194 624 1,157 1,523 1,616 1,232 767 477 766
Mean 0.35 4.60 5.98 9.47 8.67 7.49 5.67 9.67 9.59
95–100% 0.113*
N 112 139 213 302 340 215 138 120 206

Correlation Between
Percent Correct and 0.119* 0.153* 0.157* 0.193* 0.199* 0.169* 0.156* 0.229* 0.217*
NCE Change

*Correlation significant at the p<0.001 level.


Cells with fewer than five students excluded.
Table 14c: Combined Impact of Engaged Reading Time and Percent Correct on Reading Improvement (NCE Change)
for Students in Grades 6 through 12 (n=17,943)

Engaged Reading Time in Minutes Correlation


Between Time
10 or More and NCE
11–20 21–30 31–40 41–50 51–60 61–70 71–80
Fewer Than 80 Change
Below Mean –3.36 –4.61 –3.70 –1.57
60% –0.016
N 1,459 182 51 12
Mean –2.22 –1.89 –4.19 –3.87 –4.94 –17.47
Average Percent Correct Range

60%–69% –0.071**
N 986 450 155 48 24 5
Mean –1.25 –1.31 –3.15 –1.69 –7.64 5.82 -12.25
70–74% –0.055**
N 592 437 180 64 27 11 6
Mean 0.35 –0.98 –2.83 –3.34 –3.00 –3.73 –0.86 –14.66
75–79% –0.123
N 543 648 390 157 75 36 15 9
Mean 0.37 0.21 –0.18 –1.59 –2.29 –5.09 –5.02 0.59 1.03
80–84% –0.060*
N 547 931 659 427 190 81 51 22 40
Mean 0.49 0.74 0.89 0.31 0.11 –1.51 –0.73 –0.85 –1.27
85–89% –0.038**
N 363 898.00 894 628 357 238 103 74 89
Mean 0.60 1.75 2.45 2.43 1.88 1.97 3.79 2.47 3.03
90–94% 0.032
N 327 587 728 638 490 309 162 80 146
Mean 0.63 1.77 3.09 4.34 5.60 2.44 2.77 3.19 5.73
95–100% 0.105*
N 228 208 213 201 164 101 55 36 66

Correlation Between
Percent Correct and 0.136* 0.126* 0.166* 0.164* 0.227* 0.185* 0.239* 0.179* 0.239*
NCE Change

*Correlation significant at the p<0.001 level.


**Correlation significant at the p<0.05 level.
Cells with fewer than five students excluded.
37
There are several things remarkable and perhaps even surprising about Tables 14a–c. First are the
high gains in reading achieved for students with both high average percent correct and high amounts
of reading practice time. Second, these tables demonstrate that unless students are attentive readers
with high levels of comprehension and achieve high levels of success on AR quizzes, they do not
benefit from increased reading practice. In fact, at low levels of percent correct, more time spent
reading may actually cause reading achievement to decline relative to a student’s peers. It’s not
practice which makes perfect, but successful practice which develops successful readers.

Tables 14a–c show why it is important to have systematic quantitative information feedback
on students for an effective independent reading program. Tables 14a–c also confirm what we
teach in Reading Renaissance: quality before quantity—and guided independent reading with
teachers watching over the AR information closely, intervening and instructing as appropriate
vs. unguided reading.

Optimum Student Goals for Percent Correct and ZPD

The factors that impact reading development are incredibly complex, dynamic, and non-linear.
In addition, tools for measuring reading ability are anything but precise. Large data sets such as
the RPD help to filter out noise caused by measurement error, but do not begin to approach the
precision necessary. In addition, at the individual student level, we are no longer dealing with
statistical averages but with complex individuals. Given this, there is temptation to throw up
one’s hands and say nothing, taking refuge in the rule, “the first rule is to do no harm.”

Yet the data and research literature tell us something of value. Inevitably, offering some suggestions,
some rules of thumb, based on a reasonable reading of the evidence, wins out, and so we do here.

The 92% Recommendation

In Reading Renaissance, the recommendation has been that teachers consider guiding students
to more challenging books when their average percent correct exceeds 92%. Unlike the 85% rule
though, this has never been a “hard” recommendation as it has been clear many students score
higher on their AR quizzes almost without regard to text difficulty. Regardless, an examination
of the RPD quickly disposes of even a “soft” 92% recommendation. Clearly, as seen by looking at
Table 14(a), significant growth continues to occur above 92%. A closer examination shows that for
all grades, students with average percent correct up to 97% continue to make gains.

The 85% Rule

We next look at the recommendation that 85% is the minimum level for average percent correct
on AR quizzes students need to achieve growth in reading. Unlike the 92% recommendation, 85%
minimum has been treated as more of a hard and fast rule. It has been the Reading Renaissance
recommendation for almost 10 years. The Topping and Sanders study (2000) confirmed that an
average of 85% correct was the level at which literature-based reading practice began having
a positive effect. The RPD yields a more complex answer.

38
What we find is the minimum percent-correct goal depends on a variety of factors, most
importantly grade level and ability level.

Specifically, the data show that students in grades K–3 who are “learning to read” will achieve
maximum growth when average percent correct on AR quizzes is 85% or above. However, students
in grades 4–12 who are “reading to learn” should strive for even higher comprehension rates (90%
or above on AR quizzes). The recommendation of 90% for fourth grade and above varies somewhat
by achievement level.

Tables 14a–c contain data on students spanning a broad range of developmental reading abilities.
To provide more precision, Table 15 looks at students in grades 1–8 and compares the percentage
of students who average 85% or greater on AR quizzes and show gains in reading to the percent
of students who average 90% or greater on AR quizzes and show gains in reading.

Table 15: Percent of Students with Positive NCE Gain


Above 85 and 90 Average Percent Correct1
Students above 85% Correct Students above 90% Correct
Grade Percent with Average NCE Percent with Average NCE
Positive Gain Change Positive Gain Change
First 86.7 16.37 87.6 16.80
Second 74.2 8.71 78.4 10.30
Third 65.6 4.68 68.1 5.44
Fourth 65.0 4.38 68.1 5.33
Fifth 57.7 2.28 63.5 3.51
Sixth 51.0 0.48 55.8 1.75
Seventh 58.1 2.45 60.5 3.15
Eighth 55.1 1.76 56.3 2.29
1
Includes only students with an average of less than 98% correct and an engaged practice time of at least 20 minutes a day,
as students with less time spent reading may not gain due to low reading time vs. low percent.

It would be impossible to expect 100% of students to show positive gain with any program.
Even 60% achieving positive gain would be acceptable in many situations. However, 50% having
gain would clearly be unacceptable in most situations, as this would mean that as many students
are helped as not by the treatment.

Consistent with the research on the stages of reading development (Chall, 1996) this figure
shows that students in the upper grades need higher percent correct (90%) to experience greater
gains in reading.

39
Tables 16a–b take this analysis one step further to show that percent-correct recommendations
are different for high- and low-achieving students.

Table 16a: Percent of High-Achieving Students with Positive NCE Gain


Above 85 and Above 90 Average Percent Correct1
Students above 85% Correct Students above 90% Correct
Grade Average NCE Percent with Average NCE Percent with Average NCE
Positive Gain Change Positive Gain Change
First 67.05 83.3 11.44 85.3 12.38
Second 69.08 74.8 8.28 79.0 9.58
Third 65.59 58.8 2.67 63.2 3.71
Fourth 65.19 64.6 4.06 68.4 5.09
Fifth 63.62 58.9 2.26 63.0 3.18
Sixth 63.36 51.5 0.34 56.6 1.49
Seventh 63.23 59.8 2.57 62.9 3.34
Eighth 64.67 54.7 1.28 54.9 1.89
1
Includes only students with an average of less than 98% correct and an engaged practice time of at least 20 minutes a day, as students
with less time spent reading may not gain due to low reading time vs. low percent.

Table 16b: Percent of Low-Achieving Students with Positive NCE Gain


Above 85 and Above 90 Average Percent Correct1

Students above 85% Correct Students above 90% Correct


Grade Average NCE Percent with Average NCE Percent with Average NCE
Positive Gain Change Positive Gain Change
First 37.34 91.3 22.84 91.2 23.75
Second 37.97 73.3 9.31 77.1 11.63
Third 34.72 75.2 7.52 78.0 8.89
Fourth 34.49 65.6 4.95 67.3 5.95
Fifth 33.41 55.5 2.31 64.6 4.39
Sixth 31.21 50.0 0.74 53.6 2.47
Seventh 31.69 55.2 2.24 55.2 2.73
Eighth 31.41 55.9 2.66 59.6 3.21
1
Includes only students with an average of less than 98% correct and an engaged practice time of at least 20 minutes a day, as students
with less time spent reading may not gain due to low reading time vs. low percent.

Tables 16a–b show that there is a significant difference between low- and high-achieving students
at the third- and fourth-grade level. In third grade, more low-achieving students make gains in
reading when they average 85% correct on AR quizzes than do high-achieving students who need
to average 90% correct or higher. This suggests that high-achieving students have already learned
to read and are transitioning to reading to learn, which is more difficult. At the same time,

40
low-achieving students are still showing gains in second and third grades at 85%, suggesting that
these students are still learning to read.

So, while the research literature focuses on the “fourth-grade slump” that happens with low-
achieving students, there is evidence here that high-achieving students have a slump too, but
it happens in third grade.

Among third-graders with quiz scores averaging at least 85% correct, 75% of the low-achieving
students achieved NCE gains, compared to just 59% of the high-achieving students. This analysis
suggests that the recommended minimum percent should be higher than 85% for high-achieving
third-graders and low-achieving fourth-graders and higher than 85% for all students fifth grade
and up.

Appropriate Book Level

The analysis of appropriate book level is perhaps the most complex. Although readability formulas
measure the level of the book a student reads, they do not tell us anything about the student’s moti-
vation or interest.

An equally difficult problem has to do with measurement error. To match a book to a student, two
measurements are needed: (1) the reading ability of the student, and (2) the readability of the text.
Neither of these are 100% accurate because every assessment has some measurement error. For
example, even high-quality standardized reading tests can have ± one grade level measurement error.

And what about the readability estimate for books? At Renaissance Learning, we think we do
a better job of estimating readability level than anyone. First, we use the ATOS readability formula
that was especially designed for measuring book readability (see Appendix A on page 67). Second,
we scan the entire book (not just three 100-word samples, which is more typical). ATOS considers
the reading level of the words in the book, the word length, sentence length, and the book length.
While ATOS is one of the most accurate readability formulas available, all existing readability
formulas have a fundamental problem. This is illustrated by the following two passages.

“Four score and seven years ago our fathers brought forth upon this continent a new nation,
conceived in liberty and dedicated to the proposition that all men are created equal. Now we
are engaged in a great civil war, testing whether that nation or any nation so conceived and
so dedicated can long endure.”

“Endure long can dedicated so and conceived so nation any or nation that whether testing,
war civil great a in engaged are we now. Equal created are men all that proposition the to
dedicated and liberty in conceived, nation new a continent this upon forth brought fathers
our ago years seven and score four.”

The first passage is very readable; the second completely unreadable. The first passage is of course
the first paragraph of the Gettysburg address. The second passage is the Gettysburg address in exact
reverse word order. Yet all readability formulas would score the two passages exactly equal in terms
of reading difficulty!

41
The problem with readability formulas is that they do not take into account many of the variables
that affect a person’s ability to comprehend text, such as: Does the story make sense? Is it a good
story? Is it well written? Reading formulas are indifferent to all these questions and therefore are at
best only decent estimates of readability. At worst, as shown, they can considerably underestimate
the difficulty of a text.

So at the very outset, there is a question as to just how well we might be able to match books
to students in the best of circumstances. In spite of these difficulties, the large amount of data
provided by the RPD enables us to explore this further.

To start our analysis of the effect of challenge on reading outcomes, we calculated the expected
average book level for every student within grade by regressing average book level against their
pretest NCE score. Next, we created a challenge index for each student which is the ratio of their
actual average book level to their expected average book level based on their pretest score. The
scatter plot in Figure 4 shows the relationship between the challenge index and reading growth.

Figure 4: Scatter Plot of NCE Change as a Function of Challenge Index


Students in Grades 3 through 12 with at least 20 Minutes of
Daily Engaged Reading Time
80

60

40
Adjusted NCE Gain

20

-20

-40

-60

-80
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Challenge Index

42
Topping and Sanders (2000) found book level to be negatively related to reading improvement—
but concluded it was an anomaly and that teachers should continue to encourage students to read
progressively harder books. Consistent with Topping and Sanders (2000), we found that the simple
linear regression was negative. However, we then discovered that the quadratic regression had a
significantly better fit and has a shape very similar to Figure 1 on page 13. The quadratic regression
formula is NCE Change = –21.3 + 51.6 (Challenge Index) – 27.8 (Challenge Index2). As we would
expect, students who are at either a very high challenge level or a very low challenge level make less
gains in reading than students who are appropriately challenged. So the concept of an optimal range,
a ZPD range, is confirmed.

However, the challenge index explains only a small amount of students’ reading gains (R2=0.002).
Also, as shown in Figure 4, the quadratic curve is quite flat at its peak. This tells us that students can
read a very broad range of books, both easy and very hard books, and still have successful reading
experiences and achieve significant gains in reading. In fact, although teachers often worry that
students, particularly high-achieving students, do not read challenging books, the peak value on the
plot of the quadratic curve is attained when students read books slightly below their current average
at a challenge index of 0.932. This suggests, on average, students are being slightly overchallenged.
We did this analysis for both high- and low-achieving students, and in both cases, students were
being slightly overchallenged.

However, given the broad book reading ranges possible without impacting gain, the slight overchal-
lenge is not significant. Students in the RPD are reading well within the ZPD ranges we currently
recommend. Therefore, there is no indication our Goal-Setting Chart needs to be changed or modified.

There is good and bad news here. The bad news is that we can only justify very general guidelines
and broad readability ranges in which to place students in appropriate books. Furthermore, it
turns out that a better tool for monitoring the appropriateness of a book is average percent
correct on AR Reading Practice Quizzes.

The good news is that reading improves by reading a wide range of books of varying levels of
difficulty. That’s the advantage of literature-based reading with trade books. Trade books often
contain vocabulary that is more advanced than the book level (i.e., third-grade book with some
sixth- to seventh-grade vocabulary words), and amount of exposure to new words is critical to
learning those words (Anderson, 1996; Nagy & Herman, 1987). Furthermore, words encountered
within a story have a greater chance of being remembered (Anderson, 1996).

This analysis indicates that teachers can and should allow students a fair amount of freedom to
choose their books and follow their interests. When a student really gets into a book, it’s hard not
to do well on the AR quiz. Therefore, success in quiz-taking does not just provide teachers informa-
tion about comprehension. It also provides information about enjoyment.

In the end, teachers and students will benefit from allowing students to read a broad range of
books. Teachers should focus on what really makes a difference: directing students to books that
are engaging, talking to students about what they read, teaching them strategies to become attentive,
meta-cognitive readers, and providing instruction as necessary to hone their decoding, comprehen-
sion, fluency, and vocabulary skills.

43
Effect Size and Expected Improvement for High-Fidelity Implementation

In Reading Renaissance, fidelity of implementation is measured by the Implementation Index,


which is the percentage of students in a class or grade averaging above 85 percent correct on AR
quizzes multiplied by the percent of 60 minutes the class or grade has achieved for a period of time
(as measured by AR points). The Implementation Index provides a reasonably objective and compa-
rable estimate of the level of implementation and is predictive of achievement gains. A 100% imple-
mentation would consist of all students in a class or grade at or above 85% correct and reading an
average of 60 minutes per day.

A more accurate index will likely be developed now with the availability of the RPD, but the
current index is attractive as it is easy to understand and apply. We continue to use it here.

Reported effect size for improvement of reading scores from implementation of Reading
Renaissance at the school level are 0.50 to 0.57 (Holmes & Brown, 2003; Sadusky & Brem,
2002). We examine here effect sizes at the classroom level by comparing the average adjusted NCE
improvement in Model- or Master-Certified classrooms7, to non-certified classrooms, and to non-
implementing classrooms which we define as those classrooms which averaged less than 10 minutes
of engaged reading time.

7
Model- and Master-Certified classrooms are classrooms led by educators who have met clear, objective Renaissance
implementation standards.

44
Table 17: Effect Sizes for NCE Differences
Certified vs. Non-Certified and Non-Implementing Classrooms1
Certified Non-Certified Effect Size Non-Implementing Effect Size for
Mean Mean for NCE difference— Mean NCE difference—
Grade2
N Implementation N Implementation Certified vs. N Implementation Certified vs. Non-
Index Index Non-Certified Index Implementing

First 161 0.77 86 0.45 0.00 5 0.06 0.52


Second 247 0.76 157 0.39 0.29 7 0.05 1.25
Third 247 0.67 129 0.34 0.46 10 0.04 0.86
Fourth 164 0.55 139 0.33 0.26 16 0.04 0.91
Fifth 138 0.59 116 0.28 0.53 9 0.03 1.34
Sixth 72 0.56 120 0.22 0.27 8 0.01 1.38
Seventh 12 0.46 119 0.27 0.19 3 0.01 1.10
Eighth 13 0.55 77 0.27 0.18 3 0.06 –0.40
Second–
893 0.65 857 0.31 0.49 56 0.03 1.04
Eighth
1
For classrooms with between 12 and 30 students.
2
Grades with fewer than five certified classrooms excluded.
45
Table 17 shows Reading Renaissance-certified classrooms outperformed non-certified and
non-implementing classrooms by wide margins. For grades 2–8, the effect size for certified versus
non-certified is 0.49, and for certified versus non-implementing is 1.04. Only in the eighth grade
did the certified classrooms achieve lower than the non-implementing classrooms, and this was
a relatively small sample size. (As mentioned in the Methodology section, both the number of
students and Model Classrooms drops off significantly after the seventh grade.)

Remarkably, these high effect sizes favoring certified classrooms were achieved even though the
average Implementation Index for certified classrooms was only 0.65. (Full or 100% implementation
would be an index score of 1.00.) A classroom is certified as Model or Master based on 9- or 12-
week results, whereas the index score reported in Table 17 is for the implementation level between
the pre- and post-test or on average eight months.

Table 18 shows implementation levels by grade for students in the RPD.

Table 18: Reading Renaissance Implementation Level by Grade


Actual Average
Percent Percent of Implementation
Grade Adjusted Percent
Above 85% 60 Minutes Index
NCE Gain Correct
First 15.10 88.67 80.28 74.02 0.59
Second 6.74 87.44 74.26 77.59 0.58
Third 3.60 86.53 69.11 72.80 0.50
Fourth 2.92 85.49 63.21 62.65 0.40
Fifth 0.70 85.12 62.51 64.13 0.40
Sixth –1.05 83.55 55.77 53.07 0.30
Seventh 0.18 82.53 52.62 44.36 0.23
Eighth –0.06 82.27 54.25 45.21 0.25
Ninth–
0.11 74.55 31.60 23.25 0.07
Twelfth

As you can see, the level of implementation falls as grade level increases. By the sixth grade, imple-
mentation is only at 30% and in 9–12, only 0.07. Both the percent above 85% and minutes spent
reading tends to drop off by grade six.

What might the expected NCE growth be for 100% implementation as currently defined? This is
shown below in Table 19, based on analysis of students in the RPD scoring above 85% correct and
averaging 60 minutes of engaged reading time per day within each grade.

46
Table 19: Estimated Impact of 100% Implementation of Reading Renaissance
Actual Average Adjusted NCE Average Percent
Grade Adjusted Percent Gain for 100% Correct for 100%
NCE Gain Correct Implementation Implementation
First 15.10 88.67 15.93 91.47
Second 6.74 87.44 8.98 90.26
Third 3.60 86.53 4.94 89.95
Fourth 2.92 85.49 4.87 90.24
Fifth 0.70 85.12 2.46 90.42
Sixth –1.05 83.55 0.81 91.05
Seventh 0.18 82.53 2.57 91.18
Eighth –0.06 82.27 1.61 91.17
Ninth–
0.11 74.55 4.70 91.56
Twelfth

But how likely is a 100% implementation?

At the elementary grades, we see levels of 90–100% implementation fairly often. However, high
levels of implementation are very rare at the middle school level and non-existent so far at the high
school level. The reason for this is that no high school has felt reading practice was important
enough to allocate 60 minutes, and only a few middle and junior high schools have done so. This is
why we have recently changed our recommendation of 60 minutes to 45 minutes and 30 minutes for
middle and high school respectively.

This analysis is interesting in several respects. First, it shows the very high annual gains possible
with good implementation. In grade 1–4, annual gains are extraordinarily high. It suggests the
toughest years to achieve gains are the middle grades, 5–8, where predicted gain for 100% imple-
mentation is 1.86 NCE. Predicted gain for 100% implementation for grades 9–12 are at 4.7 NCE.
On the other hand, in looking at the effect size in Table 17 which measures relative vs. absolute
gains, some of the highest relative gains are in the fifth, sixth, and seventh grades.

47
Differential Effects

High- vs. Low-Achieving Students

An additional question we looked at is who would benefit most from a high-fidelity implementation
(as measured by the current implementation index), low-, average-, or high-achieving students?
Table 20 provides an estimate.

Table 20: Comparison of NCE Gain for High- and Low-Achieving Students at Current
Implementation Levels and with 100% Implementation
Current Implementation 100% Implementation
Grade Above Below Above Below
Median Median Median Median
First 10.46 19.87 11.33 22.42
Second 7.06 6.41 8.35 9.91
Third 1.79 5.36 2.87 7.77
Fourth 2.76 3.09 4.54 5.53
Fifth 0.95 0.43 2.41 2.55
Sixth –0.89 –1.19 0.51 1.33
Seventh 0.65 –0.23 3.00 1.91
Eighth –0.23 0.10 1.47 1.91
Ninth–Twelfth 0.12 0.11 5.10 3.96

In general, Table 20 shows that the low-achieving student is likely to improve the most with a high-
fidelity implementation. The advantage for the low-achieving student is especially significant in the
third grade. However, the advantage for low-achieving students seems to disappear after the sixth
grade. This is an important finding which is covered further in the discussion section.

48
Gender

Shown below in Table 21 is a comparison of boys and girls by grade for various reading statistics.

Table 21: Summary Statistics for Boys and Girls


Average Daily Engaged
Average Adjusted
Grade Percent Time Spent Percent Fiction
NCE NCE Gain
Correct Reading
Male 51.43 14.81 88.22 43.60 81.35
First
Female 54.47 15.19 88.98 45.12 85.56
Male 52.36 7.36 86.93 45.74 85.65
Second
Female 54.64 6.36 88.12 46.14 89.70
Male 48.67 4.05 85.91 42.42 88.09
Third
Female 50.81 2.87 87.31 43.31 91.69
Male 50.25 3.49 84.92 36.50 89.03
Fourth
Female 50.87 2.75 86.41 38.60 93.05
Male 48.44 1.01 84.39 37.91 89.10
Fifth
Female 49.09 0.61 86.19 38.34 92.59
Male 45.34 –0.86 82.53 29.66 91.02
Sixth
Female 46.84 –0.92 84.34 33.61 95.04
Male 45.49 –0.39 80.74 25.66 89.32
Seventh
Female 45.36 0.18 83.15 28.14 94.09
Male 45.94 –0.23 81.21 26.45 89.05
Eighth
Female 46.63 0.27 83.93 29.11 92.57
Male 44.48 –0.31 73.12 13.51 81.63
Ninth
Female 45.28 0.62 77.17 16.83 89.99
Male 48.76 0.17 72.81 12.25 80.31
Tenth
Female 48.96 0.91 77.12 16.30 91.60
Male 50.29 1.03 71.79 10.63 84.19
Eleventh
Female 51.08 1.41 76.22 13.97 91.66
Male 50.00 –2.76 72.45 12.00 85.65
Twelfth
Female 50.95 –0.68 76.21 13.27 90.05

49
As shown in Table 21, girls not only read more than boys, they read differently, scoring higher on
their AR quizzes and also reading more fiction. Boys score significantly lower than girls on stan-
dardized measures of reading achievement. These results provide several reasons as to why this
might be.

In spite of these differences in reading behaviors, however, when we looked at the impact of guided
independent reading for similar levels of percent correct and time, we found both boys and girls
achieved similar NCE gains.

Race or Ethnicity

As with gender, the effects of guided independent reading practice with AR and Reading Renaissance
do not appear to differ by race or ethnic background. In general, African American and Hispanic stu-
dents in the RPD tend to begin with lower pretest scores. However, their gain in reading ability is
comparable to the gain experienced by students who are white at the same achievement level with
similar engaged reading time and percent correct.

Nonfiction Reading

Almost everyone agrees students should do a significant amount of nonfiction reading. The National
Assessment of Educational Progress (NAEP) reports that students with access to a broader range
of reading materials in the home score higher on standardized assessments (U.S. Department of
Education, 2001). Also, nonfiction texts make up a large part of most state and national standard-
ized tests (Dreher, 1998; Parkes, 2001). If students are not reading nonfiction in their daily
classroom reading, they may not perform as well on standardized reading comprehension
passages or on tests in subjects other than reading or language arts (Snow, 2002).

Despite these apparent benefits for reading nonfiction, there has not been much, if any,
experimental research done specifically on the effect of nonfiction reading on reading achievement.
Reading Renaissance currently recommends that students read approximately 20% nonfiction,
however recent analyses of the data available in the RPD suggest that this recommendation should
be made with caution.

Data from the RPD show that reading high amounts of nonfiction may have deleterious effects on
student reading growth. Seventeen percent of the students read 20% or more nonfiction; however,
those seventeen percent gained significantly less on STAR Reading from pre- to post-test (t= –6.33; p
≤ 0.001). Nonfiction reading is negatively correlated to reading growth across all grades (grades 2–12
combined, r = –0.020, p < 0.001). Nonfiction reading is also negatively correlated to percent correct
(grades 2–12, r = –0.135, p < 0.001) and engaged reading time (grades 2–12, r = –0.055, p < 0.001).

These are startling findings to say the least—only made slightly less startling by the fact that the
negative correlation between percent nonfiction reading and NCE gain is moderate to small. On
the other hand, the amount of nonfiction reading students do is also small at about 10%.

The question is why does nonfiction reading seem to have negative effects? Are the books too hard?
Do students not know how to read nonfiction? Is there something teachers can do about it? The
next set of analyses helps us get closer to being able to answer these questions.

50
Tables 22(a), (b), and (c) compare nonfiction to fiction book reading by grade. Separate comparisons
are also made for males and females.

Table 22a: Fiction to Nonfiction Comparison—All Students


Nonfiction Nonfiction Fiction
Percent Fiction Average
Grade Percent Average Percent
Nonfiction Book Level
Correct Book Level Correct
First 16.77 87.50 1.8 88.71 1.9
Second 13.16 83.04 2.7 88.12 2.6
Third 10.61 80.74 3.3 87.52 3.2
Fourth 8.98 79.81 3.9 86.53 3.8
Fifth 9.03 80.38 4.4 86.04 4.2
Sixth 7.14 79.03 4.7 84.32 4.6
Seventh 8.64 78.77 5.2 83.00 4.9
Eighth 9.19 79.94 5.5 82.88 5.2
Ninth 13.71 72.39 5.6 74.59 5.4
Tenth 13.86 71.73 6.1 73.39 5.6
Eleventh 11.46 68.07 5.4 72.39 5.7
Twelfth 11.95 69.81 5.9 72.52 6.4

Table 22b: Fiction to Nonfiction Comparison—Males


Nonfiction Nonfiction Fiction
Percent Fiction Average
Grade Percent Average Percent
Nonfiction Book Level
Correct Book Level Correct
First 18.65 87.19 1.9 88.27 1.9
Second 14.35 82.51 2.7 87.72 2.6
Third 11.91 80.26 3.3 87.07 3.2
Fourth 10.97 79.65 3.9 86.15 3.8
Fifth 10.90 80.16 4.4 85.41 4.2
Sixth 8.98 78.32 4.7 83.38 4.6
Seventh 10.68 76.67 5.1 81.28 4.8
Eighth 10.95 78.51 5.4 81.89 5.2
Ninth 18.37 72.01 5.7 71.92 5.4
Tenth 19.69 70.27 6.1 70.23 5.7
Eleventh 15.81 64.95 5.4 69.21 5.6
Twelfth 14.35 67.29 5.8 70.39 6.5

51
Table 22c: Fiction to Nonfiction Comparison—Females
Nonfiction Nonfiction Fiction
Percent Fiction Average
Grade Percent Average Percent
Nonfiction Book Level
Correct Book Level Correct
First 14.44 87.64 1.8 89.04 1.9
Second 10.30 83.28 2.7 88.71 2.7
Third 8.31 81.02 3.3 88.16 3.3
Fourth 6.95 80.57 3.9 87.25 3.9
Fifth 7.41 81.28 4.4 86.97 4.2
Sixth 4.96 79.73 4.8 85.07 4.7
Seventh 5.91 79.11 5.2 83.57 4.9
Eighth 7.43 82.12 5.7 84.42 5.2
Ninth 10.01 73.16 5.7 77.52 5.4
Tenth 8.40 74.12 6.2 77.11 5.7
Eleventh 8.34 70.90 5.6 76.11 5.7
Twelfth 9.95 74.32 6.2 75.69 6.5

What’s interesting is that nonfiction book levels are about the same as fiction through the sixth grade.
At the seventh through tenth grades, nonfiction book levels are higher than fiction. More surprising
though is that the difference between average percent correct for nonfiction reading vs. fiction
reading actually narrows and in the case of boys, nonfiction percent correct is slightly higher in the
ninth and tenth grades than fiction. In grades 3–5 though, average percent correct is 6% higher on
fiction books than nonfiction while the books are about at the same level of difficulty. This supports
what many commentators say about nonfiction reading—in the early grades, students just don’t know
how to read nonfiction. Nonfiction is more difficult to read until you learn it requires a different approach.

If this supposition were true, one would also expect that effective teachers could help overcome the
initial disadvantage of nonfiction reading by providing instruction in nonfiction reading methods.
Table 23 provides some support for this position.

Table 23: Descriptives by Classroom Quintile Sorted by


Percent Correct on Nonfiction Books Grades 4 through 6
Average Percent Average Percent
Adjusted NCE
Classroom Quintile Correct— Correct—
Change
Fiction Only Nonfiction Only
Lowest 79.23 68.60 –0.26
Fourth 85.37 77.35 1.21
Middle 87.39 80.65 1.29
Second 88.55 83.02 1.35
Highest 88.55 86.88 2.02

Table 23 shows that when classrooms are ranked by average percent correct on nonfiction, low-
quintile classrooms have both lower average percent correct overall and an almost 11-percent
difference between fiction and nonfiction average percent correct. The high-quintile classroom,
on the other hand, has both high percent correct overall and less than 2% difference between
average percent correct for fiction and nonfiction reading.

52
Discussion

Causation

The National Reading Panel concluded there was not sufficient evidence to support the idea that
independent reading practice causes improved reading achievement because the studies reviewed by
the Panel were correlational and not experimental.

However, when one looks at the broader research literature both within and outside the U.S., there
are many studies supporting causation, including 18 independent quasi-experimental studies on
Reading Renaissance. In addition, a recent major international study of reading, the PISA study,
concluded that independent reading was a primary factor explaining differences in the reading
achievement of 15-year-old students between countries and within countries (Kirsch et al., 2002).

Yet, as we indicated, even Anderson (1996), who co-authored the 1985 report Becoming a Nation
of Readers, which heavily endorsed independent reading, found in later research that the whole
language-type of independent reading without systematic feedback garnered inconsistent results.
While there is a large amount of experimental research showing the importance of feedback in
general, the only experimental study we are aware of that specifically addresses feedback with
respect to independent reading was done by Samuels et al. (2003a) which showed feedback had
a significant positive effect on student achievement.

This study of the Reading Practice Databases provides causative evidence supporting the benefits
of guided independent reading with feedback. Additionally, this study shows why having students
“just read” is unlikely to be effective. Indeed, some students who spent a lot of time reading without
high levels of comprehension actually lost ground relative to their peers—and these were the very
students who could not afford to lose ground.

This study shows it isn’t practice that makes perfect, but successful guided independent reading
practice that improves reading achievement. Engaged time, systematic feedback, guidance,
instruction, and high levels of success must come together. When they do, more reading causes
better reading, regardless of a student’s beginning reading ability.

Time

The Reading Renaissance recommendation is that 60 minutes of in-school time be allocated to


guided independent reading. Sixty minutes is more than most researchers have recommended.
Becoming a Nation of Readers recommended 25 minutes per day, though others recommend up
to 90 minutes per day (Allington, 2001).

We show there are diminishing returns to engaged reading time. This is certainly expected. The law
of diminishing returns applies to practically everything, and in this case can be explained by many
factors including boredom and ability of teachers to effectively manage and motivate engaged read-
ing time beyond some point.

53
All in all, the RPD data support our recommendation of 60 minutes of allocated time for
guided independent reading for all grades, 1–12 (except of course for children not yet reading
independently). Realistically, if you don’t allocate 60 minutes, you are unlikely to achieve 45 minutes
of engaged time. Even though it is harder for middle and high schools to allocate 60 minutes, the
gains in reading achievement for guided independent reading at both the middle and high school
are substantial, and therefore allocating 60 minutes is well supported. Very few treatments are likely
to be as effective in raising test scores as guided independent reading at all levels, elementary, mid-
dle, and high school.

Fourth-Grade Slump

The statistical examination of the RPD helped confirm and uncover important aspects of the aver-
age percent correct metric. We learned that a minimum goal of 85% correct stands up pretty well as
a general rule in the early grades. Students in grades one through three will do well with this goal.
Indeed, we found that for first- and second-graders who are learning decoding skills and improving
fluency, an average of 80% correct or higher on AR quizzes will lead to reading growth. However,
these young readers typically average above 90% anyway, and therefore there is no need to drop the
minimum below 85% in grades one through three.

We paid particular attention to the transition stage between “learning to read” and “reading to learn”
which usually happens around third or fourth grade. We found average percent correct on AR
quizzes should be higher for students who have progressed into the “reading to learn” stage. At the
“reading to learn” stage, decoding skills become automatic and the focus switches to learning new
vocabulary from context. At this stage, students encounter more and more words that are not in
their listening vocabulary. The data indicates that this dip occurs in third grade for high-achieving
students. For low-achieving students, the dip occurs in the fourth grade, confirming the classic
fourth-grade slump. We discovered that the significant year-to-year dip in the percent of students
achieving gains in reading likely indicated that year of transition from “learning to read” to “reading
to learn.”

The RPD data show that when students make the transition to reading to learn, they need
to comprehend at higher levels (demonstrated by an average of 90% or above on AR quizzes) to
achieve consistent gains. However, we found that during this transition, average percent correct
scores actually drop below 90%. Furthermore, high school students average below 80%.

This drop in percent correct is likely the result of two factors. First, reading does in fact get more
difficult as one’s reading vocabulary catches up and eventually surpasses one’s listening vocabulary.
Second, there is a misperception that independent readers, especially older independent readers, do
not need to be monitored and guided in their reading as much as younger readers. From extensive
classroom observations, we have found that teachers tend to do less monitoring, active intervention,
and direct instruction as the grade level goes up. While teachers trained in Renaissance techniques
monitor much more than those not trained, the pattern is the same.

The good news is that the RPD shows teachers have a great deal of influence over how well students
do on their AR quizzes. The variance explained by the teacher and classroom environment is very
high. In other words, teachers make a huge difference. In fact, a student’s reading growth is more
related to the impact of the teacher than to the student’s beginning reading level.
54
While we typically think the “learning-to-read years” (first through third grades) are the
most difficult, the drop in growth rates after third or fourth grade suggests that “reading to learn”
is more challenging.

Average Percent Correct

Reading Renaissance has always recommended that teachers increase a student’s book level when
their average percent correct on AR quizzes reaches 92% correct or above. The current data indicate
that this recommendation was too low. On average, the greatest gains in reading achievement occur
when students averaged 93–96% correct on AR quizzes. While gains drop off significantly once
students average 97% or above, we did not find much evidence in the data that students averaging
above 97% were underchallenged. However, we still believe it is likely a significant number of them
are underchallenged and reading books that are too easy. Another likely problem with students
scoring above 97% is they may be receiving too much “help.” Some students consistently achieved
100% on books that were two to three grade levels above where they might be expected to read.

Most importantly, we found that 85% and above is a reasonable goal for students who are learning
to read. However, for students who have progressed to the “reading to learn” stage, which is around
a fourth-grade reading level, 90% on AR quizzes is a more productive goal.

ZPD and Book Level

The predicted shape of the ZPD curve (Figure 1) was confirmed. In general, students need to read
books below their frustration level and above the level at which no new vocabulary is introduced.

However, there is no precise way to measure text difficulty for a particular student. The student’s
interest and motivation are major factors. There is also considerable error inherent in measuring
student reading ability and the readability level of books. Therefore, it was no surprise that book
level is not closely related to reading gains. All in all, considering the difficulties of measurement,
the recommended ZPD ranges we currently publish are reasonable. Teachers can use them as
a guide and initial estimate for placement purposes. After the initial placement, teachers must adjust
each student’s book level up or down based on student success and motivation.

Because the data indicate that gains in reading are related to reading a fairly broad range of book
levels, students should be allowed to select from a wide range of books. Giving students choices
enhances enjoyment, interest, and motivation. Choice also greatly increases the probability of
successful reading practice. This is one of the great advantages of literature-based reading:
with some freedom to choose, students are likely to select books that are interesting and
intrinsically motivating.

55
Gender and Race

Boys do not just read less, they read differently than girls. Boys read more nonfiction books than
girls and score lower than girls on AR quizzes. The differences in reading behavior between boys
and girls may partly explain why girls consistently outperform boys on standardized reading tests.
However, there did not seem to be a difference in the treatment effect of guided independent read-
ing. This means a good implementation of Reading Renaissance can help close the gender gap.

With respect to race, except for the fact that the African American and Hispanic students
in the RPD had lower average reading scores, the impact of guided independent reading was very
comparable to white children of the same ability level.

High- vs. Low-Achieving Students

Data from the RPD indicate that low-achieving students gained as much as high-achieving
students with current levels of implementation. However, when the quality of the implementation
is very high, low-achieving students can close the gap. This is good. It means guided independent
reading can help close the achievement gap between high- and low-achieving students.

However, Table 20 on page 48 shows that while the gap between low- and high-achieving students
can be narrowed in grades 2–4, in grades 5 and above, both high- and low-achieving students
demonstrate similar growth patterns from guided independent reading. One possible reason for this
is that even when we adjusted for both percent correct and time, high-achieving students still read
more than low-achieving students, on average. The problem with this explanation is that high-
achieving students are likely to read more than low-achieving students even at the earlier grades.

The most likely explanation for decrease in rate of gain for low-achieving students concerns the
discrepancy in vocabulary size between students from advantaged and disadvantaged homes. By
age three, children from advantaged families have heard 30 million more words than children from
disadvantaged homes (Neuman & Celano, 2001). By first grade, the gap is even greater (Hart &
Risley, 1995). A disadvantaged student may enter first grade with a listening vocabulary of 3,000
words, while a student from a professional home will have a vocabulary of 12,000 words. Vocabulary
differences cause phonemic awareness differences. This is one reason why advantaged students learn
to read faster than disadvantaged students. It is the early differences in a child’s oral language envi-
ronment which sets the trajectory for later reading development and likely also sets the pace of that
development (Hart & Risley, 1995; Sticht & James, 1984). Therefore, students from advantaged
families can rely on their large listening vocabularies for much longer than students from disadvantaged
families who come to school with much smaller vocabularies.

Because of the difficulty of closing the gap during the regular school day, both extended-day and
summer programs incorporating guided independent reading would seem beneficial. However, even
with extended-day and summer programs, it is highly doubtful literature-based reading in isolation
can close the achievement gap between disadvantaged and advantaged students. The same linguists
and psycholinguists who wrote the letter to the Governor of Massachusetts referred to earlier, calling
whole language “erroneous” because learning to read is not a natural process, would on the other
hand agree learning words orally through human conversation is a natural process. The great discov-
ery of Noam Chomsky of MIT in 1957 was that humans are natural oral language learners; it is
built into our genes.
56
This suggests that to completely close the gap and maximize reading growth, one needs to attend
not only to the student’s text environment but also to their oral language environment. This is one
reason Reading Renaissance recommends reading books aloud to students throughout the school
years. It’s one way to improve the oral language environment (Armbruster, Lehr, & Osborn, 2001;
Morrow & Gambrell, 2000; Trelease, 1995).

So, independent book reading is very important, but it is not a silver bullet. To fully close the gap,
there needs to be significant improvement in the oral language environment of disadvantaged stu-
dents both before they enter school and during school.

Successful Reading Practice

In this paper, we have equated successful reading practice with scoring a high average percent
correct on AR quizzes. The ALT research, too, focuses on percent correct as a measure of success
(Berliner, 1990). We have seen that quality (high percent correct) is very important because lots of
reading at low percent correct can actually cause harm. Practicing badly creates bad habits. However,
success is not only defined by high percent correct. We also need high engaged time (quantity) and
books at an optimum level of challenge. All this needs to be balanced to be successful and to develop
students who both read well and are well read.

Yet, success is even more complex than balancing the three metrics for quantity, quality, and level.
Success is also a feeling, a feeling the student must have to sustain the motivation to achieve the
balance we seek. In addition, we have shown how important the teacher is to this process. This
in turn means that the sustainability of the program also requires that teachers feel successful too.

The purpose of this digression is to point out that there is another important element to success
that this study does not address in any detail: motivation. It is worth mentioning, though, that
our experience from implementing Reading Renaissance in thousands of schools across the country
is that measurement and information feedback are essential to provide both the means to guide
independent reading as well as to ensure the motivation to read and that feeling of success for
both student and teacher.

57
Nonfiction Reading

The negative impact of nonfiction reading on student growth is one of the surprises of this study.
We know of no other study that has investigated the impact of nonfiction reading on improvement
in reading ability. While nonfiction reading is important for many reasons, it is also more difficult
for the reader, and maintaining motivation becomes more challenging.

Consistent with the finding on time spent reading and average percent correct, we found that teach-
ers can have a major impact on students’ comprehension of nonfiction reading. Indeed, it is likely,
based on our analysis, that teachers can help students become more successful with their nonfiction
reading with the result that such reading will both improve reading achievement in general and also
significantly broaden a student’s knowledge base and horizons. Some strategies teachers can use to
facilitate more successful nonfiction reading practice include: direct instruction of comprehension
strategies; helping students to read more carefully; integrating nonfiction books into the social
studies, science, and math curriculum; and making sure students have sufficient choice when
selecting nonfiction books (Harvey, 1998).

However, the current Reading Renaissance goal of 20% nonfiction reading may be too high for
some students and classrooms. In light of the findings here, Reading Renaissance now recommends
10–15% nonfiction reading as a goal. However, this recommendation still comes with a caveat. The
books need to be available and the teachers need to do their part also. In some cases, the 10–15%
recommendation may need to be adjusted according to individual needs.

Effect Size and Reading Improvement

There are various ways to estimate the potential impact of Reading Renaissance’s guided independ-
ent reading on student achievement. Independent university research shows effect sizes of between
0.57 and 0.50 for schoolwide implementations (Holmes & Brown, 2003; Sadusky & Brem, 2002).
At the classroom level, the data from this study indicate very large effect sizes for reading growth.
For instance, when comparing Reading Renaissance-certified classrooms to non-certified and non-
implementing classrooms in grades 2–8, effect sizes of 0.49 and 1.04 respectively were found.

While these effect sizes are quite large (very few educational interventions achieve these levels),
the potential for Reading Renaissance guided independent reading to improve reading is likely
greater still. For instance, these effect sizes are for one year. One would expect the cumulative
effect of two- or three-year implementations to be greater. Also, because this study identified several
Reading Renaissance recommendations that are not optimal (such as the minimum percent correct
and the 92% rule), we expect that once teachers start implementing Reading Renaissance based
on the new recommendations, reading growth will accelerate even more. Finally, effect sizes were
calculated for certified classrooms with an average implementation level of only 65%. Over time
as teachers, librarians, and principals become more familiar with Reading Renaissance, we expect
implementation levels, and consequently effect sizes and reading gains, to increase.

Transferability

58
One important question yet to be discussed is transferability; how transferable and replicable
is Reading Renaissance guided independent reading?

There is good evidence Reading Renaissance is highly transferable and replicable. Perhaps the most
compelling evidence is that nationally more than 5,000 classrooms nationwide are certified as Model
or Master Reading Renaissance Classrooms. However, it is equally true Reading Renaissance has
not been universally successful because it is not always implemented well. Indeed, the RPD is proof
of that. Many of the classrooms in the RPD are implementing at a very low level.

School improvement is a complicated business. The more experience we have, the more we realize
how important committed teachers, along with the sustained support of principals and district
administrators, are to success. The best implementations occur when everyone pulls together.

While the results of Reading Renaissance are clearly transferable in a broad sense, it really depends
on the fidelity of treatment. This study shows the value of computerized information systems to
ensure high-fidelity treatment. Computerized information systems provide the daily and weekly
feedback teachers need to guide each student’s independent reading practice. These same informa-
tion systems can also help administrators monitor the implementation at the classroom, school,
and district level. With good information systems in place, the likelihood that everyone pulls
together to ensure fidelity of treatment increases greatly. Without such systems, it may not be
possible at all.

59
Conclusion

This study provides strong empirical support for the efficacy of the Reading Renaissance program
for guided independent reading for all grades, 1–12.

The study examined both the scientific research base and the evidence base of the Reading
Renaissance program. The first part of the study demonstrated how the six key elements of
the program are well supported by the research literature. The rest of the paper examined Reading
Renaissance’s evidence of effectiveness. Effect sizes for schoolwide implementation ranged from 0.50
to 0.57. Classroom-level effect sizes averaged 0.49 for certified classrooms vs. uncertified, and 1.04
for certified vs. non-implementing classrooms.

This examination of the Reading Practice Database (RPD) also provides strong evidence that
increased amounts of independent reading cause students to become better readers. However,
we also showed that unsuccessful reading practice (low average percent correct) can actually slow
reading improvement. Therefore, having kids “just read” is not likely to be effective. Reading practice
must be closely guided by teachers on a daily basis to ensure success for all students.

In addition, we discovered that teachers need timely and objective feedback information. Without a
computerized information system like AR, it is hard to sustain the record-keeping tasks necessary to
establish personalized goals and attend to every student’s needs.

Examination of the RPD largely supports most of Reading Renaissance’s recommendations.


However, we discovered some notable exceptions. First, students should maintain higher levels
of percent correct on AR Reading Practice Quizzes than previously recommended. Specifically, stu-
dents reading at or above the fourth-grade level should average 90% instead of 85% correct or above
on AR quizzes. We also discovered that the old recommendation to move students to the next book
level when they average 92% correct on AR quizzes should be dropped. Instead, it was discovered
that the more success students have when reading, the more growth they will make. Finally, the
recommendation that 10–15% of student reading be nonfiction reading needs to be applied with
caution. Teachers need to provide instruction to students on how to read nonfiction books to
assure high average percent correct.

The data from the RPD re-confirmed the findings of many researchers and practioners: teachers
are a very important factor in a student’s success. Teachers are also the key to successful implementa-
tion of Reading Renaissance. Most importantly, we found that teachers have even more impact on
student achievement than the student’s beginning reading ability.

Finally, this study announces a new era in educational research. The RPD is the largest database ever
assembled on student independent literature-based reading. It allows us to see patterns with a level
of clarity never before possible. Yet within a few short years, fully integrated web-based systems pro-
viding daily, continuously updated student information in multiple subjects will be available. These
newer, more powerful databases will dwarf the size of the RPD. Once this happens, educational
research, including large-scale longitudinal randomized experimental research, will be conducted
on a scale never before thought possible.

This will help us to fine tune Reading Renaissance even more and bring us even closer to the goal
of ensuring that all children learn to read well and are well read.
60
Bibliography
Allington, R.L. What Really Matters for Struggling Readers: Designing Research-Based Programs.
New York: Longman, 2001.

Allington, R.L. “Content Coverage and Contextual Reading in Reading Groups.”


Journal of Reading Behavior 16 (1984): 85–96.

Allison, P.D. “Change Scores as Dependent Variables in Regression Analysis.” Sociological


Methodology 20 (1990): 93–114.

Ames, C. “Motivation: What Teachers Need to Know.” Teachers College Record 91, no. 3 (1990): 409–421.

Anderson, R.C. Research Foundations to Support Wide Reading. Publication of the Center
for the Study of Reading, technical report no. 631. Champaign, IL: University of Illinois
at Urbana-Champaign, 1996.

Anderson, R.C., E.H. Hiebert, J.A. Scott, & I.A.G. Wilkinson. Becoming a Nation of Readers: The
Report on the Commission of Reading. Washington, DC: The National Institute of Education, 1985.

Anderson, R.C., P.T. Wilson, & L.G. Fielding. “Growth in Reading and How Children Spend
Their Time Outside of School.” Reading Research Quarterly 23, no. 3 (1988): 285–303.

Armbruster, B.B., F. Lehr, & J. Osborn. Put Reading First. Washington, DC: The Partnership
for Reading, 2001.

Barr, R., M.W. Sadow, & C. Blachowicz. “Influence of Basal Programs on Fourth-Grade Reading
Instruction.” Reading Research Quarterly 24 (1989): 44–69.

Beck, I.L., & M.G. McKeown. “Social Studies Texts are Hard to Understand: Mediating Some
of the Difficulties.” Language Arts 68, no. 6 (1991): 482–490.

Berliner, D.C. “What’s All the Fuss About Instructional Time?” In The Nature of Time in Schools,
edited by M. Ben-Peretz and R. Bromme, 3–35. New York: Teachers College Press, 1990.

Berliner, D.C. “Academic Learning Time and Reading Achievement.” In Comprehension and
Teaching: Research Reviews, edited by J. Guthrie, 203–225. Newark, DE: International
Reading Association, 1981.

Betts, E.A. Foundations of Reading Instruction, With Emphasis on Differentiated Guidance.


New York: American Book Company, 1946.

Biemiller, A. “Oral Comprehension Sets the Ceiling on Reading Comprehension.”


American Educator. (Spring 2003): 23–24.

Biemiller, A. Language and Reading Success. Cambridge, MA: Brookline Brooks, 1999.

Black, P., & D. Wiliam. “Inside the Black Box: Raising Standards Through Classroom Assessment.”
Phi Delta Kappan 80, no. 2 (1998): 139–155.
61
Block, J.M. Mastery Learning: Theory and Practice. New York: Holt, Rinehart and Winston, 1971.

Bloom, B.S. “The 2 Sigma Problem: The Search for Methods of Group Instruction as Effective as
One-to-One Tutoring.” Educational Researcher 13, no. 6 (1984): 4–16.

Bloom, B.S. Human Characteristics and School Learning. New York: McGraw-Hill, 1976.

Brophy, J., & T. Good. “Teacher Behavior and Student Achievement.” In Handbook of Research
on Teaching, edited by M.C. Wittrock. 3rd ed. New York: MacMillan, 1986.

Butler, R., & M. Nisan. “Effects of No Feedback, Task-Related Comments, and Grades on Intrinsic
Motivation and Performance.” Journal of Educational Psychology 78, no. 3 (1986): 210–216.

Campbell, D.T., & D.A. Kenny. A Primer on Regression Artifacts. New York: Guilford Press, 1999.

Chall, J. Stages of Reading Development. 2nd ed. Fort Worth, TX: Harcourt Brace, 1996.

Chall, J., V. Jacobs, & L. Baldwin. The Reading Crisis: Why Poor Children Fall Behind.
Cambridge, MA: Harvard University Press, 1990.

Chomsky, N. Syntactic Structures. Gravenhage: Mouton, 1957.

Cunningham, P.M., & J.W. Cunningham. “What We Know About How to Teach Phonics.”
In What Research Has to Say About Reading Instruction, edited by A.E Farstrup and S.J.
Samuels. Newark, DE: International Reading Association, 2002.

Denham, C., & A. Lieberman. Time to Learn. Washington, DC: U.S. Government Printing Office, 1980.

Deno, S.L. Experimental Teaching: An Approach to Improving Student Achievement, Changing


Teacher Beliefs, and Identifying Effective Practices. Minneapolis, MN: University of
Minnesota, 1985.

Din, F.S. Use Direct Instruction to Quickly Improve Reading Skills. Paper presented at the Annual
National Conference on Creating the Quality School, Arlington, VA, March, 1998.

Dreher, M.J. “Motivating Children to Read Nonfiction.” The Reading Teacher 52, no. 4
(1999): 414–415

Ehri, L.C. “Phases of Development in Learning to Read Words by Sight.” Journal of Research
in Reading 18 (1995): 116–125.

Elley, W.B., & F. Mangubhai. “The Impact of Reading on Second Language Learning.” Reading
Research Quarterly XIX (1983): 53–67.

Fisher, C.W., N.N. Filby, R.S. Marliave, L.S. Cahen, M.M. Dishaw, J.E. Moore, & D.C. Berliner.
Teaching Behaviors, Academic Learning Time, and Student Achievement, Final Report of
Phase III-B, Beginning Teacher Evaluation Study. San Francisco: Far West Laboratory
for Educational Research and Development, 1978.
62
Fry, E. How to Teach Reading, California: Teacher Created Materials, 2000.

Fuchs, L.S., & D. Fuchs. “Curriculum-Based Measurement: A Methodology for Evaluating and
Improving Student Programs.” Diagnostique 14, no. 1 (1988): 3–13.

Gage, N.L., & D.C. Berliner. Educational Psychology. 5th ed. Boston: Houghton Mifflin, 1992.

Gersten, R., & D. Carnine, Administrative and Supervisory Support Functions for the Implementation of
Effective Educational Programs for Low Income Students, Oregon: Center for Educational
Policy & Management, University of Oregon, 1981.

Gickling, E.E., & D.L. Armstrong, “Levels of Instructional Difficulty as Related to On-Task
Behavior, Task Completion, and Comprehension.” Journal of Learning Disabilities 11, no. 9
(1978): 559–566.

Goodman, K.S. What’s Whole in Whole Language? Portsmouth, NH: Heinemann, 1986.

Goodman, K.S., E.B. Smith, R. Meredith, & Y. Goodman. Language and Thinking in School:
A Whole Language Curriculum. Katonah, NY: Richard C. Owen, 1987.

Goodman, K.S., Y. Goodman, & W.J. Hood. The Whole Language Evaluation Book. Portsmouth,
NH: Heinemann, 1989.

Haller, E., D. Child, & H.J. Walberg. “Can Comprehension Be Taught? A Quantitative Synthesis.”
Educational Researcher 17, no. 9 (1988): 5–8.

Hart, B., & T. Risley. Meaningful Differences in Everyday Parenting and Intellectual Development
in Young American Children. Baltimore: Brookes, 1995.

Harvey, S. Nonfiction Matters: Reading, Writing and Research in Grades 3–8. Portland, ME:
Stenhouse, 1998.

Hattie, J.A. “Measuring the Effects of Schooling.” Australian Journal of Education 36, no. 1 (1992):
5–13.

Holmes, C.T., & C.L. Brown. “A Controlled Evaluation of a Total School Improvement Process,
School Renaissance.” Technical Report. Athens, GA: University of Georgia, 2003. Available
online: <http://www.coe.uga.edu/leadership/faculty/holmes/articles.html>.

Howell, D. Statistical Methods for Psychology, 5th ed. Pacific Grove, CA: Duxbury, 2002.

Johnson, B., & E. Stone. “Is Whole Language Restructuring our Classroom?” Contemporary
Education 62, (1991): 102–104.

Juel, C. “Beginning Reading.” In Handbook of Reading Research, edited by R. Barr, M.L. Kamil,
P.B. Mosenthal, and P.D. Pearson, 2nd vol. New York: Longman, 1991.

Kirsch, I., J. de Long, D. LaFontaine, J. McQueen, J. Mendelovits, & C. Monseur. Reading for
Change: Results from PISA 2000. France: Organization for Economic Co-Operation and
Development (OECD), 2002. 63
Kohn, A. The Case Against Standardized Testing. Portsmouth, NH: Heinemann. London:
MIT Press, 2000.

Kumar, D.D. “A Meta-Analysis of the Relationship Between Science Instruction and Student
Engagement.” Education Review 43, no. 1 (1991): 49–66.

Laufer, B. “What Percentage of Text-Lexis is Essential for Comprehension?” In Special Language:


From Humans Thinking to Thinking Machines, edited by C. Lauren and M. Nordman,
316–323. Philadelphia: Multilingual Matters, 1989.

Lewis, M., & S.J. Samuels. “Read More—Read Better? A Meta-Analysis of the Literature
on the Relationship Between Exposure to Reading and Reading Achievement.”
Minneapolis, MN: University of Minnesota, 2003. Available online:
<http://www.tc.umn.edu/~samue001/publications.htm>.

Lipsey, M.W., & D.B. Wilson. “The Efficacy of Psychological, Educational, and Behavioral
Treatment: Confirmation from Meta-Analysis.” American Psychologist 48, no. 12 (1993):
1181–1209.

Marliave, R.S., & N.N. Filby. “Success Rate: A Measure of Task Appropriateness.”
In Perspectives on Instructional Time, edited by C.W. Fisher, and D.C. Berliner.
White Plains, NY: Longman, 1985.

Marzano, R.J. What Works in Schools: Translating Research Into Action. Alexandria, VA:
Association for Supervision and Curriculum Development, 2003.

Morrow, L.M., & L.B. Gambrell. Literature-Based Reading Instruction. In Handbook of Reading
Research, edited by M.L. Kamil, P.B. Mosenthal, P.D. Pearson, and R. Barr. 2nd vol.
Mahwah, NJ: Erlbaum, 2000.

Nagy, W.E., & P.A. Herman. “Breadth and Depth of Vocabulary Knowledge: Implications for
Acquisition and Instruction.” In The Nature of Vocabulary Acquisition, edited by M.G.
Mckeown and M.E. Curtis, 19–35. Hillsdale, NJ: Lawrence Erlbaum Associates, 1987.

Nation, P. Teaching and Learning Vocabulary. Boston: Heinle & Heinle, 1990.

National Institute of Child Health and Human Development. Report of the National Reading
Panel. Teaching Children to Read: An Evidence-Based Assessment of the Scientific
Research Literature on Reading and Its Implications for Reading Instruction. Washington,
DC: U.S. Government Printing Office, 2000a. NIH Publication no. 00-4769.
Available online: <http://www.nichd.nih.gov/publications/nrp/smallbook.htm>.

National Institute of Child Health and Human Development. Teaching Children to Read:
An Evidence-Based Assessment of the Scientific Research Literature on Reading and Its
Implications for Reading Instruction: Reports of the Subgroups. Report of the National
Reading Panel. Washington, DC: U.S. Government Printing Office, 2000b. NIH
Publication no. 00-4754. Available online:
<http://www.nichd.nih.gov/publications/nrp/smallbook.htm>.
64
Neuman, S.B., & D. Celano. “Access to Print in Low-Income Communities: An Ecological Study
of Four Neighborhoods.” Reading Research Quarterly 36, no. 1 (2001): 8–26.

Nunnery, J.A., S.M. Ross, & E. Goldfeder. ”The Effect of School Renaissance on TAAS Scores in
the McKinney ISD.” Memphis, TN: Center for Research in Educational Policy, University
of Memphis, 2003. Available online:
<http://www.people.memphis.edu/~coe_crep/research.htm>.

Palincsar, A.S. & A.L. Brown. “Reciprocal Teaching of Comprehension-Fostering and


Comprehension-Monitoring Activities.” Cognition and Instruction 1, (1984): 117–175.

Parkes, B. “Real-World Reading & Writing: Nonfiction Opens Doors to a Life of Learning.”
Creative Classroom (May/June 2001): 46–50. Available online:
<http://www.pearsoned.com.au/schools/primary/pdf/pdArticles/Realworld.pdf>.

Paul, T. “National Study of Literature-Based Reading: How Literature-Based Reading Improves


Both Reading and Math.” Madison, WI: Institute for Academic Excellence, 1993.

Paul, T. “National Reading Study and Theory of Reading Practice.” Madison, WI: Institute for
Academic Excellence, 1992.

Rich, S. “Whole Language: The Inner Dimension.” English Quarterly 18, no. 2 (1985): 23–37.

Rosenshine, B., & R. Stevens. “Classroom Instruction in Reading.” In Handbook of Reading


Research, edited by P.D. Pearson, 745–798. New York: Longman, 1984.

Sadusky, L.A., & S.K. Brem. “The Integration of Renaissance Programs into an Urban Title I
Elementary School, and its Effect on School-wide Improvement.” Technical Report. Tempe,
AZ: Arizona State University, 2002. Available online: <http://www.public.asu.edu/~sbrem/>.

Samuels, S.J., M. Lewis, Y. Wu, J. Reininger, & A. Murphy. “Accelerated Reader vs. Non-
Accelerated Reader: How Students Using Accelerated Reader Outperformed the Control
Condition in a Tightly Controlled Experimental Study.” Technical Report. Minneapolis,
MN: University of Minnesota, 2003a.

Samuels, S.J., M. Lewis, Y. Wu, J. Reininger, & A. Murphy. “Effect of Increased Individual Reading
Time on Reading Achievement.” Technical Report. Minneapolis, MN: University of
Minnesota, 2003b. Available online: <http://www.tc.umn.edu/~samue001/>.

Sanders, S.J. & K.J. Topping. “Teacher Effectiveness and Computer Assessment of Reading, Technical
Report.” Knoxville, TN: University of Tennessee Value Added Research and Assessment Center, 1999.

Scheerens, J., & R. Bosker. The Foundations of Educational Effectiveness. Oxford, UK: Pergamon, 1997.

Shany, M.T., & A. Biemiller. “Assisted Reading Practice: Effects on Performance for
Poor Readers in Grades 3 and 4.” Reading Research Quarterly 30 (1995): 382–395.

Skinner, B.F. The Technology of Teaching. Englewood Cliffs, NJ: Prentice-Hall, 1968.

Skinner, B.F. “The Science of Learning and the Art of Teaching.” Harvard Educational Review (1954): 86–97.
65
Slavin, R.E., N.L. Karweit, & N.A. Madden. Effective Programs for Students at Risk. Boston:
Allyn and Bacon, 1989.

Snow, C.E. Reading for Understanding: Toward an R&D Program in Reading Comprehension.
Santa Monica, CA: Rand, 2002.

Snow, C.E., M.S. Burns, & P. Griffin, eds. Preventing Reading Difficulties in Young Children. Washington,
DC: National Academy Press, 1998. Available online: <http:// www. nap. edu/catalog/6023.html>.

Stahl, S.A. “Four Questions About Vocabulary.” In Learning From Text Across Conceptual Domains,
edited by C.R. Hynd, 73–94. Mahwah, NJ: Erlbaum, 1998.

Stahl, S.A., M. McKenna, & J. Pagnucco. “The Effects of Whole-Language Instruction: An Update
and Reappraisal.” Educational Psychologist 29, no. 4 (1994): 175–185.

Stahl, S.A., & P.D. Miller. “Whole Language and Language Experience Approaches for Beginning
Reading: A Quantitative Research Synthesis.” Review of Educational Research 59, (1989): 87–116.

Sticht, T.G., & J.H. James. “Listening and Reading.” In Handbook of Reading Research, edited by
P. David Pearson, 293–318. New York: Longman, 1984.

Topping, K.J., & W.L. Sanders. “Teacher Effectiveness and Computer Assessment of Reading:
Relating Value-Added and Learning Information Systems Data.” School Effectiveness and
School Improvement 11, no. 3 (2000): 305–337.

Trelease, J. The Read-Aloud Handbook. 4th ed. New York: Penguin Books, 1995.

U.S. Department of Education. Office of Educational Research and Improvement. National


Center for Education Statistics. “The Nation’s Report Card: Fourth-Grade Reading 2000,”
NCES 2001-499, by P.L. Donahue, R.J. Finnegan, A.D. Lutkus, N.L. Allen, and J.R.
Campbell. Washington, DC: 2001.

Vygotsky, L. Mind in Society: The Development of Higher Psychological Processes. Cambridge:


Harvard University Press, 1978.

Walberg, H.J. “Productive Teaching.” In New Directions for Teaching Practice and Research, edited by
H.C. Waxman and H.J. Walberg, 75–104. Berkeley, CA: McCutchen Publishing Corporation, 1999.

Walberg, H.J. “U.S. Schools Teach Reading Least Productively.” Research In the Teaching of English 30, no. 3
(1996): 328–343.

Walberg, H.J. “Improving the Productivity of America’s Schools.” Educational Leadership 41, no. 8
(1984): 19–27.

Wise, K.C., & J.R. Okey. “A Meta-Analysis of the Effects of Various Science Teaching Strategies on
Achievement.” Journal of Research in Science Teaching 20, no. 5 (1983): 415–425.

Yu, V. “Extensive Reading Programs—How Can the Best Benefit the Teaching and Learning of
English?” TESL Reporter 26, no. 1 (1993): 1–9.

66
APPENDIX A
ATOS Readability Formula for Books

Matching books to students is a two-step process in which the second step, continuous adjustment,
is most important. However, teachers must still make initial estimates of ZPD ranges, and readabili-
ty formulas are an important tool for making this estimate. In 1998, Renaissance Learning with sev-
eral outside readability experts, including Touchstone Applied Science Associates (TASA), conduct-
ed a study of readability that resulted in the ATOS Readability Formula for Books. There is also an
ATOS Readability Formula for Text that is very similar but does not consider things such as book
length, extreme sentence length, and other special issues involved with early reader type books.
ATOS for Books is thus the only readability formula which is specifically designed to measure the
text difficulty in the context of book reading.

The first step in creating ATOS was to analyze variables used in readability research as indicators
of text difficulty. All readability formulas measure “semantic” and “syntactic” difficulty (difficulty
of words and difficulty of sentence structure). More than 100 different measures of semantic and
syntactic components were correlated with the difficulty levels of 650 leveled items. These items
were drawn from STAR Reading and TASA’s Degrees of Reading Power (DRP) test.

The table below shows the percentage of variance in difficulty in the DRP items explained by several
measures of text difficulty. This helped us identify the most important predictors of difficulty.

Variables Correlation (R2)


Sentence length 0.897
Average grade level of words 0.891
Percent of familiar words 0.867
Syllables per word 0.839
Word length 0.839
Word frequency 0.769

Additional analysis of the DRP items using regression modeling showed that the combination of
sentence length (natural log transformation), average grade level of words, and word length (natural
log transformation) explained the most variance in difficulty. The resulting formula was validated
using STAR Reading items, and passages from several standardized tests.

The formula was further refined and validated by looking at the actual reading of more than 30,000
students who read more than 950,000 books.

The final formula includes adjustments for book length, books with extreme sentence length, and
books for emergent readers.

67
Book difficulty as measured by ATOS is calculated by:

1. Compute raw difficulty of text:


Raw score = -8.54 + 1.95 x Adjusted Sentence Length + 0.46 x
Average Grade Level + 1.74 x Ln(Average Word Length)

where sentence length is adjusted if it is extreme for the semantic difficulty of the text.
Otherwise, Adjusted Sentence Length is simply Ln(Average Sentence Length). The
adjustment works as follows:

Predicted Maximum = 0.45 x [0.46 x Average Grade Level + 1.74 x Ln(Average Word
Length)] + 0.83 + 0.2
Predicted Minimum = 0.45 x [0.46 x Average Grade Level + 1.74 x Ln(Average Word
Length)] + 0.83 – 0.8
If Ln(Sentence Length) > Predicted Maximum then
Adjusted Sentence Length = Predicted Maximum + 0.25 x [Ln(Sentence Length) –
Predicted Maximum]
If Ln(Sentence Length) < Predicted Minimum then
Adjusted Sentence Length = Predicted Minimum + 0.25 x [Ln(Sentence Length) –
Predicted Maximum]

2. Convert to grade-level scale:


ATOS Grade Level = 5.86 + 2.86 x Raw score + 0.32 x Raw score2
(The grade-level scale is designed to equate to the Flesch-Kincaid scale for difficulty.)

3. Adjust for book length:


a. ATOS proportion = 0.0648 x Ln(Book Length) + 0.0894
b. Book Length proportion = 1 – ATOS proportion
c. For books with 500 or more words:
Book Length Grade Level = 0.68 x Ln(Book Length) – 1.87
For books with fewer than 500 words:
Book Length Grade Level =0.004 x Book Length + 0.4
d. Final ATOS Grade Level = ATOS proportion x ATOS Grade Level + Book Length
proportion x Book Length Grade Level

68
APPENDIX B
page 1 of 2

Reading Practice
TOPS Report for Sara Shin
Accelerated Reader®: Monday, 09/22/03, 10:05 AM

Lake View School


Class: Section 4 G3 Teacher: Mrs. Sally Jacobson
ID: 2369 Grade: 3 Team: Mustangs

Good Job, Sara! You have answered 9 out of 10 questions correctly on quiz number 34685 for the
book A Reward for Josefina by Valerie Tripp.

Reading Practice Quiz Results


Book Level: 4.2
Number Correct / Possible: 9 / 10 (90%)
Points Earned / Possible: 0.9 / 1.0
TWI: Read Independently
F/NF: Fiction

Marking Period Results to Date (48% of Marking Period One)


Marking Period Goal
Average Book Level: 4.1 3.3
Average Percent Correct: 92.1% 85.0%
Points Earned: 11.6 18.0 (64.4% of goal)
Reading Practice Quizzes Passed / Taken: 14 / 14

School Year Results to Date (8% of school year)


Average Book Level: 4.1
Average Percent Correct: 92.1%
Points Earned: 11.6
Reading Practice Quizzes Passed / Taken: 14 / 14

Certification Level
Last Certification & Date Current Certification Goal
Rising Reader 9/17/2003 Rising Reader 2

Monitor Signature Teacher Signature


Comments:

69
APPENDIX B
page 2 of 2

Diagnostic Report - Reading Practice Page 1

Accelerated Reader®: Tuesday, 10/14/03, 03:39 PM


Report Period: 9/2/2003 - 10/14/2003 Marking Period One
Lake View School

Class: Language Arts 5th Hour Teacher: Mr. Mike Garcia


RP Book Certification
Quizzes % Correct RP Points Level Level

Earned

Fiction
Passed

Indep.
Taken
Codes
Diag.

Read
Working

% of
Goal

Goal

Goal

Goal
Avg.

Avg.
%

%
Student Name Toward
Adams, Emily 13 13 85.0 94.2 30.8 68.4 222.1 100.0 96.1 5.0 6.0 Adv(3)
Ashbeck, Mary 7 7 85.0 95.7 21.0 25.7 122.4 100.0 89.5 4.5 4.8 Adv
Conner, Tina B% 8 9 85.0 71.1* 17.3 22.6 130.6 100.0 90.7 3.9 4.3 Adv
Jenkins, Evan 10 10 85.0 93.0 16.5 25.0 151.5 100.0 100.0 3.7 4.0 Adv(2)
Khang, Mao BD% 5 6 85.0 75.0* 11.0 4.1* 37.3 87.8 70.7 2.9 3.6 Sup
Lee, Michael C% 6 6 85.0 68.3* 17.0 13.8 81.2 94.2 94.2 3.8 4.7 Achieved
Roberts, Tim 7 7 85.0 91.4 16.0 19.8 123.8 100.0 90.4 3.7 3.9 Adv
Rodriguez, Maria 9 9# 85.0 88.9 18.8 28.0 148.9 100.0 93.9 4.2 4.6 Adv
Wilder, Jason B% 9 10 85.0 74.0* 12.5 11.7 93.6 92.3 88.0 2.9 3.7 Adv

Totals 74 77 85.0 83.5 160.9 219.1 97.1 90.4 3.8 4.4

Summary
Total number of students: 9
Median of points earned: 22.6
1/2 of median points earned: 11.3
Number of students at risk: 4 ( 44.4%)
% of nonfiction: 9.6
Number of students below 85% correct: 4 ( 44.4%)
Number
Diagnostic of
Codes Students Description
A 0 No quizzes taken during period
B 3 Low average percent correct (70% to 79% )
C 1 Very low average percent correct (below 70%)
D 1 Low points earned - less than 1/2 of median
E 0 Low percent correct with above median points
F 0 Very low percent correct with above median points
% 4 Average Percent Correct below 85%

RP Book Certification
Quizzes % Correct RP Points Level Level
Earned

Fiction
Passed

Indep.
Taken
Codes
Diag.

Read

Working
% of
Goal

Goal

Goal

Goal
Avg.

Avg.
%

Toward
Report Totals 74 77 85.0 83.5 160.9 219.1 97.1 90.4 3.8 4.4

Code Explanation Code Explanation


* Trouble value "Rdy" Ready Reader certification
@ Includes quizzes without book levels "Ind" Independent Reader certification
# Spanish quizzes included "Ris" Rising Reader certification
-- No data "Sup" Super Reader certification
"Adv" Advanced Reader certification
"Sta" Star Reader certification
"Cla" Classic Reader certification
"Sta(2)" Indicates this is the second Star Reader certification

70
APPENDIX C
Goal-Setting Chart
Use the chart and guidelines below to help plan goals on students’ reading growth. If a student struggles to
for your students based on their reading levels and maintain the minimum average, talk to her and identify
amount of daily reading practice. the cause. Then decide on a strategy to help her succeed.

Identify ZPD Point goals—The chart shows the number of points stu-
Get each student’s Grade-Equivalent (GE) dents are expected to earn depending on how much time
score from a standardized assessment, such they read and at what levels. These are estimates—set
as STAR Reading, or estimate a GE based on goals that are realistic for your individual students.
past performance. The corresponding ZPD is a
Provide Daily Reading Practice
recommended book-level range for the student. If books
Recommendations for students reading
in that range seem too hard or easy for the student,
independently:
choose a new range or create a wider one that better
matches his abilities. • Elementary school—60 minutes
• Middle school (6th–8th)—45 minutes*
Set Goals
• High school (9th–12th)—30 minutes*
Average percent correct—The most important goal for
all students is to average 85% or above on Reading * Students reading below grade level need
Practice Quizzes. Meeting this goal has significant impact at least 60 minutes of daily reading practice.

60 Min. Daily Practice 45 Min. Daily Practice 30 Min. Daily Practice


Grade- Points Points Points Points Points Points Points Points Points
Equivalent Suggested per per per per per per per per per
Score ZPD Week 6 Weeks 9 Weeks Week 6 Weeks 9 Weeks Week 6 Weeks 9 Weeks
1.0 1.0 – 2.0 1.7 10 15 1.3 7.5 11.3 0.9 5.0 7.5
1.5 1.5 – 2.5 1.9 11 17 1.4 8.3 12.8 1.0 5.5 8.5
2.0 2.0 – 3.0 2.1 13 19 1.6 9.8 14.3 1.1 6.5 9.5
2.5 2.3 – 3.3 2.3 14 21 1.7 10.5 15.8 1.2 7.0 10.5
3.0 2.6 – 3.6 2.5 15 23 1.9 11.3 17.3 1.3 7.5 11.5
3.5 2.8 – 4.0 2.7 16 24 2.0 12.0 18.0 1.4 8.0 12.0
4.0 3.0 – 4.5 2.8 17 25 2.1 12.8 18.8 1.4 8.5 12.5
4.5 3.2 – 5.0 3.2 19 29 2.4 14.3 21.8 1.6 9.5 14.5
5.0 3.4 – 5.4 3.5 21 32 2.6 15.8 24.0 1.8 10.5 16.0
5.5 3.7 – 5.7 3.9 23 35 2.9 17.3 26.3 2.0 11.5 17.5
6.0 4.0 – 6.1 4.2 25 39 3.2 18.8 29.3 2.1 12.5 19.5
6.5 4.2 – 6.5 4.6 28 41 3.5 21.0 30.8 2.3 14.0 20.5
7.0 4.3 – 7.0 4.9 29 44 3.7 21.8 33.0 2.5 14.5 22.0
7.5 4.4 – 7.5 5.3 32 48 4.0 24.0 36.0 2.7 16.0 24.0
8.0 4.5 – 8.0 5.6 34 50 4.2 25.5 37.5 2.8 17.0 25.0
9.0 4.6 – 9.0 6.3 38 57 4.7 28.5 42.8 3.2 19.0 28.5
10.0 4.7 – 10.0 6.9 41 62 5.2 30.8 46.5 3.5 20.5 31.0
11.0 4.8 – 11.0 7.6 46 68 5.7 34.5 51.0 3.8 23.0 34.0
12.0 4.9 – 12.0 8.3 50 75 6.2 37.5 56.3 4.2 25.0 37.5

71
Acknowledgements
A study like this has many authors. This one has even more than most because it is the culmination
of many other previous studies and events.

The most important event was of course the invention of the Accelerated Reader information
system, to which I owe my wife, Judi, thanks. She wouldn’t give up until she had a system to
motivate and manage a summer reading program for our four children. With her degree in elemen-
tary education, experience in working in schools, understanding of kids, and a determination only
mothers have, she didn’t quit until she had invented the paper system which later became
Accelerated Reader.

Next was the invention of Reading Renaissance, which over the years has so many contributors—
teachers, librarians, principals, and Renaissance Learning employees—that they are truly too numer-
ous to mention. Suffice it to say, there is the blood, sweat, and tears of a hundred people who helped
create the Reading Renaissance program for guided independent reading.

Although I take sole responsibility for any inaccuracies or misstatements contained in this report,
there are certain individuals who played an important role both helping to reduce the inaccuracies,
improve the prose, create the Reading Practice Database, and do much of the statistical analysis who
deserve specific mention. These are Teri Thill, Eileen Hannigan, and Laurie Borkon of our Research
and Evaluation Department at Renaissance Learning.

Terrance D. Paul
Co-Founder
Renaissance Learning, Inc.
August 11, 2003

72
PO Box 8036
Wisconsin Rapids, WI 54495-8036
(866) 846-7323
www.renlearn.com

L1818.1003.SP.4M

You might also like