Вы находитесь на странице: 1из 46

Effective Schools and Accomplished Teachers: Lessons about Primary-Grade Reading

Instruction in Low-Income Schools

Author(s): Barbara M. Taylor, P. David Pearson, Kathleen Clark and Sharon Walpole
Source: The Elementary School Journal, Vol. 101, No. 2 (Nov., 2000), pp. 121-165
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/1002340
Accessed: 11-12-2017 14:39 UTC

Linked references are available on JSTOR for this article:
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and
extend access to The Elementary School Journal

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms
Effective Schools Abstract

and Accomplished We investigated school and classroom factors re-

lated to primary-grade reading achievement in
Teachers: schools with moderate to high numbers of stu-
dents on subsidized lunch. 14 schools across the

Lessons about United States and 2 teachers in each of grades

K-3 participated. 2 low and 2 average readers
per class were tested individually in the fall and
Primary-Grade spring on measures of reading accuracy, flu-
ency, and comprehension. The teachers were
Reading Instruction observed
in 5 times by trained observers between
December and April during an hour of reading
Low-Income Schools instruction, completed a written survey, com-
pleted a weekly log of reading/language arts
activities in February and again in April, and
were interviewed in May. Each school was
identified as most, moderately, or least effective
Barbara M. Taylor based on several measures of reading achieve-
University of Minnesota ment in the primary grades. A combination of
school and teacher factors, many of which were
intertwined, was found to be important in the
P. David Pearson most effective schools. Statistically significant
Michigan State University school factors included strong links to parents,
systematic assessment of pupil progress, and
strong building communication and collabora-
Kathleen Clark tion. A collaborative model for the delivery of
University of Minnesota
reading instruction, including early reading
interventions, was a hallmark of the most effec-
tive schools. Statistically significant teacher fac-
Sharon Walpole tors included time spent in small-group instruc-
tion, time spent in independent reading, high
University of Virginia
levels of student on-task behavior, and strong
home communication. More of the most accom-
plished teachers and teachers in the most effec-
tive schools supplemented explicit phonics in-
struction with coaching in which they taught
students strategies for applying phonics to their
everyday reading. Additionally, more of the
most accomplished teachers and teachers in the
most effective schools employed higher-level
questions in discussions of text, and the most
accomplished teachers were more likely to ask
students to write in response to reading. In all
The Elementary School Journal
Volume 101, Number 2 of the most effective schools, reading was
? 2000 by The University of Chicago. All rights reserved.clearly a priority at both the school and class-
0013-5984/2001 /10102-0001$02.00 room levels.

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

We all want the best schools

wide organizational issuespossible,
and improved
schools that help children acquire the
classroom reading instruction.
knowledge, skills, and
In their dispositions they
special-strategies study, String-
will need to pursue et al. (1997) found dreams
whatever that reform pro- and
paths they wish. Yet that focused
many on the primary
children aregrades
reading well enough had larger
to keep achievement
up gains in reading
with the
demands of schoolthan programs that spread
(Campbell, their efforts
across the elementary
Reese, & Phillips, 1996; Donahue, grades or into the sec-
Campbell, & Mazzeo, ondary1999),
grades. They also
let found that schools
alone the
demands of society or adoptedpersonal
their externally developed
dreams. pro-
In the recent national grams had greater
report, achievement gainsRead-
Preventing than
schools that
ing Difficulties in Young developed theiraown
Children, programs.
Academy of Science In a report of a large national
committee study of
that "quality classroom 400 Chapter 1 schools, researchers
instruction infound
dergarten and the primary that higher poverty,
grades greater
is application
the sin- of
gle best weapon against reading failure" policies, and more student
(Snow, Burns, & Griffin, disciplinary actions
1998, werep. related to lowerThe
committee recommended that the number student achievement (Puma et al., 1997).
one priority for funding research should be five schools in the pool of 400 were
to improve classroom reading instruction in as exceptional. These schools
tended to have a "more experienced prin-
kindergarten and the primary grades.
In a recent report of a 3-year studycipal,
of a schoolwide Chapter 1 program,
schools implementing special strategies to
tracking by ability in grades 1-6,
lower rates of teacher and student mobility,
improve reading achievement, researchers
described classroom instruction as "in one a balanced emphasis on remedial and
higher-order thinking in classroom instruc-
sense, distressing" (Stringfield, Millsap, &
tion, and higher levels of community and
Herman, 1997, p. 2). In the elementary
parent support" (p. 62). Except for first
schools, instruction was predominantly
grade, in which grouping was used, whole-
teacher-led, focused on discrete skill in-
class instruction was the dominant practice
struction, and driven by management con-in all schools.
cerns. There were relatively few observa-
In a recent study of nine high-
tions of students engaged in sustained performing, high-poverty urban elementary
reading or students applying what they schools, important similarities found across
were learning. Moreover, Stringfield et al. the schools included the following: a collec-
(1997) pointed out that even in schools tive sense of responsibility for school im-
nominated as exemplary, there was ample provement including a focus on putting the
room for instructional improvement, whichchildren first, a well-behaved student body,
would, if implemented, lead to greater an increase in time spent on instruction, col-
gains in reading achievement. laborative planning and learning time for
In addition to advocating improvedteachers with a focus on instructional is-
classroom reading instruction, the Com- sues, instructional leadership and support
mittee on the Prevention of Reading Diffi- provided to teachers to help them focus on
culties in Young Children discussed the their teaching and students' learning, and
importance of systematic, schoolwide re-increased effort to collaborate with parents
structuring efforts in reading. The commit- (Charles A. Dana Center, 1999). However,
tee (Snow et al., 1998) recommended thatthis study did not focus on reading specif-
poor-performing schools consider reading ically or on instruction within classrooms.
reform efforts with a dual focus on school- These recent national reports highlight


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

* ongoing
the importance of and needcurriculum improvement;
for additional
research on schools that serve the needs of * maximum use of instructional time;
* frequent monitoring of student pro-
poor children by increasing their achieve- gress; and
ment and, hence, their educational oppor- * positive home-school relationships.
tunities. The purpose of the present study
was to examine the instructional and or-In a study of four outlier inner-city
ganizational factors that might explain
schools, Weber (1971) found strong lead-
how and why some schools are attain-ership, high expectations, positive school
climate, strong emphasis on reading, and
ing greater-than-expected primary-grade
reading achievement with students atcontinuous
risk evaluation of pupil progress
related to the identified school reading suc-
for failure by virtue of poverty. We empha-
cess criteria of median achievement level on
size the terms "instructional" and "organi-
zational," for we believe that only when a normed
we standardized reading achieve-
attend to both school-level (organizational) ment test and having a relatively small
and classroom-level (instructional) facets of
number of children with serious reading
reform do we meet our aspirations. difficulties. In a study of five schools found
Within this broader framework, we to be most effective out of a sample of 741
were, like the researchers in the special-schools that were part of a study of com-
strategies study (Stringfield et al., 1997), in-pensatory reading programs, Wilder (1977)
terested in both imported models of reformfound the following factors common to all
(where schools had adopted an external schools: reading identified as an important
school reform program) and homegrowninstructional goal, leadership in the reading
reform efforts. To that end, we soughtprogram provided by either the principal or
schools in both categories. We were also in- reading specialist, attention given to basic
terested in schools that had adopted early skills, breadth of materials made available,
reading interventions, both tutorial andand communication of ideas across teach-
small-group interventions for at-risk chil-ers, a process typically fostered by the pro-
dren. As it turned out, we had a number ofgram leader.
schools that had adopted early reading in- In a more recent longitudinal study on
tervention programs set within a home-schools implementing special strategies for
grown schoolwide reform effort. educating disadvantaged children, String-
field et al. (1997) found that the schools dem-
Effective Schools
onstrating the greatest achievement gains
Research on effective schools relevant to
worked hard at both initial implementation
reading achievement, much of which was
and long-term maintenance of an innova-
conducted in the 1970s and early 1980s,
tion. But the researchers also noted the im-
was documented in a review "Teacher and portance of systematic self-improvement in
School Effects in Learning to Read" by
these schools, in which the innovations con-
Hoffman (1991) in the Handbook of Reading
tinued to evolve and expand. Externally de-
Research, volume 2. Hoffman described veloped research-based programs and pro-
eight attributes of effective schools fre-
grams that focused on whole-school reform
quently summarized in the literature (e.g.,
were related to greater achievement gains
Shavelson & Berliner, 1988), including: than locally developed programs and in-
novations composed of pull-out programs.
* a clear school mission;
The study also found support for the prem-
* effective instructional leadership and ise that students at risk of academic failure
* high expectations; could achieve at levels that met national
* a safe, orderly, and positive environ- averages.
ment; In a study of five effective Title 1

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

schools, Puma et al. reported (1997) found

on the patterns ofthat high-
the most effec-
performing, high-poverty schools
tive teachers, who represented about ahad
lower-than-average teacher
of the and
sample. He found student
these teachers were
mobility, principalsbusinesslike with with more a strongexperience,
sense of task and
and more orderly direction school for themselves and their students,
than average high-poverty had high expectations
schools. for their students'
school climates and better relations with ad- achievement, and redoubled efforts when
ministration and the community were also failure was experienced, especially in low-
reported as well as greater parent involve- socioeconomic status (SES) environments.
ment and more parents with high expecta- The most effective teachers had strong man-
tions for their children's future educational agement skills, but their classrooms were
attainment. All of these high-performing not stern or oppressive. They had high pu-
schools had tracking by ability in grades pil 1- engagement and were proactive in pre-
6. In three of the five schools, teachers em- venting disruptions. The most effective
phasized basic skills and higher-order com- teachers probed students who offered in-
prehension skills in reading. correct responses instead of simply calling
Although research on effective schools on someone else or giving the answer them-
has been received favorably by school lead- selves. The students in low-SES classes of
ers and policy makers, Hoffman (1991) most effective teachers had a success
noted the limitations of this research, due
rate of about 80% correct when answering
mainly to its lack of connection to classroomteacher-directed questions, almost all of
practice and to insufficient information on
which were literal. In a follow-up interven-
the process that schools went through to be-tion study of first-grade teachers engaged in
come effective. Even so, the fact that thesmall-group instruction, Anderson, Evert-
same characteristics arise time and time son, and Brophy (1979) found that greater
again has led many reformers to suggest achievement was related to more time spent
that these findings ought to be translatedin reading groups, more active instruction,
into policies that can guide reform. shorter transitions, introduction of lessons
with an overview, and follow-up by teach-
Effective Teachers
ers to incorrect responses with attempts to
In addition to research on effective schools,
improve upon them.
Hoffman (1991) summarized a consider- In a study of 166 first- and third-grade
teachers of children who had been in Head
able body of research, spanning the 1960s
through the 1980s, on teachers who were Start, variables positively related to gains
exceptionally effective in helping students
in reading included time spent in academic
activities, frequency of small-group in-
learn to read. Hoffman reported on a litera-
ture review of effective teaching by Rosen-
struction in basic skills, and frequency of
shine and Furst (1973) in which they found
supervised seat work activities (Stallings &
Kaskowitz, 1974). The lowest-SES students
several teacher behaviors consistently re-
lated to student achievement: clarity, vari-
benefited most from intense, small-group
ability, enthusiasm, task orientation, teacher
In a study of 25 second-grade and 21
directness, student opportunity to learn cri-
terion material, use of structuring com-fifth-grade classrooms, Fisher et al. (1980)
ments, multiple levels of questions, and
found that the more effective teachers allo-
criticism (which was negatively relatedcated
to more time to academics and had
achievement). higher pupil engagement than less effect
In a study of the achievement of stu-
teachers. High success rates on tasks wer
dents of 165 second- and third-grade teach-
also found to be related to learning gain
ers conducted over 3 years, Brophy (1973)
with higher optimum success rates foun


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

for low-ability thanlearn.

for Theyhigh-ability
encouraged self-regulation
dents. teaching their students to monitor their
Knapp (1995) studied
learning, the140 grade
quality of 1-6
their work, and their
classrooms in 15 high-poverty schools
work time. They also encouraged self-regu-in
California, Maryland,
lation by and
teaching Ohio
students to over 2
use strategies
years. Knapp found that
to be goodstudents
readers and to fixin grades
problems they
1, 3, and 5 who were exposed
encountered as they to meaning-
were reading. The best
oriented reading instruction performed
teachers had high expectations 5.6
for their stu-
national curve equivalents (NCEs)
dents and masterful higher,
classroom manage-
and students in grades ment 2,
Theyand 6, 1.4
were skilled NCEs
in managing
higher, at the end of time as well
the as behavior.
school They were
year, well
students in classrooms with skills-oriented prepared for their lessons, and they men-
approaches to reading instruction. They tioned the importance of routines in terms
also studied effects in math and writing of andactivities and expectations. Finally, the
concluded that meaning-oriented instruc- most effective teachers were clear about the
tion was effective in high-poverty class- purposes of their activities and practices.
rooms. The teachers they observed teach- In the conclusion to their article,
ing for meaning wanted to give childrenWharton-McDonald et al. (1998) pointed to
more responsibility for learning, wanted toneed for additional research on the ef-
provide academic tasks that asked morefects
of of school factors and district policies
students, and sustained engagement on
in teacher practices and student perfor-
learning among children. mance. Hoffman (1991) also observed that
The work of Wharton-McDonald, Pres- there has been a paucity of research simul-
sley, and Hampston (1998) both echoes and taneously investigating both school and
extends the earlier research on effective classroom factors affecting reading achieve-
teachers of beginning reading. Because they ment. Clearly, more research operating at
worked out of a constructivist rather than a the effective school/effective teacher nexus
behaviorist perspective, they were able to is needed. Such research would, in a single
focus their work on a broader range of effort, examine school-level factors (e.g.,
teacher actions and intentions than would building climate, home-school relations,
have been possible 20 years earlier; even so,
schoolwide organization for reading col-
they were able to look closely at those as-laborative efforts) while examining class-
pects of instruction that distinguished theroom/teacher factors (e.g., time spent in
best teachers from their colleagues. Three reading
of instruction, time on-task, student
the nine first-grade urban teachers in their
engagement, approaches to word recogni-
sample were identified as most effective tion and comprehension instruction, teach-
based on their students' end-of-year reading
ers' interactive styles).
and writing achievement. These teachers In this study we attempted to wed these
demonstrated instructional balance, focus- important but seldom-integrated lines of
ing on both literature and skills. They taught
inquiry. We used quantitative and descrip-
decoding skills explicitly and also provided
tive methods to examine the programs and
their students with many opportunities practices
to in 11 moderate- to high-poverty
engage in authentic, integrated reading and
schools selected because of their dual rep-
writing activities. In contrast, the otherutation for implementing recent reading
teachers in the study either focused on reform and for promoting greater-than-
skills or whole language approaches expected
or primary-grade reading achieve-
combined the two in disjointed ways. Thement. We also examined three schools
three most effective teachers extensively
chosen because they allegedly produced
used scaffolding to help their students
ordinary achievement. However, during

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

the course of the data collection, some their reading program. Additionally, we se-
schools surfaced as more effective than oth- lected two schools that had implemented
ers (see Stringfield et al. [1997] for a similar
externally developed, nationally recognized
phenomenon). Therefore, rather than rely schoolwide reform programs-Success for
on a priori labels, we sought to pinpoint All (Madden, Slavin, Karweit, Dolan, &
and explain school (i.e., program) and class- Wasik, 1993) and Core Knowledge (Hirsch,
room (i.e., teachers' instructional practices)1987). Also included in the sample was one
factors that distinguished the most effectivehomegrown reform school with its own
schools from other schools in the study. homegrown early intervention. Operating
on the assumption that all 11 of these re-
form schools might demonstrate similar
Participants achievement profiles, we recruited three
Fourteen schools geographically dis-
typical schools for comparison purposes;
persed throughout the country took part in were schools with similar populations
the study, including schools in Virginia,
but with no history of either exceptionally
Minnesota, Colorado, and California. A high achievement or reform activity. Two of
summary of the characteristics of each the typical schools were in large urban dis-
tricts and one was in a rural area. We
school, including type of intervention and
wanted to include typical schools to p
type of schoolwide reform, if any, in read-
vide a comparison base (both in terms
ing appears in Table 1. Schools ranged from
28% to 92% poverty and included four ru- achievement and instructional practices) f
ral, four small-town, and one suburban the schools that had already undertak
school, as well as five inner-city schoolsand achieved some reform. All three of
from three large metropolitan districts. these schools were nominated by district
We started by trying to identify schools administrators as meeting our criteria in
with two characteristics: (a) those that hadterms of primary-grade students' reading
recently implemented reform programs to achievement.
improve reading achievement, and (b) those Thus we began the study with 11 exper-
with a reputation for producing higher-than- imental schools and three control or com-
expected results in reading with low-incomeparison schools. However, as Stringfield et
populations. Because we were interested in al. (1997) found in their work, not all
special interventions for students most at schools believed to be exemplary in our
risk for failure, we selected eight schools study were, in fact, found to be so. Rather
that had carefully implemented one or an-than rely on reputation, we decided to de-
other externally developed model of early fine school exemplarity empirically. We
reading intervention; our sample included used a combination of (a) gain scores from
one Book Buddies school (Invernizzi, Juel, our own classroom reading measures and
& Rosemary, 1997), two Early Intervention(b) scores on whatever achievement test the
in Reading schools (Taylor, Short, Frye, & district normally used. Based on this aggre-
Shearer, 1992), three schools with Rightgate index, four schools in the present study
Start in Reading (Hiebert, Colt, Catto, &were determined to be most effective. These
Gury, 1992), and two Reading Recoveryschools were doing as well as or better than
schools (Pinnell, Lyons, DeFord, Bryk, &others in our sample in reading growth
Seltzer, 1994). In six of these schools, the and/or doing better than average for their
interventions were set within a broader district, considering their poverty level. We
context of homegrown program reform identified
in six additional schools as moder-
reading. The other two schools in this group ately effective (neither exceptionally high
of eight had implemented early reading norin-low on the two indices that made up
terventions without schoolwide reform of our school effectiveness rating), and four


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms




0c 0 cn


0 0 0 Q) ..)Q) Q
0::(too 00 0 000o E-4
00~ -


7? C) 0 00 \ C4 f
a E- n 4- 0

0 0 0 o
g" 3 0 b
O ~~~ U
OL . NON .* ekeO

N e nO 4 e 07N o 4

0 Q0

00 *.--- 0 ;5
H0 PH4Hch UUUU H0H u
u to 0
LP I cl eo P e

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

schools as least effective-lower than other two low and two average students. If these
schools in our study on our composite in- children were no longer in schools as we
prepared to collect posttest data, we used
dex, but typical for their district in primary-
grade reading achievement. achievement and gender balance to reduce
Within each school in each of gradesthe classroom pool to four (two average-
K-3, the principal was asked to identify
and two low-achieving) students. We were
two good or excellent teachers who would only able to test four children per class in
the spring due to resource limitations.
be willing to take part in the study. We did
not include all teachers because we wanted However, from a sampling standpoint it
to focus on exemplary practice. Similarwas
to preferable to obtain fewer data points
what we discovered for schools, however,
in more classrooms than more data points
in fewer classrooms.
not all teachers were found to be exemplary,
at least according to our ratings of teacher
accomplishment. They varied widely alongData Collection Tools
the scale of accomplishment that we used to Fall and spring outcome measures. Chil-
dren were pretested in November and
characterize their practices. The principals
contacted the teachers they had nominated again in May. All tests were administered
to request their participation. A total ofby22members of our research team who had
kindergarten, 23 first-grade, 25 second- been trained for this project and the admin-
istration of these tests
grade, and 22 third-grade teachers partici-
pated in the study. All teachers were female Grade 1: In the fall, children were indi-
except for two male second-grade and two vidually tested on upper- and lower-case
male third-grade teachers. Because of the letter-name identification, phonemic blend-
ing, phonemic segmentation (Pikulski,
detail and complexity of the study, results
in this article related to pupil performance
1996), and a list of preprimer words (de-
and classroom observations are limited to scribed below).
grades 1-3. Information about the kinder- In the spring, children were individually
garten portion of the study can be foundassessed
in on a specially constructed word
a related technical report (Taylor, Pearson,
reading test and reading passages from the
Clark, & Walpole, in press). QRI-II (Leslie & Caldwell, 1995). An in-
Each principal was also asked to partici-
structional level (highest level with 90%
word recognition accuracy or better) was
pate in the study. The principal recruited the
determined for each student. Then, irre-
teachers, responded to a survey, completed
an interview, and provided demographic spective
in- of decoding ability, each student
formation about the school, including the.
was asked to read a grade 1 passage so we
number of students on free and reduced- could obtain a common fluency measure
(wcpm) for all students-the number of
price lunch, the school's overall performance
on district tests, and grade 3 standardized
words a child could read correctly in 1 min-
ute (Deno, 1985).
test achievement (expressed as a percentile).
Mindful of problems with attrition and Also, children were asked to retell each
absences, we asked teachers in the fall topassage they read. A four-point holistic
divide their students into thirds, represent-
scoring rubric (see App. A, and Colt, 1997)
ing high-, average-, and low-performing was used to score the retellings on the pas-
readers and to identify four typical average-
sage that proved to be at their instructional
achieving and four typical low-achieving level, 90%. All of the retellings were scored
children, based on teachers' perceptions by
of a single member of our research team. A
reading performance (or emergent literacy second member read and scored 15% of the
performance) who would complete pre- retellings to establish interrater reliability
tests. From this pool we randomly selected(91% agreement).


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

The reading words test Observers practiced coding
was developed
by the research team video
for theofproject
segments in
instruction until order
they had
at least 80%
to ensure that our word agreement
test with the principal
included an ap-
propriate mix of decodable
investigators on and
the coding
of teacher
student behaviors.
words. Half of the words at each Beginning in December,
grade level
were high-frequency a member
words of our team
fromconducted a 1-hour
the QRI-
II and half were decodable observation of words garnered
instruction during the basic
from an extensive review of the decodable reading program in every classroom once a
patterns introduced in the four most pop- month for 5 months. The observations were
ular basal series. The decodable words were scheduled for an hour in which reading in-
controlled to match the QRI-II words in struction was occurring. Due to resource
terms of frequency (Carroll, Davies, & Rich- limitations, we were not able to observe for
man, 1971). There were 20 words at each more than 1 hour per classroom during
grade level, preprimer through grade 3, for each monthly visit. During the observa-
a total of 100 words (see App. B). tions, the observers focused on the teacher
Grades 2 and 3: In the fall, grade 2 chil- and the children. Observational notes in-
dren were tested individually by a member cluded classroom dialogue as well as com-
of our research team on the reading words ments about general classroom activity,
test for grade 1 and a grade 1 passage from children's involvement in the lesson, and
the QRI-II. The number of words the child other events that seemed noteworthy. The
read correctly in the first minute was re- observer recorded what the teacher was say-
corded. The child's word recognition accu- ing and doing as well as what the children
racy on the passage was also recorded. Each were saying and doing during the lesson.
child was asked to retell the passage, and Additionally, every 5 minutes the observer
the four-point scoring rubric was used to recorded any of the following teacher be-
score the retellings. haviors that were observed in the previous
In the spring, children were tested in- 5-minute segment: coaching/scaffolding,
dividually by a member of the research modeling, engaging the children in recita-
team, starting with the reading words test. tion, explaining how to do something,
On the QRI, each child began with a grade telling, or engaging the children in a dis-
1 passage and continued until an instruc- cussion. A description of each of these be-
tional level was found, after which each haviors is provided in Table 2. After the
child read the grade 2 passage (to obtain the observation, the observer completed a sum-
fluency measure) if it had not been read as a mary of the lesson that required a statement
part of the procedure seeking to establish in- about each of these characteristics: overall
structional level. For each passage, we asked impression, teacher instruction and teacher-
the child for a retelling, and the retelling for student interaction, activities and materials,
the instructional level passage was scored student engagement, classroom manage-
using the four-point rubric described earlier. ment, and classroom environment.
Grade 3 children followed the same pro-
cedure as grade 2 children except for the Logs
passages read: in the fall, they read a grade We asked the teachers to keep a log of
2 passage and word list, and in the spring, daily instructional activities in the class-
they read a grade 3 passage to obtain a flu- room for 1 week in February and 1 week in
ency (wcpm) measure. April. We asked them to indicate how long
they spent on various activities, including
reading instruction (teacher-directed read-
Members of our research team at each ing of narrative and expository text; instruc-
site were trained to conduct classroom ob- tion in phonics, vocabulary, and comprehen-

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 2. Teachers' Interaction Style

Style Description

Teacher suppor
tries to perform
Modeling/ demonstrating
Teacher shows
performing a t
Engaging students in engage
Teacher recitat
which he or sh
Telling students information
Teacher gives f
on how to use the information or what to do with it.
Explaining how
Teacher providesto do
direct explanation somet
of processes
involved in a task.
Engaging Teacher leads students in a formal
students in discussiondiscus
in which
the social conventions of a discussion apply.

sion; literature circles); student independent * ways to increase the reading achieve-
ment of struggling readers;
reading; writing in response to reading;
* consequences of assessment and as-
other written composition; spelling; read- sessment training;
ing aloud to students; and other academic
activities. Teachers also indicated the group
The teacher survey embraced most of the
setting in which each activity occurred: stu-
dents working as a whole class, working samein topics and some additional ones:
small groups, or working independently.
Teachers recorded activities in 15-minute * types, frequencies, and purposes of as-
intervals and could include more than one
* types of reading/language arts helpers
activity during a time period. We divided and their activities;
the number of minutes for an interval by the
* home-school communication; and
number of activities coded to get number of* community activities.
minutes spent on an activity during that in-
terval. For example, if a 15-minute interval
Across the 14 schools, the return rate was
was coded as whole-class reading instruc-
tion, independent reading, and writing in
response to reading, we coded each activity
as occurring for a child for 5 minutes.
We interviewed all principals and at
least three teachers from each school. Prin-
cipal questions focused on the community
In April or May, the principal and teach-
and school links to parents, the principal's
ers from each school completed surveysviewde- of his or her leadership role, factors
veloped by a team of CIERA researcherscontributing
for to the school's success, chal-
a broader national survey of "beatlenges the as well as things on which the school
odds" schools. The principal survey dealt
was still working, and advice to schools
with the following topics: that wanted to significantly improve their
reading achievement. Teacher questions
* attributes of effective schools and ef- were similar but also included questions
fective instruction;
about a teacher's general approach to
* use of goals and standards;
teaching reading, behavior management
* factors/reasons for success in improv-
ing reading/language arts instruction; systems, and her or his expectations for stu-
* approaches to professional develop- dents. We transcribed the interviews and
ment; used them as a source of information for


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

writing case studies(More

for detail can also school
each be found in the
generating several nical
school variables.
report, Taylor, Pearson, Clark, & Wal-
pole [1999].) We used the data emerging
Case Studies from the observations, surveys, and inter-
views to create and operationalize both
For every school, a common outline
school- and classroom-level variables
was followed to guide our research team
in crafting a case study. A model caseConstructing school variables
study was also provided to create someSchool effectiveness rating: To categorize
schools as most effective, moderately effec-
common expectations for content, format,
tive, or least effective, we used two school-
and depth. Major topics within each case
study included based measures: (a) students' growth on
project measures of reading and (b) stu-
(1) school demographics; dents' performance on district measures of
(2) history of the school;
reading. These two scores were combined
(3) primary-grade reading performance
data; to create a general rating of effectiveness
(4) community/home/school relation- (see Table 5 and App. C).
ships; School efforts to link to parents: An initial
(5) school factors (leadership, effective reading of the case studies indicated that
schools/school change factors, profes-
schools varied in the extent to which they
sional development, format of school
reading program); reached out to parents. Schools were judged
(6) grade-level classroom practices (in- to be high on the linking to parents factor if
struction, curriculum, student en- a large percentage of teachers communi-
gagement, classroom management, cated regularly with parents, had an active
classroom environment);
(7) other factors believed to be enhanc- site council in which parents helped to
ing beginning reading achievement; make school decisions, and used practices
(8) challenges; and such as focus groups, phone surveys, or
(9) advice to other schools. written surveys to find out parents' needs
and concerns.
Case studies were written by the principal
Schools were judged to be average or
investigator or by the research team mem-
low on the linking to parents measure if
ber who had spent the most time in a par-
ticular school.
fewer of the above factors were present. Six
schools were determined to be high on the
linking to parents factor, five were deter-
Creating Variables for Data Analyses mined to be average on this factor, and
Overall strategy. Quantitative and de-
three schools were determined to be low.
scriptive analyses were conducted using The two raters evaluating schools on this
multiple sources of information for bothfactor achieved 93% agreement in their cat-
types of analyses. Analyses were conducted
at the school and classroom level. To ex- Systematic, internal assessment of pupil
amine the natural variability within our
progress: A school was coded as systemati-
sample, we developed empirically driven
cally assessing pupil progress if at least two-
indices of effectiveness for both schools thirds
and of teachers on the survey perceived
teachers, grouped the schools or teachers this to be an attribute of their school and if
into categories of effectiveness (most, mod-
comments in the case studies and/or inter-
erately, and least), and then investigated views
the supported this perception. Across
systematic differences among schoolsschools,or all but four were coded as system-
teachers on other sets of variables. The pro-
atically assessing pupil progress. These were
cedures and criteria used to determine these
not externally imposed standardized testing
variables are summarized in Tables 3 andsystems;
4. they were internally developed sys-

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms
cL a) nC

6 3

C) L44 Z, -, w .
ccc ?t>
a) CL4
U ?n
-4--4 4-


o k
04 0i s)
C4 7)-a
C 4m 4 -4
2Ue~ ?d
3 O ~
V e~
u u
(d ()d (dU U


eOi -,c~a

V ~k~ V
( rCI c
-g & ~ l
Y ~~,?
C Eb.0u

) a
Z U ~E ,ua
O~ k t ) c k( ~

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

tems for monitoringgrades 1-3 in this project,

individual student 22% received
pro- a
gress within a schoolwide
rating of 5, curriculum.
21% a rating of 4, 35% Thea rating
data from these systems were
of 3, 14% a rating used
of 2, and fre-
8% a rating of 1.
quently by individual teachers
The mean rating was 3.35to
(SD =make
1.22). One
grouping and instructional research team member rated all teachers,
Building (school) communication and interrater
and a second rated a 25% sample; col-
laboration: In a number of the interviews agreement was 95%.
and surveys, staff commented about either Student time on task: From the observa-
good or poor collaboration and/or com- tions, we searched for comments about chil-
munication among teachers. Schools judgeddren's on-task behavior. Most helpful in
to be high on the building communication/this regard were the summaries that each
collaboration scale had a high mean scoreobserver completed at the end of each ob-
on teachers' perceived building communi- servation session; one of their tasks was to
cation rating (from the teacher survey) and
summarize and point to evidence to sup-
positive comments in the case study or port
in any conclusions about students' level
of on-task behavior. Based on these sum-
interviews about collaboration among teach-
ers within and across grades. Schools judged
mary comments a teacher received a high
to be average or low on the building com-(3), average (2), or low (1) rating in main-
munication factor had fewer positive and/taining students' on-task behavior. One
member of our research team rated all
or more negative comments in the case study
or interviews. Two raters achieved 86% teachers, and another team member rated
agreement on this judgment. Across 25% the of
14the teachers on this dimension.
schools in the project, six schoolsAcross werethe pairs of ratings for each teacher,
judged to be high on the building commu-
there was 100% agreement. Across teachers,
nication factor, five were judged to be the mean rating was 2.30 (SD = .76), with
age, and three were judged to be low.48% of teachers receiving a rating of 3
Use of an externally developed early (high),
read- 34% a rating of 2 (average), and 18%
ing intervention: Within each grade alevel a of 1 (low) in level of on-task student
school was coded as either having behavior.
or not
having an externally developed earlyPreferred
in- interaction style: As reported
tervention in place. Out of the 14 schools,
earlier (under observations), at the end of
10 had an externally developed interven-each 5-minute segment during the class-
tion, five across two or more grades, room observations, observers coded in-
in grade 1 only, and one in grade stances 3 (in aof interactions observed during that
grade 3-6 school). Four schools had no ex-
segment, using these categories: coaching/
ternally developed intervention in place.scaffolding, modeling, engaging students in
Constructing classroom (teacher) recitation, engaging students in discussion,
explaining how to do something, or telling
Home communication: On the survey students information (these activities are
teachers indicated how often they com- described in more detail in Table 2). The to-
municated with parents and in what ways. tal number of times a teacher was coded as
The areas in which there appeared to beengaging
the in each of these behaviors (more
most variability were the frequency with than one was possible within each segment)
which teachers reported (a) calling home,was calculated, and the behavior coded
(b) sending a letter or newsletter home, or frequently across all five observations
(c) sending a traveling folder home. A five-
was determined to be the teacher's pre-
point scale was used to rate teachers on ferred
the interaction style. Across teachers in
extent to which they communicated with grades 1, 2, and 3, 24% had a preferred in-
parents. Across the sample of teachers in
teraction style of coaching, 31% had a pre-

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms


~m ZX
0 0

k~ cr
H 0 0


0 w0
0 0*j.*
tt 4 %k
Lr m

0 00 0~-

- S


Zln c' o (flG e)




0 0


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms


C4 L C) 75 0 a

b3 O2~ 0 ~ C~3 0U ,

u P-.oP-.

CI A aa

- ,5 ~o

C6 075
0Om 0
L N~


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 5. Z-Scores for School Effectiven

Z-Score for Sum of

Standardized Residual Mean z-Scores School Effectiveness School Effectiveness
School Composite z-Scorea Percentileb (Cols. 2 and 3) Scorec Rating (Three Levels)d
1 - .74 - .22 - .96 - .58 1
2 .74 .53 1.27 .77 3
3 - .22 - .36 - .58 - .35 2
4 - .01 - .52 - .53 - .32 2
5 .03 - .91 - .88 - .53 1
6 1.33 2.04 3.37 2.04 3
7 - 1.39 - 1.28 - 2.67 -1.62 1
8 - .41 1.17 .76 .46 2
9 1.16 - .56 .60 .36 2
10 1.74 - .39 1.38 .82 3
11 .28 1.45 1.73 1.05 3
12 - .39 - .36 - .75 - .45 2
13 - 1.91 -.75 - 2.56 -1.56 1
14 -.32 .15 -.17 -.10 2

aFrom z-scores (by g

bOn district standardized reading test (controlling for poverty).
cStandard score (from col. 4).
d3 (most effective if > .50 SD above mean of 0); 2 (moderately effective if < .50 SD above mean but > -.50
SD below mean of 0); 1 (least effective if < -.50 SD below mean of 0).

ferred interaction style of engaging studentsunknown words as they were reading text
in recitation, 39% had a preferred interaction(b) providing explicit phonics instruction,
style of telling students information, and and (c) practicing words.
6% had a preferred style of modeling. To Coaching in word recognition strategies
evaluate the trustworthiness of these rat- involved prompting children to use a vari
ings, three members of the research team ety of strategies as they were engaged in
read the observational notes and rated a reading during small-group instruction or
sample of 25% of the teachers; they agreed one-on-one reading time. They fell into sev
with the original coder on 82% of theeral subcategories, as listed below:
nal codings. If and when mismatches oc-
curred, we reverted to the code most fre-
Metacognitive Dialogue on Strategies
quently marked by the classroom observer. A teacher reviews independence in
Approaches to word recognition and com-
using word recognition strategies with
prehension instruction: The data from thehis
ob-students: "The point is to be able to
read on your own this summer. What if
servations were analyzed to determine how
you come to a big long word? Yes, sound
teachers provided word recognition and
it out. What else can you do? Yes, you
comprehension instruction. A numbercanof
twist it a little (e.g., try a different
approaches were coded on a frequency
vowel sound in 'terrible'). Also you can
scale, and the frequencies were collapsed
ask yourself if it makes sense. And if you
try these things [and still don't know the
into three categories: frequently (observed
word], then what do you do? Yes, skip it,
in two or more of the five observations),
or what else? Yes, you can ask someone."
occasionally (observed in at least one ob-
Metacognitive Review of Strategies Used to
servation), and never. The three word rec-
Figure Out a Word
ognition approaches that occurred most
After a child came up with "squirt"
frequently were selected for more elabo-
while reading, the teacher asked, "How
rate analysis; these were (a) coaching chil-
did you figure out 'squirt,' Tom?"
dren in the use of strategies to figure outTom: I sounded it out.


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

Teacher: You could also look at the teacher uses a sliding device so the first
picture. letter can be changed. The children con-
Tom: Also make sure it makes sense. tinue to read: Dan, fan, man, can, pan,
ran. "Good for you! Excellent. One more
Praise for Use of Strategies family." The children read: Pig, dig, rig,
After a child has read, the teacher big, twig.
says to the group, "I noticed that Mara
got stuck and skipped it and read around Word Study
it and then came back to it. That's good A child sorts picture cards by first let-
thinking." ter: lamp, letter, game, gate, girl, October,
otter, officer. "Let's check and see if you
Prompts to Figure Out Words: General got them right. "g" makes the /g/
A child is stuck on a word while sound. Game, October, gate, girl. There
is one that doesn't belong. Let me say
reading, and the teacher asks, "What
could give you a clue on that word?"them. Do you hear one that doesn't be-
long? Right, where would October go?"
Prompts to Figure Out Words: Specific
The teacher is helping a group as
Practice on sight words involved teach-
they are reading aloud. "Whoa, back up
there. Frame the word with 'i-n.' What ers using flash cards, a pocket chart, or a
is the first sound? What is the second word wall to review words the students
sound? What's the word?"
were expected to recognize instantly as
sight words.
Explicit phonics instruction included
work on a chart, whiteboard, worksheet, or
Flash Cards
word cards dealing with word study, word The teacher reviews sight words with
his group. He gives them the word card
families, introducing or comparing phonic
elements (i.e., er, ir, and ur all have the sameif they say it first. "These two look like
they rhyme (he points to goes and does),
sound), making words (Cunningham but & do they?"
Cunningham, 1992), writing words, and
reading words with a particular phonic ele-
ment in isolation. The following were the The teacher puts a chart of No Ex-
cuses words on the board: a, an, and, for,
most common approaches observed: in, will. She calls on individual children
to read the words. "Yesterday we read
this word in our story: will. Now use
Working on Phonic Elements your white board and write 'will.' "
"There are a lot of words that don't
have the long or short vowel sound be-
cause they have a bossy letter. 'Ur' says For comprehension instruction, eight in-
/ur/ in 'hurt'. In 'born', 'or' says /or/. structional practices were observed and
In 'her' if it was a long e it would say
/here/ but 'er' says /er/. Can anyone coded: doing a picture walk; asking for a
think of an /ur/, /or/, or /er/ word?" prediction; asking a text-based question;
asking a higher-level, aesthetic-response
Making Words question; asking children to write in re-
Children get letter cards: a, I, g, k, n,
p, r. "Let's start with two letters to make
sponse to reading (including writing an-
'in'. Change one letter to make 'an'. Add swers to questions about what they had
one letter to make 'pan'. Rearrange these read); doing a story map; asking children
to make 'nap.' " The children continue to retell a story; and working on a compre-
until they end up with a word from the hension skill or strategy. For further anal-
story, "parking."
ysis, we focused on those categories for
Working on Word Families which 10 or more teachers were frequently
"Let's warm up with a few word fam- observed using the strategy. The practices
ilies." Children read: cat, hat, fat, mat. The that met the 10-teacher criterion were ask-

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

Results and Discussion

ing text-based questions, asking higher
level questions, and asking
The results children
are organized to
by level of anal
write in response to what
ysis. First wethey had
report and read.
discuss result
One member of the research team coded
largely descriptive, at the school level. Sec
all teachers, and a second member coded ond, we use both descriptive and statistic
25% of the teachers, achieving 96% agree- tools to examine the variation in instruc-
ment on coding of the approaches to word tional practices among teachers within the
recognition instruction and 97% agreement levels of school effectiveness. Third, we re-
on the coding of the approaches to compre- port and discuss the variation we found in
hension instruction.
instructional practice as a function of teacher
Teacher accomplishment rating: Two ex- accomplishment. Throughout the results we
perts in teacher supervision at the elemen-
refer to the most effective schools by pseu-
tary school level read all teacher observa-
donym: Hilltop, Stevenson, Wheeler, and
tions. One expert was a whole language
advocate, and the other was a reading Analyses related to student performance
skills advocate. They used the checklist of
at the school and all analyses at the teacher
elements of effective instruction from the level were based on classroom mean scores.
principal survey (based on the work of An-
Mean classroom performance measures by
derson et al., 1979; Barr & Dreeban, 1991;
school effectiveness and teacher accom-
Hoffman, 1991; Pressley, Rankin, & Yokoi,plishment are reported in Tables 6-8.
1996; Roehler & Duffy, 1991; Wharton-
McDonald et al., 1998; see App. D). School-Level Analyses
Based on a thorough reading of the ob- Linking to parents. Parent links were
servations and summaries, each rater rated positively and statistically significantly re
each teacher separately on each item on the lated (r = .73) to the school effectiveness rat-
list, providing a judgment about whether ing and to all measures of student growth
there were many, some, or only a few in- fluency (r = .60), retelling (r = .37), and
stances of that attribute in the data set (or reading
if words (r = .41) (see Table 9).
there were no data, the rater coded the itemThe most effective schools reported more
as "could not determine"). The two raters links with parents than the moderately an
least effective schools. Three of the four most
then arrived at an overall rating (on a three-
point scale) based on the mode (the most effective schools reported having an active
commonly occurring point on the scale site
of council on which parents served with
many, some, or only a few) of the individ-
teachers and other school staff and helped to
ual items. For the overall rating, the two
make decisions concerning school practices.
raters demonstrated 80% agreement. Four In of six moderately successful schools
those instances in which disagreements also
oc- reported having an active site council,
curred, a third judge conducted an inde-
but only one of the four least successful
schools reported having such a body in
pendent rating. In all instances, the third
judge agreed with one of the other two
judges, and that rating was used in the fi- Additionally, the most effective schools
nal analysis. Across teachers in gradesreached
1- out to parents in other ways.
3 in our sample, 42% were identifiedWheeler
as scheduled focus groups to learn
demonstrating many of these elements how to better meet parents' and students'
(most accomplished), 34% demonstrated needs. Woodlawn had regular focus groups
some of these elements (moderately and ac- conducted phone surveys to determine
complished), and 24% demonstrated only parents' concerns and needs. Stevenson
a few of the elements of effective instruc- school officials sent a written survey home
tion (least accomplished). to parents, and the principal attempted to


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 6. Grade 1 Means (and Standard Deviations) fo

Effectiveness Rating and Teacher Accomplishm

Fall Spring

wcpm on
Number of Phonemic Preprimer Reading Grade 1 Retelling at
Rating Teachers Letter ID Awareness Words Words Passage Reading Level
Least effective 8 47.95 9.38 9.90 56.28 34.68 2.52
(3.51) (1.71) (4.58) (14.40) (12.09) (.64)
Moderately effective 8 44.75 6.56 7.38 55.87 41.19 2.34
(5.82) (3.85) (4.44) (16.89) (14.99) (.30)
Most effective 7 45.80 7.48 9.94 68.29 50.75 2.51
(3.97) (2.51) (3.87) (9.01) (17.03) (.35)
Least accomplished 3 43.00 4.52 4.58 41.58 25.58 2.17
(1.64) (4.68) (2.18) (14.63) (11.22) (.14)
Moderately accomplished 7 45.83 7.11 7.67 66.18 44.40 2.38
(4.42) (1.96) (3.80) (17.70) (15.96) (.37)
Most accomplished 12 46.74 8.57 10.67 60.48 44.44 2.59
(5.13) (2.87) (4.23) (10.32) (15.39) (.52)

NOTE.-wcpm = words correct per minute.

TABLE 7. Grade 2 Means (and Standard Deviations) for Pre- and Post
by School Effectiveness Rating and Teacher Accomplishment Rati

Fall Spring

wcpm on wcpm on
Number of Grade 1 Grade 1 Reading Grade 2 Retelling at
Rating Teachers Words Text Words Text Reading Level
Least effective 8 13.88 50.46 79.59 66.30 2.56
(3.59) (18.92) (5.88) (19.24) (.69)
Moderately effective 10 15.88 59.77 86.98 73.35 2.99
(2.18) (13.07) (6.04) (14.65) (.36)
Most effective 6 14.42 47.98 84.58 70.17 3.06
(3.46) (17.31) (7.99) (13.18) (.40)
Least accomplished 8 13.91 56.83 80.22 64.47 2.66
(2.50) (13.38) (8.09) (17.37) (.78)
Moderately accomplished 8 15.20 55.63 84.82 71.40 2.99
(3.58) (14.38) (7.00) (17.20) (.39)
Most accomplished 8 15.44 48.68 86.72 74.75 2.95
(3.12) (21.31) (4.73) (11.80) (.30)

NOTE.-wcpm = words correct per minute.

communicate with all parents

home in regularly.
English or Spanish s
could read to their
None of the moderately successful orchildren.
and Stevenson also
successful schools reported engaging in any cited successful at-
of these practices. home reading partnerships, all with h
Hilltop, another of the parent
effectivein which parents re
schools, had developed anularly
to their children read.
partnership in which books were
When asked sent
about reasons for their suc-

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 8. Grade 3 Means (and Standard Deviatio

Effectiveness Rating and Teacher Accomp

Fall Spring

wcpm on wcpm on
Number of Grade 2 Grade 2 Reading Grade 3 Retelling at
Rating Teachers Words Passage Words Passage Reading Level
Least effective 7 16.95 73.19 90.99 85.70 2.80
(2.15) (14.48) (3.76) (18.98) (.29)
Moderately effective 2 15.27 75.58 89.89 91.13 3.06
(3.74) (22.79) (8.14) (29.13) (.43)
Most effective 3 16.81 72.58 96.25 105.38 3.31
(2.03) (8.82) (1.17) (16.92) (.47)
Least accomplished 5 15.95 71.70 90.60 88.49 2.96
(2.27) (11.67) (3.91) (27.83) (.38)
Moderately accomplished 9 16.28 75.88 91.44 92.08 3.14
(3.67) (17.37) (8.21) (23.59) (.50)
Most accomplished 7 16.20 77.11 92.33 94.58 2.92
(3.23) (23.06) (46.48) (31.35) (.35)

NOTE.-wcpm = words correct per minute.

TABLE 9. Correlations for School Effectiveness Rating and S

Z-Score Z-Score Z-Score

School Effectiveness Residual Word
Score (Class Mean) (Class Mean) (Class Mean)

Links to parents .73 .41 .60 .37

p <.01 <.001 <.001 <.01
n 14 70 70 70
Systematic ass
of pupil progress .42 .21 .53 .37
p >.05 >.05 <.01 <.01
n 14 70 70 70
Building communication .37 .07 .43 .35
p >.05 >.05 <.001 <.01
n 14 70 70 70
Research-based intervention
in place .00 .14 .02 .05
p N.S. N.S. N.S. N.S.
n 14 70 70 70

cess, Steven
to com
have i
tions, Alth
them feel
does n
the comm
ing closely
in as many
an ex
explaining a


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

exception in this respect, countabilitymade(to one'stime

colleagues) while
to call all homes justproviding teachers with a usefulparents
to compliment bench-
on the academic or social achievements of mark on each student's progress. The pub-
their children. lic sharing of the data was important in all
Systematic internal assessment. Sys-four schools. The principal at Hilltop re-
tematic assessment of pupil progress figured lated that the teachers had learned how to
prominently in our findings. Systematic as-confront data, to keep data in front of them,
sessment was related to students' growth into use data to identify specific strategies to
reading fluency (r = .53, p < .01) and to their help struggling readers, to provide support
retelling performance (r = .37, p < .01). Allin the implementation of strategies, and to
four of these schools used some form of sys- align major school events and celebrations
tematic assessment of student progress. Inaround the meeting of schoolwide goals. In
all four cases, this meant that all the teach- other words, the staff at this particular
ers in the school regularly (at least threeschool had learned how to make perfor-
times throughout the year) administeredmance data a useful ally rather than a cause
some sort of common classroom-based as- of constant alarm or frustration.
sessment to all students and shared the Building communication and collabo-
ration. The building communication and
information about classroom-level perfor-
mance with the principal and fellow collaboration
teach- rating was positively related
ers. Wheeler implemented four assess-
to the fluency measure (r = .43, p < .01)
ments across the school year, relying andon a
retelling (r = .35, p < .01). Five of the
fluency measure (wcpm), sight words, and
10 moderately effective or least effective
letter identification in grades 1 andschools
2, andrevealed concerns about communi-
letter identification and concepts aboutcation
print across teachers and program articu-
in kindergarten. Woodlawn used a fluencylation across grades. The interviews with
measure as a curriculum-based indicator staff in one moderately and one least effec-
tive school revealed several instances of
five times a year (three times in first grade).
Hilltop used an informal reading inventory
negative communication and collaboration
three times a year. Stevenson used an including
infor- low morale among teachers du
to factions among the staff and perceived
mal inventory (grade 1) or basal tests (grades
1-2) three times a year as well as a devel-
lack of cooperation among teachers. In on
opmental spelling test three times a yearof the inleast effective schools, all of the teach
ers rated communication of ideas across
grades 1-3 and a words-in-isolation-and-in-
context test in grade 1 every 6 weeks. teachers
Four as moderate to low in their school.
of the six moderately effective schools also
In another least effective school, working as
reported having systematic assessment of across the entire school was high-
a team
pupil progress in place, and two of the fouras something the school had just be-
least effective schools had an assessmentgun to work on; seven of nine teachers in
tem in place. that school rated the presence of commu-
We emphasize that these were nication as moderate to low in their school.
curriculum-based, classroom assessments In contrast, teachers in all four of the
intended to provide information for moni- most effective schools reported collabora-
toring individual student progress and to tion within and across grades as a reason
shape individual and classroom (and occa-for their success. Factors such as peer coach-
sionally schoolwide) curricular and instruc- ing, teaming within and across grades,
tional decisions; they were not external,working together to help all students, and
accountability-focused assessments. Instead program consistency were mentioned as
of external accountability, these classroom- aspects of collaboration that teachers val-
level data provided a form of internal ac- ued in these most effective schools. In con-

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

of reading/language
trast, in the moderately and least arts instruction
effectiv an
schools, concerns about
practice. A program
Title 1 teacher or compati
aide worked
bility, instructional
classroom for 50 minutes and com
each mornin
The Title 1 teachers returned to
mon instructional terminology the first- mor
and second-grade
prevalent. One of the teachers classrooms for 30 min-
at Wheeler
one of the most effective schools,
utes in the afternoon to providesummed
or one-on-two
it up this way: "Teaming withhelp to other
struggling readers.
important. You can't do
At Hilltop, it by
teachers yourself
also used a collabora-
Teaming also builds tive
sensebut in this
ofcase the children who
were struggling most in reading left the
If the children see us working together an
getting along, that classroom
means during the 21/2- to 3-hour literacy Th
a lot to them.
children also get to see
block other
to receive teachers
small-group an
instruction for
get to know them. 45 That builds caring
minutes. This small-group instruction, and
delivered to two or three
community." This sentiment was children at a time, b
teachers at Hilltopwas compatible with
School, who the instruction
suggeste the
that peer coaching andreceived
children collaboration,
in their regular class-be
cause they led to schoolwide
buy-in, were
A common element of all four of these
key factors in their success.
Collaboration played an approaches
important was the focus
role on
in the delivery of reading instruction
small-group instruction. in
Additionally, in all a
of the most effective schools. All four of the four of these most effective schools, teachers
most effective schools had reorganized their
spent considerable time, averaging 134 min-
instructional delivery system within the utes a day, on reading instruction. In inter-
past few years to make use of a collabora-views, teachers in three of these schools
tive model for reading instruction. In threementioned that reading was a priority at
schools, this meant that special personnel-their school; their time allocation to reading
a Title 1, reading resource, or special edu-is strong evidence of this commitment.
cation teacher-went into the classroom for Early reading interventions. The nature
an hour a day to help provide instruction of the interventions implemented in the
for small, ability-based groups. schools in our sample is presented in Table
Wheeler deployed a resource teacher 1. inThe use of interventions was not limited
the classroom with the classroom teacher to the most effective schools; in fact, the pro-
during reading time for 1 hour and anportion
aide of schools with externally developed
for a second hour of reading. Children re-
interventions did not vary dramatically by
ceived guidance in small groups with a of effectiveness: most = 3/4; moderate
teacher or a resource teacher, or with =the 4/6; least = 2/4. However, the nature of
aide in one-on-one settings. At this school,
the interventions and their implementation
did vary by effectiveness. In the three most
21/2 to 3 hours a day were spent on reading/
language arts instruction and practice. effective schools with interventions, they
Clearly, helping all children learn to read
tended to be small group, locally or region-
was a priority. Woodlawn sent a reading ally developed, and implemented across the
specialist and a special education teacher
primary grades; in the moderately or least
into the classroom to work with small effective schools, interventions tended to be
groups along with the classroom teacher for
one-on-one, either regional or national in
an hour each day. In this school, 21/2 hoursand implemented in grade 1 only.
a day were spent on reading/languageTeachers arts and administrators in the most
instruction in grades 1 and 2. At Stevenson,
effective schools felt strongly that the early
using a similar push-in collaborativeinterventions
model, in place in their schools were
children also received about 21/2 hourskey
a day
to their success. In the words of one Hill-


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

top teacher, "Our early intervention

in the survey makes
about preferred approaches to
all the difference in professional terms development.
of taking In three care
of the of
kids' needs." One Woodlawn teacher
most effective schools, a majority of put
teachers it
this way: "With our rated early "visits reading interven-
to schools with innovative pro-
tion kids experience grams success
followed by andsharinglook for-
of observations
ward to the time [in with the intervention]
colleagues" as an effective approach to ...
because of the teacher and the success." professional development that had been
Teachers or administrators from the mod- used in their school. Also, in three of the four
erately or the least effective schools did
most effective schools, a majority of teachers
not express such enthusiasm or regard for rated "district- or school-sponsored year-
their interventions. Although the sample long workshops" or "graduate-level
of schools in this study was limited, the courses" as an effective approach to profes-
data nonetheless point to reading success in sional development, which had been used in
the most effective schools achieved through their school.
a combination of regionally (or in the case Summary of school factors. Across the
of one school, locally) developed small-four most effective schools in this study,
group interventions set within a homegrownreading was clearly a priority. The teachers
school reform model (see Table 1). Thisand principals considered reading instruc-
stands in contrast to a national push (e.g., thetion their job and they worked at it. They
17 programs listed in the Comprehensive worked together, worked with parents, and
School Reform Demonstration Program leg- worked with a positive attitude to reach the
islation; see Herman, 1999) for off-the-shelf goal of all children reading well before they
reading intervention programs and schoolleft the primary grades. They were able to
reform models (American Federation of reach consensus on schoolwide monitoring
Teachers, 1997; Herman & Stringfield, 1997; systems, a collaborative approach for deliv-
Slavin & Fashola, 1998; Wang, Haertel, & ering reading instruction, and professional
Walberg, 1998). development, with the constant goal of im-
Ongoing professional development. proving an already effective reading pro-
During interviews, teachers and/or princi- gram.
pals in three of the four most effective
schools cited a yearlong staff development Instructional Practices within Levels of
effort related to their early intervention pro- School Effectiveness
gram as responsible for their success, indi- Multivariate analysis. To investigate
the relation between school effectiveness
cating that it helped them "stay in a learner
mode," and "all be of one accord." For ex- and classroom instruction, we initially con-
ample, at Hilltop the teachers took two or ducted a multivariate analysis of variance
three yearlong courses (at the beginner, ex- (MANOVA) with the school effectiveness
perienced, or advanced level) on the phi- rating serving as the independent variable
losophy of the intervention and the imple- and eight teacher variables serving as out-
mentation of the strategies within the come measures (see Table 10). To ensure
classroom. Teachers were encouraged to that we were focusing on potentially pow-
take at least two of the three courses, which erful variables, only those classroom factors
met for 2 hours a month during the school that were statistically significantly related
year. There was also a time for every class to one or more of the measures of student
participant to meet with a peer coach for 45 or teacher accomplishment (school effec-
minutes once every 2 weeks. tiveness rating; teacher accomplishment
In addition to the information about pro- rating; fluency, retelling, or reading words
fessional development related to the early measure) were included in the MANOVA
reading interventions, teachers were asked (see Table 11). A statistically significant

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 10. ANOVAs on Teacher Factors with

Teacher Factor Number of

and School Ratinga F p Mean SD Teachers
Home communication (mean is
based on 5-level rating)b 5.25 .008
1 3.09 1.22 22
2 2.95 1.12 27
3 3.75 1.03 19
Time on-task (m
based on three-level rating) .82 N.S.
1 2.18 .73 22
2 2.24 .74 29
3 2.40 .84 19
Time in small-
(mean is minutes/day)b 9.63 .0002
1 37.94 27.19 20
2 26.00 20.16 24
3 59.82 28.46 19
Time in whole-class instruction
(mean is minutes/day) 2.20 N.S.
1 30.20 18.42 20
2 36.66 20.94 24
3 24.86 14.64 19
Time in independ
(mean is minutes/day)c 4.24 .02
1 18.63 9.43 20
2 27.04 13.30 24
3 28.14 10.50 19
Preferred interac
Coaching 2.32 N.S.
1 .23 .43 22
2 .21 .41 29
3 .47 .51 19
Telling 2.01 N.S.
1 .41 .50 22
2 .41 .50 29
3 .16 .37 19
Recitation .17 N.S.
1 .32 .48 22
2 .34 .48 29
3 .26 .45 19

al = least ef
b3 0 2 = 1.
c3 = 2 #1.

MANOVA, F(14, 108) = 2.56, p < .01, led in the most effective schools
were more likely to send a letter or news-
us to conduct follow-up univariate analyses
of variance (ANOVAs; see Table 10). letter home weekly and call home regularly
Univariate analysis than teachers in the other schools. Teachers
Home communication: The follow-up in the
AN- most effective schools had a higher
OVA on home communication was statis- mean score on the home linkage scale than
tically significant, F(2, 65) = 5.25, p teachers
< .01. in either the moderately or least
effective schools. In three of the four most
Tukey post hoc tests revealed that the teach-
ers in the most effective schools communi- effective schools, more than half of the
teachers reported calling home at least
cated more with parents/caretakers than
teachers in the moderately effective or least
once a month. In only one of six moderately
effective schools. effective schools and one of four of the least


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 11. Correlations for School Rating, Teacher R

Growth Scores, and Teacher Factors

Z-Score Z-Score Z-Score

Teacher School Teacher Residual Fluency Residual Retelling Residual Words
Factor Rating Rating (Class Mean) (Class Mean) (Class Mean)
Teacher rating .23 .27 .10 .18
p .05 <.001 N.S. N.S.
n 71 69 69 69
Home communication .26 .27 .06 .25 -.09
p .03 .02 N.S. .04 N.S.
n 68 68 66 66 66
Time on-task .14 .79 .21 .17 .18
p N.S. <.001 N.S. N.S. N.S.
n 70 70 68 68 68
Time spent in small-g
instruction .30 .31 .41 .10 .22
p .02 .02 <.01 N.S. N.S.
n 63 63 61 61 61
Time spent in whole
instruction - .11 - .41 -.03 -.22 .06
p N.S. <.01 N.S. N.S. N.S.
n 63 63 61 61 61
Time spent in indep
reading .32 .10 .33 - .11 .20
p .01 N.S. <.01 N.S. N.S.
n 63 63 61 61 61
Preferred teaching s
Coaching/ scaffolding .20 .40 .07 - .18 - .03
p N.S. <.01 N.S. N.S. N.S.
n 70 70 68 68 68
Telling information - .20 - .53 - .27 - .14 - .01
p N.S. <.01 .03 N.S. N.S.
n 70 70 70 70 70
Recitation - .04 .04 .18 .24 .07
p N.S. N.S. N.S. .05 N.S.
n 70 70 68 69 68
Time spent writing
response to reading -.13 .03 -.18 .03 .01
p N.S. N.S. N.S. N.S. N.S.
n 63 63 61 61 61
Time spent reading aloud .17 .14 .19 -.07 .14
p N.S. N.S. N.S. N.S. N.S.
n 63 63 61 61 61

OVA revealed no schoo

effective statistically significant
school effectiveness effect
indicate that for the student
quently. In rating, F(2,
one 67) = .82, p > .05. o
teachers made a concerted effort to call As we discuss later, this result is to be con-
home with positive comments, and 51% of with the finding for levels of teacher
parents who were asked said they had accomplishment
re- (when teachers who ex-
ceived such calls. Looking at both teacher
hibit accomplished characteristics are con-
and school factors, the analysis of links to independently of the school in
families suggests that personnel in thewhich
mostthey work).
effective schools made a more concerted ef- Time spent in small-group and whole-class
fort than in other schools to reach out to instruction: The ANOVA on time in small-
parents. group instruction revealed a statistically
Student time on-task: The univariate AN- significant effect for school effectiveness,

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

F(2, 60) = 9.63, p < In all of the Tukey

.001. most effective schools,
post the ho
tests revealed that students of teachers in
basis for forming the small groups was per-
ceived ability, an observation suggesting
the most effective schools spent more time
that these teachers were more concerned
daily in small-group instruction (M = 59.02
minutes per day) than students of teachers
about meeting students at their instructional
in the moderately effective schools (M level
= than they were about any damage to
26.10 mpd) or the least effective schools
children's self-worth that might accrue from
(M = 37.94 mpd). The one-way ANOVA
being a part of a group socially and person-
for school effectiveness rating on time spentally sanctioned as the low group. Even so,
in whole-class instruction was not statisti- the teachers in the most effective schools
cally significant, F(2, 60) = 2.20, p > .05.
were very aware of the need to.make sure
However, even the students in the most ef-that the groups were flexible, that students
fective schools, who were spending an hourmoved to another group when their per-
a day in small groups, were in whole-classformance (as measured by the internal
instructional activities an average of 25 min-school-based monitoring system) merited
utes per day. movement. The importance of schoolwide
In addition to differences in the amount monitoring cannot be underestimated in
of time spent in small-group instruction by this regard. These data provided teachers
school effectiveness, the ratio of small- with regular, recurring opportunities to re-
group to whole-class instruction is impor- flect on the validity of their instructional
tant to consider. In each of grades 1, 2, andgroupings and modify membership accord-
3, children in the most effective schools ingly. The principal at Stevenson talked
spent more time in small-group than in about a mental change that occurred when
whole-class instruction. In grades 1 and 2, her entire school became a Title 1 school.
the small-group/whole-class ratio was 2/1 Teachers began to talk more about chil-
(see Table 12). dren's needs during the year and made
When asked on the survey to select the changes in students' reading group place-
four most important factors for improvingment. "A reading group was no longer a
struggling readers' achievement, 83% of theyearlong placement for children, particu-
teachers in the four most effective schools larly for the lowest children." At Woodlawn,
selected small-group instruction as an im-teachers also voiced commitment to the idea
portant factor. Additionally, in two of theof flexible instructional-level groups. They
schools, teachers mentioned the focus on used running records and periodic mea-
small-group instruction as a factor contrib-sures of words read correctly in 1 minute to
uting to their success. "Small, flexible move children to a higher group whenever
groups at students' instructional level are they could. Furthermore, in three of the four
important. They need to be coached at their most effective schools, early reading inter-
instructional level" contended a Hilltop ventions were in place across the primary
teacher. Comments about the virtues of grades to provide high-quality, special as-
small-group instruction were commonly
sistance to children who were struggling to
heard in the interviews at the most effective
learn to read. Recall also that in these most
schools, such as these remarks by a Wood-
effective schools, students also averaged 25
lawn teacher-"Small groups or one-on- minutes a day of whole-class instruction in
one every day really makes a difference"-
which they were interacting across ability
and a Wheeler teacher-"Small groups levels.
crucial. Children are more likely to succeedTime spent in independent reading: The
when they are in two groups of six with two on time spent in independent
teachers than when there are 12 children reading was statistically significant, F(2, 60)
with one teacher."
= 4.24, p < .05. Tukey post hoc tests re-


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms
1.\ Cf C) r-- \.o 00 ON U)N ONLO CD m 0 NC ON xU 0\ ?N C)Lr O x ON x UIt r-- 4ON
N N 6tLOO r--4 '.0 NN t X NN ON N C 00 1.0~ 00 C) Cf) \.o 1.0I Cf) 00 Id r--
c? ,6 u cfi o6 6 ci o? u? o? t< t- o? c?i cfi t- \.6 4 cji c6 6 c? t. 4 4 o? cfi
C) N '14,14 M N M~ cc r-- C4 r-- cf) N r-- C,4 e r--4 M r-- 14 r--4 r--4 N r--4 M M M r-4 M

N IV t .0 r Cf L 00 L r-A r-A ONN f)Cf 0 N C ON C ON L I t 00 N ?t C ?t

C) " ON~ r--q I d I LO LO NON AI~tN N O NI~tN 00 Cf) ON \.o Id LO 00 N N I~t Id

o6 t? c t,? t? t-? ? o? ci \.6 o6 cf cfi \.6 44 6 t< 5 4 ; 4o6 fi

r--4 N N N N N C4 r(r( ( r( VNA N N ( ( A N r( ( -A r--4 N r--4 r--

CD0 \cN O etO f 10ON LrO 1. N ) N N C LOLO C C 4?CD

I~t "oLO 1. Cf) --q ON Cf \00 -A 10 N\.c cc) LO NC f) ONIt eN tN . NL


N a)) C) N C) O N I ON c) Lo ONb \ ONf NO qN \O d NLrN CC . ION1 ON 0e\L f DN 4

N 1.0 m I~ t m C C C r fO N L~O N N N N 114 N f N N N LO N III N

0) Ut
c~ 0;0
-" -~ ) O f)
h h h h
~L~L C ~~C e~dr L e\ C C
h h
U) O'.0 00 Q)
00 c) ~-N
c; x4LU)
9 O .ox C N10m . SU
r ~ ~ \ ~\ c (o c 0r o~\ (c~c~t ~~c~o c ~ oceL




23 S
H -- -- -
rcl~~~~~~~~~~~~~~~ 06r3 O~\ OO~r c OO~O0 ~~ 0C~r ~L ~C~Od~O\ ~\ C
- - cc OL~C~~ ~ C C reC~L~L~\OCC 0 C~ ~ \ O00d ~ 0 r r r U
(6~ u;~c
ct Cj ~ ci \ a))
C"~~~~~~~~~~~ ~ 0
eNc ~dc~
~\ c c ~c ~i~d
c c~c ;
~r ~cc~cc~
edr ~r L~c~ ?dI ~e~~
C,5 vv 0 u vv

U tb
U cu
a) a)
a) a)
a).r a a

..j (6 bCCa)
U= a); rn u 0
a) a) a) -a) .
-, a) a) a) a) a) r(\ 0 d\ C e~0 ~~\ 0 OC OC~\ ~0
t0a -
a) h U

c~ cj o d~d d ddid; 0
-/3 bi 0e 0~ C ~ 0e o~ oe C~ oe _~C~C ~C edr eC~C eC~r c ~C(r
00 0; 0vvv v

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

vealed that students in the most effective ing. The time devoted to reading activities
schools (M = 28.14 minutes per day) and
indicates that reading was an "operational"
moderately effective schools (M = 27.04
priority in the schools as a whole as well as
in the classrooms of the teachers in the most
mpd) spent more time in independent read-
ing than students in the least effective
effective schools. In the words of one
schools (M = 18.63 mpd). teacher at Stevenson, "My advice to othe
In three of the most effective schools,
schools is let kids READ, READ, READ! WRIT
teachers mentioned providing time for WRITE,
dents to read authentic texts as a factor con-
Preferred interaction style: Although the
tributing to their school's success. "I give trends were provocative, with half t
my students lots of time to engage in read- teachers in the most effective schools p
ing/writing opportunities. Lots of oppor-ferring coaching compared to about a qua
tunities to read all kinds of texts," explained
ter of teachers in the moderately and le
a Hilltop teacher. Or, as a Woodlawn effective schools, the ANOVAs on pr
teacher put it, "You become a better readerferred interaction styles by school effe
by reading. My students read at least 20-30tiveness were not statistically significan
minutes a day. Also, partner reading-theycoaching, F(2, 67) = 2.32, p > .05; tellin
love it." "Everyone in the whole school is
F(2, 67) = 2.01, p > .05; or recitation,
taking books home at night for reading. It's
F(2, 67) = .17 (see Table 10).
one of our school improvement goals," Supplementary analyses of reading-spe-
pointed out another Hilltop teacher. Thesecific teaching strategies. In addition to the
findings complement earlier research doc-question of the use and effects of more ge-
umenting that time spent in independent
neric teaching practices, we were able to
reading in school does make a difference in
apply nonparametric analyses to two addi-
students' reading achievement (Anderson,
tional reading-specific teaching domains-
Wilson, & Fielding, 1988; Elley & Man-
word recognition and comprehension in-
gubhi, 1983; Taylor, Frye, & Maruyama,
struction. In the case of word recognition,
we were limited to grades 1 and 2 because
Reading as a priority: When one looks
of the paucity of word recognition instruc-
across time spent in a variety of categories
tion observed in grade 3.
that fall roughly under the general rubric of
Approach to word recognition: Chi-square
reading instruction-whole-class instruc-
tests revealed that in comparison to the mod-
tion, small-group instruction, independent
(seatwork) activities, independent reading, erately effective schools (X2 = 5.0, p < .05)
and writing in response to reading-the av- and least effective schools (X2 = 5.4, p < .05),
erages across levels of school effectiveness more grade 1 and 2 teachers in the most ef-
were: most effective-134 minutes; mod- fective schools were frequently observed
erately effective-113 minutes; and least ef- coaching in the use of word recognition
fective-113 minutes. Across the four most strategies as children were reading in order
effective schools, teachers were averaging to teach them word recognition. More
134 minutes a day on reading activities. teachers in the most effective (X2 = 5.5) and
Eighty-five minutes of this was either small- effective schools (X2 = 8.5) were fre-
quently observed practicing sight words
group or whole-class instruction, and almost
30 minutes of the total was independent than teachers in the moderately effective
schools. There were no differences in the
reading. These times, based on teachers' logs
from 2 weeks, do not include time spent number of teachers in grades 1 and 2 who
reading aloud to children, time spent provided
in explicit phonics instruction (e.g.,
composition (in contrast to writing in re- focusing on letter-sound correspondences
or word families on the board, on a work-
sponse to reading), and time spent in spell-


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

sheet, or with lettermoderately

cards) across
effective the three
schools primarily pro-
types of schools (seevidedTableexplicit13).
phonics instruction, with
In third grade, little a few adding
word the coaching component.
recognition in-
struction was observed In the leastanywhere. In pri-
effective schools, teachers one
most effective school, marily provided explicit phonics
third-grade instruc-
provided small-group tion,instruction
with about half adding topractice
gling readers in which they coached them
sight words.
to decode multisyllabic Whenwords
asked about reasons
as for their suc-were
cess, teachers in
reading. In another most effective school two of the most effective
in one moderately effective schools mentionedschool,
the importancestrug-
of teach-
gling third-grade readers received
ing students strategies, not skills, as small-
a factor
group instruction from resource
contributing to their success.teachers
As one Hilltop
who used a combination of
teacher put coaching
it, "I focus on strategieswhile
reading and workthan on word families
specific skills-metacognitive strate- to
teach word recognition. gies, demonstrations,
In one most how to do think-
tive school and three alouds. I am process oriented
moderately so kids be-
schools, third graders continued
come independent rather than reliant to ondo the
word study as a subject separate
teacher." This sentiment was echoed fromby a
reading. Woodside teacher, "You're not going to im-
Although not universal across teachers, prove as a teacher if you don't get into
there was a definite trend in the most effec- teaching strategies. I teach my students
tive schools for grade 1 and 2 teachers to strategies to become independent."
combine (a) explicit phonics instruction in Approach to comprehension instruction: Chi-
isolation with (b) coaching students to usesquare tests revealed that more teachers in
a range of strategies to figure out unknown the most effective schools were frequently
words when they encounter them in every-observed asking higher-level questions
about stories students had read than teach-
day reading. In contrast, the teachers in the

TABLE 13. Number and Percentage of Teachers in Grades 1 and 2 (by School Effectiveness Rating) Frequently
Observed Using Three Approaches to Word Recognition Instruction

in Word
Recognition Providing
Strategies Explicit
during Phonics Practicing
Reading Instruction Sight Words
Number of
School Rating Teachers N % N % N %

Least effective (1) 15 2 13 9 60 6 49

Moderately effective (2) 18 3 17 11 61 0
Most effective (3) 15 8 53 9 60 4 27
3 > 2:
X2 5.0 5.5
P <.05 <.05
3> 1:
X2 5.4
P <.05
1 > 2:

Z2 8.5
P <.05

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

erately effective
ers in moderately effective (2 schools).
of 29, To gain
x2 insight
= 9.1
p < .01) or least effective (0 that
into other factors of 22,
might X2
explain how=to 9.8
p < .01) schools (seenurture
Table teaching
That thatsaid,
promote w
must reiterate the low learning,
rate ofwe undertookmore
these an analysis
nitively challenging of activities
instructional practicesin the
that overa
was indepen-
sample. dent of student achievement. Instead, we
Word recognition work and reading used the ratings assigned to teachers on the
practice were much more the focus of read- effective instruction scale to classify teach-
ing instruction in grades 1-2 across all ers into three levels of accomplishment
schools in this study than was comprehen- (most, moderately, and least); these levels
sion. Explicit instruction in comprehension were used as predictor variables to explain
strategies was seldom witnessed across variation in the instructional practices
grades 1-3. Discussions that stretched chil- teachers used (see App. D).
dren's thinking were also infrequent across To investigate the relation among the in-
grades 1-3. dicators of teacher expertise and classroom
Instructional practice and teacher ac- practices, we subjected this large set of
complishment. The analyses of instructional teacher variables to a MANOVA. We used
practices within levels of school effective- three levels of the teacher accomplishmen
ness documented the fact that, on average, rating (most, moderately, and least accom
teachers in effective schools operated dif- plished) as the independent variable and
ferently than did teachers in other schools. eight scores from the set of generic teach-
These average differences, however, masked ing practices derived from our empirica
instructional variation among teachers data (time spent in small-group instruc
within schools. Not all of the best teachers tion, time spent in whole-class instruction,
worked in the most effective schools. In fact,
time spent in independent reading, student
engagement rating, home communication
only 52% of the teachers in grades 1-3 in the
most effective schools were perceived from
rating, preferred style of telling, preferred
the observations to be the most accom- style of recitation, and preferred style of
plished teachers (as compared to 36% coaching)
in the as the set of dependent mea-
least effective schools and 34% in the mod- sures. The MANOVA was statistically sig

TABLE 14. Number and Percentage of Teachers (by School Effectiveness Rating) Frequently Observ
Three Approaches to Comprehension Instruction

Asking Students
Asking Higher- Write in
Text-Based Level Response
Questions Questions to Reading
Number of
School Rating Teachers N % N % N %

Least effective (1) 22 10 45 0 6 27

Moderately effective (2) 29 10 34 2 7 7 24
Most effective (3) 19 7 37 7 37 9 47
3 > 2:
X2 9.1
p <.01
x2 9.8
p <.01


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

nificant, F(14, 108) = 10.77,

guardians. Recallp < this
that .001,
factor thus
was statis-
follow-up univariate ANOVAs
tically were
significant in both con-
school-level anal-
ducted (see Table 15).
yses; that is, most effective schools had
Communicating to higher compositeThe
parents: school ratings
ANOVA and teach-
on level of home communication was noters in the most effective schools had higher
statistically significant, F(2, 65) = ratings
2.40, on the same home communication
p > .05. That is, no differences were ob-The differences between the school-
served in the frequency with which teach-
level and the accomplishment analyses sug-
gest either that the most accomplished
ers of different levels of accomplishment
communicated with students' parents or were not necessarily the best com-

TABLE 15. ANOVAs on Teacher Factors with Teacher Rating

Teacher Factor Number of

and Ratinga F p Mean SD Teachers

Home communication (mean is based on 2.4 N.S.

five-level rating)
1 2.88 1.22 17
2 3.23 1.23 23
3 3.64 1.03 28
Time on-task (mean is based on 85.41 < .001
three-level rating)b
1 1.29 .47 17
2 2.21 .51 23
3 2.93 .26 29
Time in small-group instruction 3.08 .05
(mean is minutes/day)c
1 25.35 26.09 13
2 38.67 33.00 23
3 48.25 21.51 27
Time in whole-class instruction 8.66 < .01
(mean is minutes/day)d
1 47.94 18.38 13
2 28.98 17.12 23
3 24.69 15.78 27
Time in independent reading .87 N.S.
(mean is minutes/day)
1 23.97 11.77 13
2 22.51 10.98 23
3 26.92 12.81 27
Preferred interaction
Coachinge 5.92 < .01
1 .06 .24 17
2 .21 .41 24
3 .48 .51 29
Tellingf 16.60 < .001
1 .745 .440 17
2 .375 .491 24
3 .069 .258 29
Recitation 1.00 N.S.
1 .18 .39 17
2 .38 .49 24
3 .34 .48 29

al = least accom
b3 # 2 (1).
c3 # 1.
d3 = 2# 1.
e3 # 2 = 1.
f3 0 2 1.

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

municators or that teacher effects were ulum, instruction, and interaction tools
moderated by a school-level ethic for this
required to keep students on task.
type of activity. Grouping practices: The ANOVA on time
Student time on task: The ANOVA for spent
the in a whole-class format, F(2, 60) =
8.66, p < .01, indicated that students with
student time on task rating was statistically
significant, F(2, 67) = 85.41, p < .001.teachers rated as least accomplished spent
Tukey post hoc tests revealed that the most
more time in whole-class instruction (M =
47.94 mpd) than teachers rated as moder-
accomplished teachers had a higher rating
for maintaining student on-task behaviorately accomplished (M = 28.98 mpd) or
(M = 2.93 out of a possible 3) than the mod-
teachers rated as most accomplished (M =
erately accomplished teachers (M = 2.21),
24.69 mpd). A two-way (teacher accom-
who, in turn, had a higher mean score plishment
than x grade) ANOVA showed a sta-
the least accomplished teachers (M = 1.31).tistically significant effect for grade level,
To shed more light on this teacher factor,
F(2, 54) = 7.90, p < .01, with a strong ten-
in six of our sites we were able conduct a dency for whole-class time allocations to in-
special analysis to learn more about the ef-
crease with grade level, but no statistically
fects of the student time on task variable,significant grade x teacher accomplish-
using a procedure first used by Pressley ment
et interaction. Means and standard de-
al. (in press). During the last two observa-
viations by grade level and teacher rating
are shown in Table 16.
tions, observers were asked to interrupt
their normal observational protocol every 5 The ANOVA on time spent in small-
minutes, scan the room quickly, and record group instruction revealed an effect for level
the proportion of children in the class whoof teacher accomplishment, F(2, 60) = 3.08,
were perceived to be on-task, that is, pro-p = .05, with students in the classrooms of
ductively engaged in their assigned activity.
teachers rated as most accomplished spend-
Grade 1-3 teachers rated as most accom- ing more time in small-group instruction
plished were found to have an average(M of= 48.25 mpd) than students with teach-
96% of their students on-task when the 5- ers rated as moderately accomplished (M =
minute counts of students on task were 38.67 mpd), who, in turn, spent more time
taken. By contrast, the time on task rates than students with teachers rated as least
were 84% and 61%, respectively, foraccomplished
the (M = 25.35 mpd).
moderately accomplished and least accom-The ANOVA on time spent on indepen-
plished teachers. Because these numbersdent
are reading indicated no statistically sig-
nificant differences between teachers at
based on only 30 teachers and 60 observa-
tions, they should be interpreted cautiously.
different levels of accomplishment. Stu-
Even so, they underscore the importance of averaged from 23 to 27 minutes a day
student time on task as a curricular and in independent reading across all condi-
management concern for teachers as they tions of teacher accomplishment. This find-
implement reading programs in their class-
ing is at variance with the parallel analysis
rooms. The findings suggest that, unlike fora teachers in the most effective schools,
variable such as parent communication, where reliable school effectiveness differ-
wherein individual teacher practices appear
ences emerged for independent reading. As
to be moderated by school efforts, promot-
with the parent outreach finding, it suggests
ing student on-task behavior is a teaching
that teacher practices for independent read-
practice not easily influenced by school-ing may have been moderated by school ini-
level practice. Instead of putting a school-
tiatives, interactions, and/or philosophies.
level practice in place (e.g., calling homeInteraction styles: The ANOVA on pre-
monthly), an individual teacher must ferred
de- interaction style of coaching and
velop the disposition as well as the curric-
teacher accomplishment was statistically


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 16. Mean Time (and Standard Deviation) (in Minutes)

by Grade Level and Teacher Rating

Time in
Time in Time in Activity Writing in
Number of Whole Small (excluding Independent Response to To
Teachers Class Group nos. 5 and 6) Reading Reading in R
Grade/Teacher Rating (1) (2) (3) (4) (5) (6) (7)a
Grade 1:
Least accomplished 3 50.00 32.83 15.4 15.97 25.70 123.70
(3.06) (35.92) (5.70) (10.06) (11.29) (38.78)
Moderately accomplished 7 23.53 49.17 10.34 23.44 9.87 116.36
(15.71) (47.98) (4.23) (13.28) (7.61) (48.02)
Most accomplished 12 17.33 52.00 7.88 26.50 19.98 123.69
(14.64) (20.24) (11.27) (11.90) (11.80) (30.90)
Grade mean 22 23.76 48.49 9.69 24.09 17.55 123.57
(17.48) (32.21) (14.32) (12.13) (11.58) (39.91)
Grade 2:
Least accomplished 6 36.45 28.25 9.05 20.95 14.25 108.95
(18.09) (27.46) (10.67) (10.75) (10.09) (22.17)
Moderately accomplished 7 26.79 47.34 15.06 21.19 16.76 127.14
(13.12) (29.00) (16.55) (11.42) (7.58) (24.67)
Most accomplished 9 23.06 52.78 5.58 25.82 12.74 119.98
(11.05) (22.66) (5.26) (13.06) (9.83) (32.37)
Grade mean 22 27.90 44.35 9.71 23.02 14.25 119.23
(14.37) (26.89) (11.67) (11.64) (8.93) (27.26)
Grade 3:
Least accomplished 4 63.63 15.38 10.03 34.50 16.80 140.34
(14.25) (20.05) (8.84) (7.98) (8.52) (26.60)
Moderately accomplished 9 34.93 23.79 3.30 22.81 9.81 94.64
(20.53) (19.85) (4.86) (9.95) (5.08) (23.47)
Most accomplished 6 41.85 33.97 3.27 29.40 19.53 128.02
(11.80) (19.16) (5.69) (16.11) (16.50) (30.62)
Grade mean 19 43.16 25.23 4.71 27.35 14.35 114.80
(19.68) (19.80) (6.36) (12.24) (10.94) (32.08)
Pooled across grades:
Least accomplished 13 47.94 25.35 10.82 23.97 17.68 125.76
(18.38) (26.05) (8.91) (11.77) (10.20) (28.33)
Moderately accomplished 23 28.98 38.67 9.15 22.51 11.79 111.11
(17.12) (33.91) (15.30) (10.98) (7.07) (34.59)
Most accomplished 27 24.65 48.25 6.09 26.92 17.47 122.38
(15.78) (21.51) (8.48) (12.81) (17.35) (33.87)
Mean across teachers 31.05 40.03 8.20 24.70 15.44 119.42
(18.78) (28.40) (11.50) (11.94) (10.47) (33.19)

aSum of columns 2-6. Columns may not total exactly due to rounding.

the Tukey
significant, F(2, 67) = 5.92, p < .01. least accomplished teachers (75%
post hoc tests revealed that more ferredoftelling
the than moderately accom
most accomplished teachers had a teachers
preferred(38%), who, in turn exceede
interaction style of coaching (48%)mostthanaccomplished
the teachers (7%).
were also more moderately accomp
moderately (21%) or least accomplished
teachers (2%) (see Table 17 for teachers
a complete
with a preferred interaction st
telling than teachers rated as most a
presentation of the interaction preferences).
The ANOVA on preferred interaction
plished. (The ANOVA on preferred
style of telling (see Table 17) was style of recitation and teacher
cally significant, F(2, 67) = 16.60, p < .001. was not statistically signif
Tukey post hoc tests revealed thatThe data of
more on coaching and telling may

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 17. The Relation between Teacher Accomp

Teacher Rating

Least Moderately Most

Accomplished Accomplished Accomplished
(N = 17) (N = 24) (N = 29)

Interaction Style N % N % N %

Yes 1 6 5 21 14 48
No 16 94 19 79 15 52
Yes 13 75 9 38 2 7
No 4 25 15 62 27 93
Yes 3 18 9 38 10 34
No 14 82 15 62 19 66

be two to use different of

sides word recognition
the strategies
our observation to figure out unknown words while they
sibility that
were reading connected atext.teac
By contrast, of
codes within the 15 moderately accomplisheda giv
teachers in
tional block, grades 1 and 2, threewe nev
(20%) were observed
individual frequentlyteachers,
coaching children as they were
ternative reading.to Of thethe 11 teachers perceived
oth as
ferred least accomplished,of
styles none was frequently
predicted observedby using the teache
In contrast strategy to teach
to word recognition. Chi-
differences for teachers within levels of square tests confirmed that these differ-
school effectiveness, these statisticallyences
sig- were statistically significant.
nificant differences among teachers across Chi-square tests revealed no systematic
schools suggest that a teacher's preferredrelation between teacher accomplishment
style of interacting with students is a teach- explicit phonics instruction. In fact, a
ing dimension that is less influenced by the of teachers in both grades 1 and 2
practice of others at the teacher's schooland across the three levels of accomplish-
than other dimensions of teaching being ment were frequently observed providing
investigated in our study such as time explicit phonics instruction (Table 18).
spent by students in independent reading A chi-square test indicated that, com-
or degree of home communication. As pared with to teachers identified as moderately
high levels of student on-task behavior, accomplished,
a more teachers identified as
preferred style of coaching during readingmost accomplished were frequently ob-
instruction may be a teaching skill that served
re- engaging students in sight word
quires time and/or support from more ac-
complished teachers in order to develop.What emerges is a pattern in which the
most accomplished teachers demonstrate a
Approaches to word recognition instruction:
Of the 22 teachers in grades 1 and 2more
who balanced portfolio of approaches to
assist in word identification (i.e., more of
were rated high on the composite teacher
accomplishment rating, 10 (45%) werethem
fre- do a little of each practice) and are, by
quently observed coaching children onand
how large, the only group of teachers across


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 18. Number and Percentage of Teachers (by T

Using Three Word Recognition Instructional

on Word
Recognition Providing
Strategies Explicit Practicing
during Phonics Sight
Reading Instruction Words
Number of
Teacher Rating Teachers N % N % N %

Least accomplished (1) 11 0 0 5 45 1 9

Moderately accomplished (2) 15 3 20 11 73 1 7
Most accomplished (3) 22 10 45 13 59 8 36
x2 7.17
p <.05
3 > 2:
x2 4.27
p <.05

across the three levels of teacher accom-

plishment, chi-square tests revealed that
thethese practices were not alp randomly distrib-
ing tasks. uted. Compared to least accomplished
Comprehension instruction: Across all teachers (0%), more most accomplished
schools, comprehension instruction was teachers (31%) were frequently observed
minimal in grades 1-3. Primary modes of asking higher-level questions and having
working on comprehension included ask- students write in response to reading,
ing questions (many of which were literal) where the difference was 17% (vs. 48%).
about the story as children were reading, The most accomplished teachers also dif-
either in small groups or in a whole-class fered from the moderately accomplished
setting, and having children write in re- teachers on the higher-order scale. Chi-
sponse to stories they had read. This writ- square tests for text-based questions were
ing was most typically in the form of a not statistically significant (see Table 20).
journal entry or answers to written ques- Another way to examine these data is to
tions. Twenty-nine of 70 teachers were fre- suggest that the most accomplished teach-
quently observed asking text-based ques- ers were more balanced in their employ-
tions, and 27 were frequently observed ment of comprehension-fostering activi-
having children write in response to what ties; they tended to emphasize some of
they had read. Only 11 of 70 teachers were each of these three common strategies,
seen frequently asking higher-level ques- whereas the other two levels of teachers
tions about children's feelings or about their tended to omit at least one of the catego-
lives in relation to a story they had read. ries. In the bigger picture, however, what
Only five teachers were frequently ob- we found in comprehension instruction is
served providing instruction (not including disconcerting; only 16% of the teachers in
worksheet completion) about a comprehen- grades 1-3 in this study were frequently
sion skill or strategy (see Table 19). observed asking higher-level, aesthetic-
Nonetheless, when looking at teachers response questions.

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 19. Number of Grade 1-3 Teachers (Ou

Comprehension Technique

Frequently Occasionally
Technique Observed Observed

Doing a picture walk 5 9

Asking for a prediction 6 20
Asking text-based questions 29 22
Asking higher-level, aesthetic-response questions 11 16
Having children write in response to reading 27 26
Doing a story map 2 16
Having children retell a story 2 20
Working on a comprehension skill or strategy 5 22

Summary and Conclusion cially those most in need of extra help, to

One contribution of the current studytwo
have blocks of small-group instruction.
is that
it focused on classrooms as well as schools Although ability-grouped instruction
has been criticized because it has been
to get a richer picture of what was happen-
ing in high-poverty schools that excelfound
in to doom struggling readers to a l
promoting growth in reading among timestu- in low groups (Anderson, Hieb
dents. Because we have already discussed & Wilkinson, 1985; Barr & Dreeba
the findings of this complex study, in1991;
this Gamoran, 1992; Hiebert, 1983; Oake
section we revisit key findings with an em- and a persistently unambitious cur
phasis on connections among classroom ulum (Allington, 1983; Hiebert, 1983), i
and school factors. We also situate our work important to consider how teachers p
ceived and implemented grouping in th
within the long-standing effective schools
and effective teaching traditions, discuss schools. First, teachers in intervi
the limitations of our study, and provide a about the importance of instruction
few broad conclusions. level grouping (a term they preferred to a
ity grouping) to meet students' needs.
Small-Group Instruction ond, teachers used systematic assessmen
Our finding that time spent in small- prevent the groups from being rigid and
group instruction characterized the most flexible, shifting group membership re
accomplished teachers and the teachers larly. in Besides establishing another layer
our most effective schools hearkens back to cultural barriers among students, the ot
important findings in the process-product major detriment of grouping is the diff
research of the 1970s (e.g., Anderson et al., ential nature of instruction accorded to dif-

1979; Stallings & Kaskowitz, 1974). The ad- ferent ability groups (see Allington, 1983).
vantage of our most effective schools in pro- Although differential treatment of groups
viding small-group time is a prime example was not a focus in our observations and
of how classroom- and school-level variables teacher logs, our data indicate that this did
interact to produce a desirable outcome. The not occur in the four most effective schools.
greater time allotted for small-group instruc-To the contrary, students in the lower in-
tion did not just happen. It was made pos-structional-level groups spent as much time
sible by the collaborative model used in allon higher-order activities as did average
four of the most effective schools. In that achievers. Also, the teachers thought that it
collaborative model, the classroom teacher, was necessary to make sure that most stu-
a resource teacher, an ESL teacher, and/ordents spent most of their time interacting
a special education teacher worked to-with books that were at their instructional
gether, often simultaneously in a single
level. It is important to note that this prac-
classroom, to enable every child, but espe-
tice is another example of the interaction of


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

TABLE 20. Number and Percentage of Teachers (by Teacher A

Using Three Approaches to Comprehension Inst

Level (Aes- Students
Text-Based thetic-Re- Write in
Oral sponse) Oral Response
Questions Questions to Reading

Number of
Teacher Rating Teachers N % N % N %
Least accomplished 17 4 24 0 0 3 18
Moderately accomplished 24 11 46 2 8 10 42
Most accomplished 29 14 48 9 31 14 48
3> 1:

X2 6.56 4.32
p <.05 <.05
3 > 2:

X2 4.11
p <.05

school and classroom variables. In this

most accomplished teachers. Perhaps coach-
case, two school-level factors, a common
ing for word recognition instruction during
assessment system that enabled flexible
children's reading of text is a place for teach-
movement between groups and a collabo-
ers to begin to develop the more general abil-
rative model that enabled flexible deploy-
ity to coach.
ment of teaching personnel, combined to
allow teachers to maximize small-group
instruction, a classroom-level factor. Phonics, Yes, but Phonics and ...
The importance of systematic phonics
Coaching instruction in learning to read has been doc-
Although different terms have been umented repeatedly (Adams, 1990; Bond &
used to describe what we have called Dykstra, 1967; Chall, 1967; Snow et al., 1998),
coaching (e.g., use of structuring com-
as has the need for phonics to be developed
ments, probing of incorrect responses, in conjunction
scaf- with real reading and writing
folded instruction), others have found (Adams, 1990). Wharton-McDonald et al.
type of "on the fly" instruction (1998) found
to be a that the most effective first-
characteristic of effective teachers (Ander-
grade teachers in their study taught decod-
son et al., 1979; Brophy, 1973; Rosenshine ing skills explicitly and provided their stu-
& Furst, 1973; Wharton-McDonald et al., dents with many opportunities to engage in
1998). Our most accomplished teachers ex-authentic reading. However, using either ap-
hibited a "general" preference for coach-proach in isolation may not be the optimal
ing over telling or recitation, whereas thepath toward helping students develop a
least accomplished teachers engaged more rich repertoire of word recognition strate-
commonly in telling. Although a coaching gies. Our data suggest that it is what teach-
preference did not emerge as a general dif-ers do to promote application of phonics
ference among teachers across school effec-knowledge during the reading of connected
tiveness ratings, we did find the practice oftext that matters most. A majority of teach-
coaching during reading to provide worders in grades 1 and 2 across all schools
recognition instruction to be characteristictaught phonics explicitly, in isolation. What
of both the most effective schools and the distinguished the most accomplished teach-

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

ers and the majority Other

of Promising
teachers Findingsin the mo
effective schools from their
The findings peers
related wasin-
to small-group the
use of coaching to struction,
help coaching
students learn
as pertains to the teach-ho
to apply word recognition strategies
ing of word recognition, to re
and higher-level
reading. Although more research
questioning were is the
found across both neede
to elucidate these coaching
fective schools and and application
accomplished teacher
strategies, our results suggest
analyses. that
Additionally, a number of conver
sations about systematic
were characteristicphonics instru
of either the teachers in
tion and opportunity to
the most practice
effective need
schools or most accom- to b
broadened to include
plished coaching during ev
teachers but not both.
eryday reading. Reaching out to parents. At the school
level, the most effective schools made more
Higher-Level Questions of an effort to reach out to parents than the
We found that more of the most accom- moderately and least effective schools. At
plished teachers and teachers in the mostthe classroom level, the teachers in the most
effective schools frequently encouraged effective schools made more of an effort to
communicate regularly with parents than
higher-level responses to text than less ac-
teachers in the other schools. These results
complished teachers or teachers in the mod-
erately and least effective schools. These
suggest that the teachers in the most effec-
findings support earlier research (Knapp,tive schools realized that good communi-
1995; Puma et al., 1997; Rosenshine & Furst,
cation and collaboration, found among the
1973) documenting the benefit of combining
staff, must also extend to the parents of the
higher-level and more basic skill instruc-
children in their schools.
tion. However, only 16% of the teachers in Independent reading. We found that the
the entire sample could be considered students
to in grades 1-3 in the most effective
truly emphasize comprehension. and moderately effective schools spent
more time in independent reading than the
Balanced Instruction students in the least effective schools.
We did not set out to examine the de- These results support findings from earlier
gree to which teachers engaged in what research that time spent in independent
has come to be called balanced reading reading
in- in school does make a difference
in students' reading achievement (Ander-
struction (Gambrell et al., 1999; McIntyre
& Pressley, 1996). Nonetheless, it is notson
dif-et al., 1988; Elley & Mangubhi, 1983;
ficult to describe our findings within aTaylor,
bal- Frye, & Maruyama, 1990). It is in-
anced instructional framework. On several
teresting to note that accomplished teachers
dimensions-variable grouping patterns, did not allocate any more time to indepen-
dent reading than their peers, suggesting,
support for reading (e.g., guided or inde-
as we proposed earlier, that this is an in-
pendent), approaches to word recognition
structional practice amenable to school-
instruction, modes of interacting with stu-
level influences.
dents, practices to support text compre-
hension-the most accomplished teachersMaintaining student on-task behavior.
and/or the teachers in the most effective As has been found in the research on
schools exhibited more balanced instruction effective teachers (Brophy, 1973; Wharton
McDonald et al., 1998), the most accom-
than their peers. Although we can only
speculate about the motives for greater bal-
plished teachers in this study managed, on
ance, the best teachers in our study said that average, to engage virtually all of their stu-
they would do whatever it took to meet thedents in the work of the classroom. By
contrast, the least accomplished teachers
wide array of individual student needs they
encountered every day in their classrooms.achieved decidedly lower rates of on-task


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

activity. This finding Knowledge,
mostChoice accom- and
other New
plished teachers did not America's Schools
extend to the models,teach-
ers in the most effective
so on. schools, who, as a
group, did not differ from the teachersand
Teacher Accomplishment in th
the other schools in this study in maintain-
Possibilities for Mentoring
ing on-task activity. This lack of consistency
Although the argument is adm
between the two teacher analyses (teachers
speculative, we want to suggest that
in schools varying in effectiveness versus
sistencies between our findings for
teachers of different levels
in the of
most effective accomplish-
schools and those for
ment) suggests that maintaining high levels
the most accomplished teachers may pro-
of time on-task is a teaching practice that is
vide encouraging news for those who re-
less amenable to school-level influence than
gard professional development as the core
other teaching practices, such as encourag-
of any reform movement. Even though
ing all students to engage in independentmany of the practices of the most accom-
plished teachers in this study, such as
Local versus national reform models.
coaching in word recognition during actual
Even though we included both national and
reading and asking higher-level, aesthetic-
local reform models in our sample, our response questions, were mirrored in our
data do not permit a definitive comparison analyses of teachers in the most effective
between the two. However, it is worth not- schools, this does not mean that all of the
ing that none of the schools in our most most accomplished teachers worked in the
effective category used national reform most effective schools. As reported earlier,
models; all were homegrown. Further- only 52% of the teachers in grades 1-3 in
more, we found no regional variations in the most effective schools were identified as
this pattern; the local reform models that the most accomplished teachers (as com-
proved most effective came from very dif- pared to 36% in the least effective schools
ferent regions of the country. We are not and 34% in the moderately effective
alone in finding support for local models of schools). It is plausible, however, that these
school reform. For example, in the Hope teachers were serving as models or coaches
for Urban Education Study (Charles A. who brought particular areas of expertise to
Dana Center, 1999), seven of the nine high- interactions with their colleagues. Our in-
performing schools had not adopted an terviews provided rich examples of this
off-the-shelf reform program. The common possibility. As the principal at Hilltop School
denominator seems to be that whatever the
explained, she had "worked to help people
model, national or local, it must privilege begin to appreciate the experts emerging
school and classroom practices that have within the building by bringing staff to the
proven effective in carefully designed and point where they acknowledged their exper-
implemented research efforts. We mention tise and by bringing teachers together to
the efficacy of local models in the face of share their expertise and learn together."
increasing pressure for schools to adopt Whatever the relationships among teach-
"research-proven" national models (Amer- ers (and more information is desperately
ican Federation of Teachers, 1997; Herman, needed about how these relationships play
1999; Herman & Stringfield, 1997; Slavin & themselves out and how to help skeptical
Fashola, 1998; Wang, Haertel, & Walberg, teachers accept the belief that even the
1998). Perhaps the "approved" lists ofpoorest children in their classes can learn),
research-proven programs should always the fact that not every teacher in the most
include "Home-Grown" (with a capital H effective schools is classified as a most ac-
and capital G) programs as a research-complished teacher should be heartening to
based option along side of Success for All, reformers who want to increase learning

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

of achievement
and achievement in the and reform did better than
our information
schools. What it suggests is that would have
a led us to be-
number of highly accomplished teachers,
lieve. What this suggests is that the static
which by our definition
assumptions means teachers
about school status that we
who possess more ofused the attributes
to select of the
schools were inappropriate.
All of these schools
canonical profile of pedagogically effective were, and are, on the
move, in one direction
teachers, may be sufficient to move or another.
a Curricu-
from the aspiring into the
lum leaders effective
and teaching cate-
staffs come and go,
and with them the energy to initiate or be
gory. Large-scale staff turnovers may not sus-
necessary. There are tain
exceptional teachers
schoolwide and classroom-level prac-in
all schools who could ticesserve
and reform. as models, peer
coaches, or demonstration Our measures were not perfect.
teachers to We help
committed teachers learn what effective in- would have liked to use more and better
struction looks like and, in the process, im-measures of a wider range of skills and
prove their teaching. strategies, including writing ability. Fur-
Limitations thermore, those who adopt a psychometric
lens would take issue with our reliance on
Although we are encouraged by our
classroom-based assessments administered
findings, this research, like all of the work in
the effective schools and effective teachingby multiple assessors in multiple sites.
tradition, comes with serious limitations. Most informal inventories, writing sam-
When all is said and done, we are examining ples, word lists, and even tests of phonemic
natural correlations between program and awareness do not possess the psychometric
teaching factors on the one hand and stu- properties of standardized multiple-choice
dent performance on the other. These cor- tests. Those who would adopt a lens of au-
relations may be useful in planning more de- thenticity would be equally disappointed
finitive research and in guiding the in our measures. Although they are class-
development of local programs and policies; room based, they are not the stuff of
however, they cannot be used to identify constructivist-based reform. Lacking is any
causes for improvements (or decrements) in appreciation of response to literature and
student achievement. For that, more system- personal engagement with text. Those who
atic experimentation is needed, including adopt a cultural lens would find both our
control groups, randomization, and careful student and our teacher measures wanting.
analyses of growth over time. It is to that They would find that our student measures
agenda that we will soon turn our attention. are not likely to be sensitive to the special
Our work carries a number of addi- skills or perspectives that children develop
tional, more specific limitations. We would in culturally rich settings. And they would
have liked to assess more students per class-that our observational lenses did not
room simply to improve the precisionguarantee and that observers would look di-
trustworthiness of our work. Because we rectly for culturally responsive (or cultur-
ally insensitive) instruction. In defense of
did not test the full range of students within
classrooms, we were unable to examine ap- what we did, all we can say is that, like all
titude x treatment interactions. school-based research efforts, we made
Also, our prior information about compromises motivated by cost and credi-
schools was unintentionally misleading. In bility. Because we were personally and
terms of selection, schools that we had ex- painfully aware of the problems of using
pected to rise to the top of our achievement group assessments, especially with young
scales (because of their reputations) did notreaders, and because we wanted assess-
always do so. Conversely, some schools
ments that would be credible with the
teachers in these schools, we were commit-
that were thought to be ordinary in terms


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

ted to one-on-one assessments, oral

ings of this study also suggest thatreadin
samples, and retellings-all
even the most effective practices
schools, have a long w
knew would appeal waytoto teachers.
go in improving reading instruction
Some of our data, insuch
the primaryas grades.
from the in-
terviews, logs, and Itquestionnaires, wer
is clear from this study that a combi-
based on self-report. of sound school-level decisions
the and inter-
view data on effective efforts as well as effective
schoolwide practic
from three teachers practices
and within
one individual
classrooms arepe
school corroborated what we found from needed if low-income schools are to beat the
odds in terms of primary-grade students'
the questionnaire data based on six to eight
teachers per school (depending on the num-reading achievement. From the descriptive
ber of grade levels involved in the study).
analyses and case studies, it is clear that the
Also, the log data on reading instruction process of becoming a school that promotes
minutes, events, and grouping patterns the learning of all its students is complex;
were based on a reporting of what actuallyequally complex are the conditions that char-
happened in the classroom every day for acterize
a an effective school. Staff in all of
these effective schools talked about the work
week as opposed to teachers' self-reporting
of how they typically spent time during they had been doing over a number of years
their reading/language arts block. Further-
to improve and the work that still lay ahead
more, the observations provided balanceoftothem. In other words, there is no single
quick answer to the question of how best to
the self-report data in the analysis of effec-
tive classroom practices. reshape a school's reading program and the
Another limitation was that we only repertoire of instructional practices teachers
conducted five 1-hour observations during employ in the quest to help all children read
the reading block in each teacher's class-
well by the time they leave the primary
room over a 5-month period. However, grades. What is equally clear is that educa-
these data were balanced with the two tors who combine a desire to improve their
weekly logs that the teachers completed to practice; a commitment to strong,
account for how they spent time on collaboratively
what forged schoolwide pro-
reading/language arts activities during grams;alland plain old-fashioned hard work
of their reading/language arts instruction.
can meet great expectations for the children
Beating the odds in reading thatwith wouldand for whom they work.
be predicted based on student poverty In all of the most effective schools and
dedication and hard work. A school needs most of the other schools in this study, the
good morale and teachers willing to work school environments were positive and the
tirelessly and collaboratively. Teachers and
schools were friendly places for children to
schools also have to go the extra milelearn.
to In most schools, the teachers and
reach out to parents. The results of this
principals were genuinely concerned about
study suggest that children in the primary
developing the abilities of struggling read-
grades make the greatest growth when ersaand, in many, were willing to take the
steps necessary to improve the reading
high proportion of their reading instruction
is delivered through small achievement- achievement of all their students. The ques-
tion that lies before us, and one to which we
based groups, when their progress is moni-
tored regularly, and when they have ample
are currently turning our attention, is how
time to read and to learn needed skills and
(and whether) the reading education pro-
strategies. Teachers who are most accom- fession can help develop the sorts of values
plished in helping children thrive in read-
and practices found in the most effective
ing are skilled in coaching and in keepingschools in other schools that only aspire to
all children academically on-task. The find-
these levels of accomplishment.

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

Appendix A

QRI Retelling Sheet


Retelling: Ask, "What happened in this story?"

Record student's response as best as possible:


1 2 3 4

Student offers Student relates Student relates some All major points
little or no details only. main ideas and some and appropriate
information about Student is unable supporting details. supporting details
the selection. to recall the gist Retelling is are included.
Retelling is of the selection. fairly coherent. Degree of
incomprehensible. Retelling is Sequence completeness and
Stated ideas do incomplete or ideas is logical. coherence.
not relate to are misconstrued. Student generalizes
the selection. Sequence is beyond the text.
not logical.

Appendix B Don't worry. I don't expect you to know all of

them. I cannot help you because I want to see
Instructions for Administering what you can do on your own. Do your very
Word Lists best. Look right above this card. That's where
you will see each word. Are you ready?
Examiner Instructions
(And after the child has missed seven con-
Directions for administering word lists are
secutive words on any list.) Thanks for your
the same for all grade levels. Begin administra-
reading. Now I'm going to let you look at the
tion with the preprimer list and have the child
rest of my words. Are there any that you know?
continue reading from one list to the next until
he or she makes seven (7) consecutive errors.
Appendix C
Count the number of correct responses for each
level and record the number at the bottom of
each sheet. Continue to administer a new list in Procedure for Creating the School
this fashion until the child misses seven or more Effectiveness Rating
words. Give the child time to attempt each word.
First, we created residual scores for all the rele-
Then uncover the remaining lists and put a check
vant spring reading measures, using appropriate
next to any additional word that the child reads.
Count those words and record the totals at the fall scores as covariates. In grade 1, spring resid-
bottom of the record sheets. ual scores for reading fluency (wcpm) and re-
telling were created by using fall phonemic seg-
Instructions for the Children mentation and blending scores as covariates. In
I have some lists of words that I want you to
grades 1-3, spring residual scores for reading
read one at a time. Some of the words will be words were created by using fall reading words
easy for you and some I expect to be veryscores hard. as a covariate. In grades 2-3, spring resid-


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

ual scores for reading fluency and retelling

were created by using fall reading fluency
as a covariate. Within each grade level, thes
sidual scores were converted to z-scores, wh
were calculated from the mean and standar
viation for the entire grade-level sample, so
the data could be aggregated across grade lev
Then, the residual z-scores for retelling (at
structional level), fluency (wcpm on a grad
level passage), and reading words were aggre
gated and standardized to create a compo
index of reading growth. 13. Effective use of praise
Second, we calculated what might be called 14. Extensive content coverage,
a primary (as in primary grades) outcome index, tional density
using the end-of-grade-3 scores on the district- 15. Explicit modeling and scaf
mandated test. As it turned out, these were, in 16. Teaching skills in context
each case, standardized achievement tests (six 17. Extra instruction for low re
schools used the Stanford Achievement Test 9; 18. Encouragement of self-reg
two used the Metropolitan Achievement Test 7; 19. Instructional balance
two used the California Achievement Test; two 20. Much reading of connected
used the Northwest Evaluation Association Lev- 21. Much writing of connected
els Test; and two used a district-normed test). A 22. Activities appropriate, me
residual mean percentile score for each school challenging
was calculated by controlling for the school's
poverty level (as indexed by the percentage of
students receiving free or reduced-price lunch).
This was done because students' achievement Note
scores are depressed in schools with 50%-75%
of students living in poverty and seriously de-
pressed in schools with 75%-100% of students The research reported in this article was con-
living in poverty (Puma et al., 1997). The residual ducted as part of the Center for the Improvement
scores were then converted to z-scores. of Early Reading Achievement and supported
Third, the z-scores on the project measures under the Educational Research and Develop-
and primary-grades outcome measure were ment Centers Program, PR/Award Number
summed and standardized. When we examined R305R70004, as administered by the Office of
these scores, we looked for natural breaks in the Educational Research and Improvement, U.S.
distribution that would divide the schools into Department of Education. However, the con-
three groups of approximately the same size. tents of the article do not necessarily represent
These breaks occurred at .5 standard deviations the positions or policies of the National Institute
above and below the mean. Breaking at those on Student Acheivement, Curriculum, and As-
two points yielded four most effective, six mod- sessment, the National Institute on Early Child-
erately effective, and four least effective schools. hood Development, or the U.S. Department of
Education, and one should not assume endorse-

Appendix D ment by the federal government.

Criteria Used to Rate Teacher
Accomplishment References
These criteria are based on the work of Anderson
et al. (1979), Barr and Dreeban (1991), Hoffman Adams, M. J. (199
(1991), Pressley, Rankin, and Yokoi (1996), Roeh- and learning abo
ler and Duffy (1991), and Wharton-McDonald et Press.
al. (1998). Allington, R. L. (1983). The reading instruction
Elements of Effective Instruction provided readers of differing abilities.
mentary School Journal, 83, 548-559.
1. Awareness of purpose American Federation of Teachers. (1997). R
2. Enthusiasm student achievement: A resource guide for rede-
3. Task orientation signing low-performing schools (AFT Item No.
4. High pupil engagement 3780). Washington, DC: Author.

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

Anderson, L., Evertson, haviors,

C., academic
& Brophy, learning timeJ. and (1979
An experimental study achievement: An overview. In C. Denham
of effective teaching&
in first-grade reading A. Lieberman
groups. (Eds.), Elementary
Time to learn (pp. 7-32).
Washington, DC: National Institute of Edu-
School Journal, 79, 193-223.
Anderson, R. C., Hiebert,cation. E. H., Scott, J. A., &
Wilkinson, I. A. (1985). L. B., Morrow, L. M., Newman,
a nation S. B., of
readers. Washington,& Pressley,
DC: M. National Institute
(1999). Best practices in literacy
of Education. instruction. New York: Guilford.
Anderson, R. C., Wilson, P. T., & Fielding, L.
Gamoran, A. (1992). Is ability grouping equita-
(1988). Growth in reading and how children ble? Educational Leadership, 50, 11-17.
spend their time outside of school. Reading
Herman, R. (1999). An educator's guide to school-
Research Quarterly, 23, 285-303. wide reform. Washington, DC: American In-
Barr, R., & Dreeban, R. (1991). Grouping stu-stitutes for Research.
dents for reading instruction. In R. Barr,Herman, M. R., & Stringfield, S. (1997). Ten promis-
Kamil, P. Mosenthal, & P. D. Pearson (Eds.), ing programs for educating all children: Evidence
Handbook of reading research (Vol. 2, pp. 885- of impact. Arlington, VA: Educational Re-
910). New York: Longman. search Service.
Bond, G. J., & Dykstra, R. (1967). The cooperative Hiebert, E. H. (1983). An examination of ability
research program in first-grade reading in- grouping for reading instruction. Reading Re-
struction. Reading Research Quarterly, 2, 5- search Quarterly, 8, 231-255.
142. Hiebert, E. H., Colt, J. M., Catto, S. L., & Gury,
Brophy, J. (1973). Stability of teacher effective-E. C. (1992). Reading and writing of first-
ness. American Educational Research Journal, grade students in a restructured Chapter 1
10, 245-252. program. American Educational Research Jour-
Campbell, J. R., Donahue, P. L., Reese, C. M., &nal, 29, 545-572.
Phillips, G. W. (1996). NAEP 1994 reading Hirsch,re- E. D. (1987). Cultural literacy: What every
port card for the nation and the states. Washing- American needs to know. New York: Vintage.
ton, DC: Office of Educational Research and Hoffman, J. V. (1991). Teacher and school effects
Improvement. in learning to read. In R. Barr, M. L. Kamil,
Carroll, J. B., Davies, P., & Richman, B. (1971). P. B. Mosenthal, & P. D. Pearson (Eds.),
The American heritage word frequency book. Handbook of reading research (Vol. 2, pp. 911-
Boston: Houghton Mifflin. 950). New York: Longman.
Chall, J. S. (1967). Learning to read: The great de- Invernizzi, M., Juel, C., & Rosemary, C. A.
bate. New York: McGraw-Hill. (1997). A community volunteer tutorial that
Charles A. Dana Center, University of Texas at works. Reading Teacher, 50(4), 304-311.
Austin. (1999). Hope for urban education: A
Knapp, M. S. (Ed.). (1995). Teaching for meaning
study of nine high-performing, high-poverty ur- in high-poverty classrooms. New York: Teach-
ban elementary schools. Washington, DC: U.S. ers College Press.
Department of Education, Planning and Eval- Leslie, L., & Caldwell, J. (1995). Qualitative read-
uation Service. ing inventory-II. New York: Harper Collins.
Madden, N. A., Slavin, R. E., Karweit, N. L., Do-
Colt, J. (1997). A scoring rubric for children's story
retelling. Longmont, CO: St. Vrain Valley lan, L. J., & Wasik, B. A. (1993). Success for
School District. All: Longitudinal effects of a restructuring
Cunningham, P., & Cunningham, J. (1992). Mak- program for inner-city elementary schools.
ing words: Enhancing the invented spelling- American Educational Research Journal, 30,
decoding connection. Reading Teacher, 46, 123-148.
106-107. McIntyre, E., & Pressley, M. (1996). Balanced in-
Deno, S. L. (1985). Curriculum-based measure- struction: Strategies and skills in whole lan-
ment: The emerging alternative. Exceptional guage. Boston: Christopher-Gordon.
Children, 52, 219-232. Oakes, J. (1985). Keeping track: How schools struc-
Donahue, P. L., Voelkl, K. E., Campbell, J. R., ture & inequality. New Haven, CT: Yale Univer-
Mazzeo, J. (1999). NAEP 1998 reading report sity Press.
card for the nation. Washington, DC: U.S. De- J. (1996) Emergent literacy survey. Bos-
partment of Education. ton: Houghton Mifflin.
Elley, W., & Mangubhai, F. (1983). The impact of G. S., Lyons, C. A., DeFord, D. E., Bryk,
reading on second-language learning. Read- A. S., & Seltzer, M. (1994). Comparing in-
ing Research Quarterly, 19, 53-67. structional models for the literacy education
Fisher, C., Berliner, D., Filby, N., Marliave, of R.,high-risk first graders. Reading Research
Cahen, L., & Dishaw, M. (1980). Teaching be- Quarterly, 29, 8-39.


This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms

Pressley, M., Rankin, J., & educating

egies for Yokoi, L. (1996).
disadvantaged children: Find- A
survey of instructional ingspractices of of
and policy implications primary
a longitudinal
study. Washington,
teachers nominated as effective in DC:promoting
U.S. Department of
literacy. Elementary School Journal, 96, 363-
Taylor, B. M., Frye, B. J., & Maruyama, G. M.
Pressley, M., Wharton-McDonald,
(1990). Time spentR., readingAllington,
and reading
R., Block, C. C., Morrow, L., Educational
growth. American Tracey, ResearchD.,
Baker, K., Brooks, G., nal,
27, 351-362.J., Nelson, E.,
& Woo, D. (in press).Taylor,
A study of P.effective
B. M., Pearson, D., Clark, K., & Wal-
first-grade literacy instruction. Scientific
pole, S. (in press). Accomplished teachers/effec-
Studies of Reading. tive schools: A look at exemplary kindergarten lit-
Puma, M. J., Karweit, N.,
instruction C.,
(CIERARicciuti, A.,
Online). Ann Arbor:
Thompson, W., & Vaden-Kieman,
University of Michigan,M.
for the Im-
Prospects: Final report on student
provement outcomes.
of Early Reading Achievement.
Washington, DC: U.S. Department
Taylor, of K.,
B. M., Pearson, P. D., Clark, Edu-& Wal-
cation, Planning and Evaluating
pole, S. (1999). BeatingService.
the odds in teaching all
Roehler, L., & Duffy, G. children
(1991). Teachers'
to read (CIERA Report instruc-
No. 2-006).
tional actions in learning to read.
Ann Arbor: UniversityIn R. Barr,
of Michigan, Center
M. L. Kamil, P. Mosenthal, & P. D. Pearson
for the Improvement of Early Reading
(Eds.), Handbook of reading research (Vol. 2, pp.
861-884). New York: Longman.
Taylor, B. M., Short, R. A., Frye, B. J., & Shearer,
Rosenshine, B., & Furst, N. (1973). The use of
B. (1992). Classroom teachers prevent read-
direct observation to study teaching. In
ing failure among low-achieving first-grade
R. M. W. Travers (Ed.), Second handbook of re-
students. Reading Teacher, 45, 592-597.
search on teaching (pp. 122-183). Chicago:
Rand McNally. Wang, M. C., Haertel, G. D., & Walberg, H. J.
Shavelson, R. J., & Berliner, D. C. (1988). Evasion (1998). What do we know: Widely implemented
of the education research infrastructure: A re- school improvement programs. Philadelphia:
ply to Finn. Educational Researcher, 17, 9-11. Center for Research in Human Development
and Education.
Slavin, R. E., & Fashola, O. S. (1998). Show me the
evidence! Proven and promising programs for Weber, G. (1971). Inner-city children can be taught
America's schools. Thousand Oaks, CA: Cor- to read: Four successful schools (CGE Occa-
win. sional Paper No. 18). Washington, DC:
Council for Basic Education. ERIC Docu-
Snow, C. E., Bums, S., & Griffin, P. (Eds.). (1998).
Preventing reading difficulties in young children: ment Reproduction Service No. ED 057 125
Report of the Committee on the Prevention ofWharton-McDonald, R., Pressley, M., & Hamps-
Reading Difficulties in Young Children. Wash- ton, J. M. (1998). Literacy instruction in nine
ington, DC: National Academy Press. first-grade classrooms: Teacher characteristics
Stallings, J., & Kaskowitz, D. (1974). Follow and student achievement. Elementary School
Through classroom observation evaluation 1972- Journal, 99, 101-128.
1973 (SRI Project URU-7370). Stanford, CA: Wilder, G. (1977). Five exemplary reading pro-
Stanford Research Institute. grams. In J. T. Guthrie (Ed.), Cognition, curric-
Stringfield, S., Millsap, M. A., & Herman, R. ulum, and comprehension (pp. 57-68). Newark,
(1997). Urban and suburban/rural special strat- DE: International Reading Association.

This content downloaded from on Mon, 11 Dec 2017 14:39:59 UTC
All use subject to http://about.jstor.org/terms