Making the Most of Interim Assessment Data

RESEARCH FOR ACTION

Making the Most of
Interim Assessment
Data
Lessons from Philadelphia

June 2009

RESEARCH FOR ACTION
Research for Action (RFA) is a Philadelphia-based, non-profit organization engaged in
policy and evaluation research on urban education. Founded in 1992, RFA seeks to
improve the education opportunities and outcomes of urban youth by strengthening
public schools and enriching the civic and community dialogue about public education.
For more information about RFA please go to our website, www.researchforaction.org.

Learning from Philadelphia’s School Reform

Research for Action (RFA) is leading Learning from Philadelphia’s School Reform, a
comprehensive, multi-year study of Philadelphia’s school reform effort under state
takeover. The project is supported with lead funding from the William Penn
Foundation and related grants from Carnegie Corporation of New York, the Samuel
S. Fels Fund, the Edward Hazen Foundation, the Charles Stewart Mott Foundation,
The Pew Charitable Trusts, The Philadelphia Foundation, the Spencer Foundation,
Surdna Foundation, and others.

Acknowledgements

We are deeply appreciative of the numerous overlapping communities of education
researchers, practitioners, and activists of which we are a part. These communities sustain
us and enrich our work. This research project, like so many others, has benefitted from
these relationships. Below, we thank those who made specific contributions to this report.

The Spencer Foundation and the William Penn Foundation provided generous financial
support for the research.

Researchers at the Consortium for Chicago School Research (CCSR) played an important
role in the quantitative analysis. John Easton contributed to the research design and analy-
sis and Steve Ponisciak, originally of CCSR and now at the Wisconsin Center for Education
Research, conducted the analysis. We thank them for their technical expertise and their
wisdom.

Many people were diligent readers and responders. The comments of two anonymous
reviewers raised important questions that helped us to sharpen and cohere the report’s con-
tent. Conversations and a joint project with colleagues at the Consortium for Policy
Research in Education were helpful. Our colleagues at Research for Action – Diane Brown,
Eva Gold, Tracey Hartmann, Rebecca Reumann, Elaine Simon and Betsey Useem – offered
sage advice.

Getting a report to press is an arduous task. Judy Adamson, Managing Director at RFA,
managed and directed the design, editing, and proofreading of the report. She was ably
assisted by Joseph Kay, Philly Fellow extraordinaire, Judith Lamirand of Parallel Design,
and Nancy Bouldin of Steege/Thomson Communications.

Most importantly, this report would not have been possible without the cooperation of the
School District of Philadelphia. Staff in the Office of Accountability and Assessment provid-
ed the data that were needed and answered many questions. Central office administrators
offered insights about the intentions of the district’s Core Curriculum and Benchmark
assessments. Staff of the district’s Education Management Organization partners helped us
gain access to schools. Special thanks to the principals, teacher leaders, and teachers in the
ten schools in our qualitative sample. All gave graciously of their time, were patient with
our many requests, and responded candidly to our questions. We are grateful to all of these
people for all that they do for Philadelphia young people everyday.

Making the Most of
Interim Assessment
Data
Lessons from Philadelphia

Jolley Bruce Christman
Research for Action

Ruth Curran Neild
Johns Hopkins University

Katrina Bulkley
Montclair State University

Suzanne Blanc
Research for Action

Roseann Liu
University of Pennsylvania

Cecily Mitchell
Research for Action

Eva Travers
Swarthmore College

The Consortium for Chicago School Research provided
technical assistance for the statistical analyses.

RESEARCH FOR ACTION
Copyright © 2009 Research for Action

A report from Learning from Philadelphia’s School Reform

The School District of Philadelphia
The School District of Philadelphia is the eighth largest district in the nation. In 2006-07 it
enrolled 167,128 students. 62.4% of the students were African American, 16.9% were
Latino, 13.3% were Caucasian, 6.0% were Asian, 0.2% were Native American, and 1.2%
classified as Other.

In December 2001, the Commonwealth of Pennsylvania took over the School District of
Philadelphia, declaring the city’s schools to be in a state of academic and fiscal crisis, dis-
banding the school board and putting in place a School Reform Commission. In 2002, Paul
Vallas became the CEO of the School District of Philadelphia. During his time as CEO from
2002 to 2007, student achievement scores rose substantially. The percentage of fifth and

Figure A.1 School District of Philadelphia 2002-2008 PSSA Results

Percentage of Students Advanced or Proficient, Grades 3-8 Combined
Initially grades 5 & 8. Grade 3 added in 2006, grades 4, 6, 7 added in 2007.

2002 2003 2004 2005 2006 2007 2008
60%

52.6%

50% 48.0%
45.8%
44.1%
47.1%
40% 42.5%
36.3%
39.1%
38.5%

30% 26.7%
30.8%
22.6%

20%
21.5%
18.6%
10% Reading I———I

Math I———I

0%

eighth graders (the grades consistently tested) scoring “Proficient” or “Advanced” on the
Pennsylvania System of School Assessment (PSSA) tests went up 26 percentage points in
math. In reading, the percentage went up by 11 points in fifth grade and 25 points in eighth
grade. The percentage scoring in the lowest category (Below Basic) dropped in all tested
grades by 26 points in math and 12 points in reading.

Test scores continued their climb in the year following Vallas’s resignation when the district
was led by an interim CEO who continued the same reforms. Achievement gains occurred
despite serious under-funding by the state (Augenblick, Palaich and Associates, Inc., 2007)
and despite the city’s high and growing rate of poverty, the highest among the nation’s 10
largest cities (Tatian, Kingsley, and Hendey, 2007).

Table of Contents
Introduction 1
The Usefulness of Interim Assessments: Competing Claims 2
Overview of Report 3

Chapter 1 - Organizational Learning: A Framework for Examining the Use of Benchmark
Assessment Data 5

Conceptual Framework 7
Research Methodology 8
Research Questions 9

Chapter 2 - Philadelphia’s Managed Instruction System 15
The Philadelphia Context 15
The Core Curriculum 17
SchoolNet 20
Benchmark Assessments 21
In Summary 27

Chapter 3 - The Impact of Benchmarks on Student Achievement 31
The Organizational Learning Framework and Key Research Questions 31
Analytic Approach 33
Findings 38
In Summary 44

Chapter 4 - Making Sense of Benchmark Data 45
Three Kinds of Sense-Making: Strategic, Affective, and Reﬂective 46
Making Sense of Benchmarks: Four Examples 48
In Summary 53

Chapter 5 - Making the Most of Benchmark Data: The Case of Mahoney Elementary /School 57
School Leaders and Effective Feedback Systems 59
Grade Group Meetings and Benchmark Discussions 61
Organizational Learning and Instructional Coherence 63

Conclusion - Making the Most of Interim Assessment Data: Implications for Philadelphia
and Beyond 65
Investing in School Leaders 65
Designing Interim Assessments and Supports for their Use 67
Implications for Further Research 68

References 69
Appendices 72
Authors 82

Three Kinds of Assessments

Tiers of Assessment

Perie, Marion, Gong, and
Wurtzel (2007)1 have
categorized the three kinds
Summative
of assessments currently Scope and Duration of Cycle Increasing
in use — summative, forma-
tive, and interim — by their
intended purposes, audi-
Interim
ences, and the frequency of
their administration. (instructional, evaluative, predictive)

• Summative assessments are
given at the end of a
Formative Classroom
semester or year to measure
students’ performance (minute-by-minute, integrated into the lesson)
against district or state con-
Frequency of Administration Increasing
tent standards. These
standardized assessments Source: Perie et al. (2007)
are often part of an
accountability system and are not designed to provide teachers with timely information about
their current students’ learning.

• Formative assessments occur in the natural course of teaching and learning. They are built
into classroom instructional activities and provide teachers and students with ongoing, daily
information about what students are learning and how teachers might improve instruction
so that learning gaps and misunderstandings can be remedied. These assessments do not
provide information that can be aggregated.

• Interim assessments fall between formative and summative assessments and provide stan-
dardized data that can be aggregated. Interim assessments vary in their purpose. They may
predict student performance on an end-of-year summative, accountability assessment; they
may provide evaluative information about the impact of a curriculum or a program; or, they
may offer instructional information that helps diagnose student strengths and weaknesses.

Figure A.2

1
Perie, M., Marion, S., Gong, B., & Wurtzel, J. (2007, November). The role of interim assessments in
a comprehensive assessment system. Washington, DC: The Aspen Institute.

Introduction
In recent years, school reformers have embraced data-driven decision-making
as a central strategy for improving much of what is wrong with public educa-
tion. The appeal of making education decisions based on hard data – rather
than tradition, intuition, or guesswork – stems partly from the idea that data
can make the source of a problem clearer and more specific. This newfound
clarity can then be translated into sounder decisions about instruction, school
organization, and deployment of resources.

In urban districts, the press for data-driven decision-making has intensified
in the stringent accountability environment of No Child Left Behind, where
schools look for ways to increase their students’ performance on state assess-
ments. These districts increasingly are turning to the significant for-profit
industry that has sprung up to sell them curricula aligned with state stan-
dards, data management systems, and interim assessments.2 Interim assess-
ments are standardized assessments administered at regular intervals during
the school year in order for educators to gauge student achievement before
the annual state exams used to measure Adequate Yearly Progress (AYP).
Results of interim assessments can be aggregated and reported at a variety of
levels, usually classroom, grade, school and district. The tools for administer-
ing and scoring the assessments and storing, analyzing, and interpreting the
assessment data are being marketed by vendors as indispensable aids to
meeting NCLB requirements.3

In this report, Research for Action (RFA) examines the use and impact of
interim assessment data in elementary schools in the School District of
Philadelphia. Philadelphia was, an early adopter of these assessments, imple-
menting them district-wide in September 2003. The report presents findings
from one of the first large-scale empirical studies on the use of interim assess-
ments and their impact on student achievement.

Interim assessments are a central component of what the School District of
Philadelphia’s leaders dubbed a “Managed Instruction System” (MIS). The
MIS includes a Core Curriculum and what are called Benchmarks in
Philadelphia. Benchmark assessments were developed in collaboration with
Princeton Review, a for-profit company, and are aligned with the Core
Curriculum. In Philadelphia, classroom instruction in grades three through
eight occurs in six-week cycles: five weeks of instruction, followed by the
administration of Benchmark assessments. In one or two days between the
fifth and sixth weeks, teachers analyze Benchmark data and develop instruc-
tional responses to be implemented in the sixth week.

The Philadelphia Benchmarks are consistent with the definition of interim
assessments offered by Perie, Marion, Gong and Wurtzel (2007) in that the

2
Burch, P. (2005, December 15). The new education privatization: Educational contracting and
high stakes accountability. Teachers College Record.

3
Burch, P. (2005, December 15).
1

Benchmarks: “(1) assess students’ knowledge and skills relative to curricu-
lum goals within a limited time frame, and (2) are designed to inform teach-
Unraveling the

ers’ instructional decisions as well as decisions beyond the classroom levels.”4
benefits of interim
(See Figure A-2 for a description of the differences among three kinds of
assessments — summative, interim, and formative assessments.)
assessment data
to improvement in

The Usefulness of Interim Assessments: Competing Claims student learning is a

The introduction of interim assessments in urban districts across the country
necessarily complex

has not been without controversy, as district leaders, teachers, and the testing
task.
industry make conflicting claims for the efficacy of these assessments for guid-
ing instruction and improving student achievement. Many educators and
assessment experts, alarmed by the growing market in off-the-shelf commer-
cial products labeled as “formative” assessments, insist that the only true
formative assessments “must blend seamlessly into classroom instruction
itself.”5 There is good evidence that these instructionally embedded assess-
ments have a positive effect on student learning.6 In theory, at least, interim
assessments could be expected to have a similarly beneficial effect on teaching
and learning as instructionally embedded, “formative” classroom assessments.
To date, however, there is not the same kind of empirical base for the claim
that interim assessments have the power of classroom-based assessments.
And, for a number of reasons, it can not be assumed that they would have the
same positive impact. For example, because interim assessments do not occur
at the time of instruction, they may not provide the kind of immediate feed-
back that is useful to teachers and students. And because they are standard-
ized tests that almost always rely on a multiple choice format, they may not
offer adequate information about “how students understand.”7

The controversy over interim assessments is growing as district budgets
shrink and there remains little empirical evidence about the efficacy of the
assessments in improving student achievement. The Providence Public School
District abandoned its quarterly assessments after three years of implementa-
tion. Researchers who documented Providence’s experience noted, “District-
level administrators provided a variety of explanations for the decision, includ-
ing a lack of evidence of effectiveness and the summative character of the
assessments, but left open the possibility of reinstating the assessments at a

4
Perie, M. et al., 2007, p. 4.
5
Cech, S. J. (2008, September 17). Test industry split over ‘formative’ assessments. Education
Week, 28(4), 1, 15, p. 1.
6
Black, P. & William, D. (1998, October). Inside the black box: Raising standards through class-
room assessment. Phi Delta Kappan.
7
Perie, M. et al., 2007, p. 22.
2

later date.”8 In January 2009, the Los Angeles teachers union threatened to
boycott the “periodic assessments” mandated by the district – a series of exams
given three or four times a year at secondary schools – claiming that the tests
are costly and counterproductive. Such district tests at all grade levels “have
become central to a debate over the proliferation of testing, whether it inter-
rupts instruction and can narrow the depth and breadth of what’s taught.”9

Overview of Report

Our research shows that Philadelphia’s elementary school teachers – in con-
trast to those in some other districts, such as Los Angeles, – have embraced
the Benchmark assessments, finding them useful guides to their classroom
instruction. However, unraveling the benefits of the Benchmark data to
improvement in student learning is a necessarily complex task. In this
study, we use data from a district-wide teacher survey, student-level demo-
graphic and achievement data, and qualitative data obtained from field
observations and interviews to examine the associations among such factors
as instructional leadership, a positive professional climate among teachers,
teacher investment in the Core Curriculum and Benchmarks, and gains in
student achievement on standardized tests.

Our analysis indicates that teachers’ high degree of satisfaction with the infor-
mation that Benchmark data provide is not itself a statistically significant pre-
dictor of student achievement gains. However, used in tandem, the Core
Curriculum and Benchmarks have established clear expectations for what
teachers should teach and at what pace. And, importantly, students in schools
where teachers made more extensive use of the Core Curriculum made greater
achievement gains than in schools where teachers used it less extensively.

Benchmarks’ alignment with the Core Curriculum offers the opportunity for
practitioners to delve more deeply into the curriculum as they review
Benchmark results, thereby reinforcing and strengthening use of the curricu-
lum. Surprisingly, however, our qualitative research showed that Phila-
delphia’s school leaders and teachers are not capitalizing on Benchmark data
to generate deep discussions of and learning about the Core Curriculum. This
suggests that continued use of Benchmark assessments in Philadelphia is not
likely to contribute to improved student learning without greater attention to
developing strong principals and teacher leaders. These school leaders need to
know how to facilitate probing conversations that promote teachers’ learning

8
Clune, W. H. & White, P. A. (2008, October). Policy effectiveness of interim assessments in
Providence Public Schools. WCER Working Paper No. 2008 Wisconsin Center for Education
Research, School of Education, University of Wisconsin-Madison http://www.wcer.wisc.edu/. p. 5.
9
Blume, H. (2009, January 28). L.A. teachers' union calls for boycott of testing. Los Angeles
Times [On-line]. Retreived on February 11, 2009 from
http://www.latimes.com/news/education/la-me-lausd28-2009jan28,0,4533508.story.
3

about curriculum and pedagogy. In this report, we use an organizational learn-
ing framework to offer specific recommendations for what district leaders can
Philadelphia’s

do to help school staff make the most of Benchmark results.
elementary school

It is important to note that while our research reviews how Philadelphia
teachers have

deployed its assessment model and examines student achievement data to embraced the
assess its impact, this report should not be seen as a review of the technical
quality of Philadelphia’s Benchmark assessments, interim assessments in
Benchmark

general, or the Core Curriculum. A close examination of the technical merits
assessments, finding
of these elements of the managed instruction system was beyond the scope of them useful guides to
this project.10 their classroom
Chapter One outlines our conceptual framework for interim assessments and
organizational learning, identifies key research questions, and summarizes
instruction.

the research methodology of this study.

In Chapter Two, we describe Philadelphia’s Managed Instruction System,
highlighting district leaders’ expectations for how school staff would use its
components. We draw on data from the district-wide teacher survey to
describe teachers’ use of the Core Curriculum and satisfaction with the
Benchmark assessments.

In Chapter Three, we address the question of whether the Managed
Instruction System and supportive school conditions for data use were asso-
ciated with greater student learning gains.

Chapter Four describes how school staff make sense of Benchmark data and
consider their implications for instruction. What do school leaders and teach-
ers talk about and what plans do they make as a result of their interpreta-
tion of the data?

Chapter Five is a case study of the Mahoney Elementary School. This case
provides concrete images of what school leaders and instructional communi-
ties can do to enrich the use of Benchmark data.

In the Conclusion, we discuss implications of this research for what needs to
be done in order for school staff to make the most of interim assessment data.

10
In 2005, Phi Delta Kappa International issued its assessment of the Core Curriculum and the
Benchmark assessments in “A Curriculum Management Audit in Literacy and Mathematics of
the School District of Philadelphia.” The report has only recently become available. Its authors
found that while the Core Curriculum had provided consistence in what is taught, 87 percent of
its instructional strategies in mathematics are at the knowledge and comprehension levels.
When the auditors observed classroom instruction, they found that 84 percent of the instruction-
al strategies used were at the knowledge and comprehension levels. Their overall judgment was
that the School District of Philadelphia was not meeting its own expectations for a rigorous cur-
riculum. In reviewing the Benchmark assessments, they also judged that most of the items
composing the test were at the levels of knowledge and comprehension.
4

Chapter One
Organizational Learning: A framework for
examining the use of Benchmark assessment data
Teaching is a complex enterprise. In order to help each student learn, a
teacher must be aware of the needs and strengths of individual students and
the class as a whole. She must note how children are making sense of newly
introduced concepts and how they are developing increasingly advanced
skills. What have children mastered and what continues to pose difficulty for
them? What is helping them learn? What is getting in their way?

The logic behind how interim Benchmark assessment data can assist teach-
ers is straightforward: a teacher acquires data about what her students have
learned; she examines the data to see where her students are strong and
weak; she custom-tailors what and how she teaches so that individuals and
groups of students learn more; and as teachers across the school engage in
this process, the school as a whole improves.

While we recognize the importance of an individual teacher’s use of student
performance data to guide her instruction, this report views use of student
data through a different lens. Specifically, we explore how an organization-
al learning framework can inform our understanding of how to strengthen
the capacity of schools to capitalize on Benchmark and other kinds of data.

Our focus on organizational learning follows from the school change litera-
ture which indicates that in order for all students to make consistent aca-
demic progress, school staff must work together in concerted ways to
advance the quality of the educational program.11 School improvement is a
problem of organizational learning, that is, the ability of school leaders and
teachers to identify and problem-solve around constantly changing chal-
lenges. From the perspective of organizational learning, urban schools – like
other organizations – will be better equipped to meet existing and future
challenges “by creating new ways of working and developing the new capa-
bilities needed for that work.”12

11
Little, J. W. (1999). Teachers’ professional development in the context of high school reform:
Findings from a three-year study of restructuring high schools. Paper presented at the Annual
Meeting of the American Educational Research Association, Montreal, Quebec.; Wagner, T.
(1998). Change as Collaborative Inquiry: A ‘Constructivist’ Methodology for Reinventing
Schools. Phi Delta Kappan, 80(7), 378-383.; Knapp, M. S. (1997). Between Systemic Reforms and
the Mathematics and Science Classroom: The Dynamics of Innovation, Implementation, and
Professional Learning. Review of Educational Research, 67(2), 227-266.; Spillane, J. P. &
Thompson, C. L. (1997, June). Reconstructing Conceptions of Local Capacity: The Local
Education Agency’s Capacity for Ambitious Instructional Reform. Education Evaluation and
Policy Analysis, 19(2), 185-203.; Senge, P. (1990). The Fifth Discipline: The Art & Practice of the
Learning Organization. NY: Doubleday.
12
Resnick, L. B. & Hall, M. W. (1998). Learning Organizations for Sustainable Education
Reform. Journal of the American Academy of Arts and Sciences, 127(4), 89-118, p. 108. 5

Recent research has begun to address the multiple factors related to overall
organizational capacity that affect data use.13 School capacity incorpo-
An important

rates multiple aspects of schools and the literature suggests that school
feature of learning
capacity has four dimensions: organizations is
the existence of a
• human capital (the knowledge, dispositions, and skills of individual actors);
• social capital (social relationships characterized by trust and collective relational culture
responsibility for improved organizational outcomes); that is characterized
• material resources (the financial and technological assets of the organiza-
by collaboration,
tion);14 and
• structural capacity (an organization’s policies, procedures, and formal openness, and inquiry.
15
practices).

An important feature of learning organizations is the existence of a relation-
al culture that is characterized by collaboration, openness, and inquiry.16
Knowledge building is a collective process that involves the development of a
shared language and commonly held beliefs. Organizational knowledge “is
most easily generated when people work together in tightly knit groups.”17
Applying this theory, we examined how formal instructional communities
made sense of data from Benchmark assessments and generated actionable
knowledge for planning instructional improvements.

A second focus of the study, also drawn from organizational learning theory
is the use of student performance data within feedback systems composed of
“structures, people, and practices” that help practitioners transform data
into actionable knowledge.18 In our effort to understand how Benchmark
data contribute to organizational learning, we applied the concept of a four-
step “feedback system” to analyze the structures and processes educators use
to engage with data collectively and systematically during the course of a
13
Mason, S. A. & Watson, J. G. (2003). Understanding Schools’ Capacity to Use Data. Paper pre-
sented at the Annual Meeting of the American Educational Research Association, Chicago, IL;
Leithwood, K., Aitken, R., & Jantzi, D. (2001). Making Schools Smarter: A System for
Monitoring School and District Progress. Thousand Oaks, CA: Corwin Press.

14
Spillane, J. P. & Thompson, C. L., 1997.

Century, J. R. (2000). Capacity. In N. L. Webb, J. R. Century, N. Davila, D. Heck, & E.
15

Osthoff (Eds.), Evaluation of systemic change in mathematics and science education.
Unpublished manuscript, University of Wisconsin-Madison, Wisconsin Center for Education
Research.

16
Senge, P., 1990; Argyris, C. & Schon, D. A. (1978). Organizational learning: A theory of action
perspective. Reading, MA: Addison-Wesley.

17
Brown, J. S. & Duguid, P. (1998). Organizing knowledge. California Management Review,
40(3), 28-44, p. 28.
18
Halverson, R. R., Prichett, R. B., & Watson, J. G. (2007). Formative feedback systems and the
new instructional leadership (WCER Working Paper No. 2007-3). [On-line]. Retreived on July
16, 7 A.D., from http://www.wcer.wisc.edu/publications/workingPapers/index.php.
6

school year. The four steps in the feedback system are; 1) accessing and
organizing data, 2) sense-making to identify problems and solutions, 3) try-
ing solutions, and 4) assessing and modifying solutions.

Conceptual Framework

The conceptual framework that guided our research, illustrated in Figure
1.1, reflects the ideas discussed above. On the left, the figure depicts the
larger policy and management context that we hypothesize will influence use
of Benchmark data – the school district’s Managed Instruction System and
the larger accountability environment of No Child Left Behind (NCLB). The
middle box represents the four dimensions of school capacity discussed
above. In this study, we focus on the role of school leaders and instructional
communities in strengthening school capacity. An organizational framework
suggests that these actors will be critical for creating the organizational
practices necessary for coherent feedback systems that strengthen organiza-
tional learning and school improvement. The four-step feedback system
described above is embedded within overall school capacity and instructional
communities. It is important to note that multiple feedback systems will be
operating simultaneously in a school; that these feedback systems do not
operate in a lock-step manner and are most likely to be iterative; and, that,
in the ideal, knowledge generated from one feedback system will inform
other feedback systems. Finally, on the right, we anticipate that the outcome
of these processes will be reflected in gains in student achievement.

This model highlights the complexity of data-driven decision-making and the
use of Benchmark data to guide instruction. For example, it implies that if
any one of the links in the feedback system in instructional communities is
missing – that is, if teachers do not examine student data or do not know
how to interpret the data they receive, or if they do not make instructional
decisions that follow logically from a careful interpretation of the data, or if
these decisions are not actually implemented in the classroom, or if their
effectiveness is not assessed – the potential to increase student achievement
is weakened. Further, it implies that the relative skill with which each activ-
ity is carried out – for example, whether the instructional decisions that
arise from the data are excellent or merely adequate – can affect how much
students learn.

The model also highlights the human, social, and material conditions in the
school that increase the likelihood of teachers being able to make good use of
student data. For example, strong school leadership is hypothesized to have
a positive effect on teachers’ opportunities to access and interpret data and
make appropriate instructional adjustments. School leadership also will affect
7

Figure 1.1 Conceptual Framework

Context School Capacity Outcome

Hu

l
ita
No

m

ap
an

lC
Child

Ca

ia
pi
Left

c
Accessing Gains in

So
ta
l
Behind and organizing Student
data
Policy Achievement
Sense-
Feedback System making
Assessing and
modifying in Instructional to identify
solutions problems and
Community solutions

School
District Trying
solutions
es

Managed
St
rc

ru
ou

Instruction
c
tu
es

System ra
lR

l
ia

Ca
er

pa
at

ci
M

ty

the extent to which teachers are encouraged to use elements of a Managed
Instruction System, including the Core Curriculum. In addition, the material
conditions of the school, including access to computers and the Internet, may
affect the extent to which teachers are able to review student data.

Research Methodology

This study includes information from the period September 2004 through
June 2007. During the first year of the project, the research was exploratory
in nature and focused on learning about the district’s Managed Instruction
System as it unfolded, identifying schools that exemplified effective use of
data, and working with the district to develop and pilot a district-wide
teacher survey that included items related to data use. The report draws on
three kinds of data:

• a district-wide teacher survey administered in the spring of 2006 and 2007;
• student-level demographic and achievement data from standardized tests; and
• qualitative data obtained from intensive fieldwork in ten elementary schools
and interviews with district staff and others who worked with the schools, as
8 well as further in-depth case study analysis of five schools in 2006-2007.

Teacher Survey Data

The district’s Office of Accountability and Assessment constructed a single
teacher survey that combined questions about different topics. From the per-
spective of this study, important survey items included questions about
school leadership, climate, and collegiality, developed and documented by the
Consortium on Chicago School Research. The survey also included several
original questions specific to Benchmarks, such as satisfaction with
Benchmarks, professional development on data use, access to technology
that could enable viewing student data online, and discussion of instruction-
al responses to data with fellow teachers and school leaders. While these
data-related survey questions provide important insights, a more complete
understanding of the use of, and professional development for, Benchmarks
and other types of student data would have required a considerably longer
set of items. However, we use what is available to us to identify associations
between data-related variables, school leadership and climate, and student
achievement. In addition, teachers were asked about the subject(s) they
taught and the grade span in which they were teaching. (NOTE: In Chapters
Two and Three, we provide more information about the district-wide teacher
survey, the sample for our study, and our analytic approach.)

Student Test Score Data

Our analysis relies on measurement of student academic growth obtained
from longitudinal data on student achievement made available by the School
District of Philadelphia. Student test score data from spring 2005, 2006, and

Research Questions
1 What were district leaders’ expectations for how school staff would use
Benchmark data and what supports did they provide to help practitioners
become proficient in using data to guide instruction?
2 Were teachers responsive to the Managed Instruction System, particularly the
Benchmark assessments? Did they use them? Did they find them helpful?
3 Did students experience greater learning gains at schools where the condi-
tions were supportive of data use: that is, where the Managed Instruction
System was more widely accepted and used and where analysis of student
data was more extensive?
4What can school leaders do to ensure that the use of Benchmark data con-
tributes to organizational learning and ongoing instructional improvement
within and across instructional communities?

9

2007 were analyzed for students who were in grades 4 through 8 during 2005-
2006 and/or 2006-2007. The tests were either the Terra Nova or assessments
from the PSSAs, depending on the grade and year. Raw scores for each stu-
dent were converted to their percentile score within the district during the
year and these scores then were converted to z-scores with a mean of zero and
a standard deviation of one. To create a measure of growth, we examine
changes in students’ performance on standardized tests given at the end of
successive school years. This strategy examines the “value added” to learning
by attending a school in a given year. In this report, we examine improve-
ment in student academic growth in two school years (2005-2006 and 2006-
2007) for students in 4th through 8th grades.

Qualitative Data

The goal of our school-based qualitative research and in-depth case study
research was to develop a fine-grained analysis of the dynamic interactions
among school leadership, data use by instructional communities (grade
groups), and instructional planning. Our aim was to identify the micro-prac-
tices of school leaders and instructional communities as they worked with
data and put into action the resulting instructional decisions. Micro-practices
refer to the routine actions that are part of the larger function of data-driven
decision-making. Examples of micro-practices include: how data are format-
ted for analysis; how leaders facilitate discussions of data among staff; and,
how they communicate messages about the importance of data.

The school sample was composed of ten elementary schools that were among
the 86 schools identified as “low performing” and eligible for intervention
under a state takeover of the School District of Philadelphia The 86 low-per-
forming schools represented 39 percent of the district’s 220 elementary and
middle schools. Like the other 76 low-performing schools, each of the ten
schools in our sample was assigned to an intervention model beginning in
the 2002-2003 school year. Seven of the schools were under management by
outside providers; two schools were part of the district’s homegrown inter-
vention under the Office of Restructured Schools; one school was a “sweet
sixteen” school – a low-performing school that was showing improvement
and therefore received additional resources for two years but was not under
a special management arrangement. We chose to take an in-depth look at
the use of Benchmark data in low-performing schools because these schools
were under considerable pressure to improve test scores and they had more
resources, including, in most cases, additional personnel to provide support

10

for data use. We believed that these two factors would increase the likeli-
hood that they would turn to the Benchmark data for guidance.19

In identifying schools to be part of the qualitative study, we sought out
schools from each intervention model that would provide insight about how
schools learn to engage with data as part of a process of school change. We
developed a purposive sample of schools that were identified by district staff,
provider staff, and other school observers as being well on the road to mak-
ing effective use of data. Criteria for selection included: data-driven decision-
making was a stated priority of school leaders; professional development on
how to access, organize and interpret Benchmark data was ongoing; and,
grade group meetings to discuss Benchmark data occurred regularly.

All of our schools served a considerably higher percentage of students living
in poverty than the district average and served student populations that
were predominantly either African American or Latino. (See Appendix A for
more information about the ten schools.) It should be noted that, during the
course of our study, the majority of these 10 schools were undergoing organi-
zational restructuring. CEO Vallas believed that K-8 schools were more hos-
pitable environments for middle grades students and either closed or con-
verted most of Philadelphia’s middle schools into K-8 schools and added
grades 6-8 to many elementary schools.

In 2005-2006, a team of at least two researchers made two one-day site visits
to each of the ten schools. During the visit, we conducted semi-structured
interviews with the principal and two or three teacher leaders. Interviews
lasted 60-90 minutes. (See Appendix C for lists of topics covered in the inter-
views.) Site visits were scheduled on days when we also could observe a lead-
ership team meeting, grade group meeting(s), or other data related event(s).

In 2006-07, we narrowed our sample to five schools for more intensive field-
work. To select the five schools, we developed criteria related to four cate-
gories; the principal’s role in data use, the strength of the professional com-
munity, the school’s AYP status, and the purposes that school staff brought
to their interpretation of Benchmark data. The research team placed schools
along continua for each category and selected schools that represented the
range of variation. Two researchers spent about four days in each school.
During these visits, we followed up with principals, observed several events
at which staff discussed data, talked extensively with teacher leaders, and
also interviewed at least two classroom teachers in each school. By June

19
In addition, an original intention of the study was to use the different management models as
points of comparison. However, this research purpose fell away when all of the provider organi-
zations, except Edison Schools, Inc. adopted the district’s Managed Instruction System.
11

Table 1.1 School-Based Interviews and Observations

Type of Respondent 2004-05 2005-06 2006-07 Total

Principal 6 17 9 32

Subject Area Teacher Leader 13 24 13 50

Teacher 5 23 28 56

Other School Leader (e.g., Ass’t Principal) 1 3 12 16

Total # of Interviews Conducted 25 67 62

Type of Observation 2004-05 2005-06 2006-07 Total

Grade Group Meeting 2 8 4 14

Leadership Team Meeting 0 5 5 10

Professional Development Session 10 3 6 19

Other Event (e.g., CSAP meeting) 0 8 3 11

Total # of Observations Conducted 12 24 18

2007, our qualitative data set included more than 150 interviews with school
staff and faculty; 54 observations of leadership team meetings, grade group
meetings, and school-wide professional development sessions; and a collec-
tion of school documents. (See Table 1.1)

RFA’s qualitative research also included six interviews with administrators
from the district’s offices of Accountability, Assessment, and Intervention;
Curriculum; and Professional Development. The topics covered included the
Core Curriculum; student performance assessments generally, as well as in-
depth probing about Benchmark assessments; professional development for
school leaders on using data; and perceptions of whether and how the differ-
ent providers operating in the district were using the district’s Core
Curriculum and Benchmark system. Researchers also interviewed staff from
the education provider organizations to understand the policies and supports
related to data use offered by these organizations to the schools that they
were managing. (See Table 1.2)

To analyze the interviews, we coded the data using a software package for
qualitative data analysis and identified themes and practices within and
across schools and providers using content analysis. We used information
from written documents and field observations to triangulate our findings.
12

Table 1.2 Central Office and Provider Interviews
Interviewee 2004-05 2005-06 2006-07 Total

Central Office 2 4 0 6

Provider 9 2 0 11

Total 11 6 0

Other analytical strategies included: case study write-ups of data use in each
of the ten schools; reduction of data into word charts (for example, a chart
describing the types of data that were attended to by school staff, the set-
tings and actors involved, and the resulting instructional decisions); and
development of extended vignettes of feedback systems in schools. More spe-
cific details on research methods, data analysis, and sample instruments can
be found in Appendices B, C, D, and E.

In the next chapter, we take a closer look at the design of the Managed
Instruction System and district leaders’ expectations for use of the
Benchmark assessment data.

13

School District of Philadelphia Timeline: September 2001 - June 2007
• January 2002
NCLB signed into law • February 2004
SchoolStat piloted in one region
• April 2002 • April 2007
Diverse providers chosen by Vallas resigns
School Reform Commission as CEO

2001 2002 2003 2004 2005 2006 2007

• July 2002
Paul Vallas appointed CEO
• September 2002 • September 2003
Core Curriculum Core Curriculum (K-8) • October 2004 • October 2006
piloted in 21 ORS & Benchmark testing SchoolNet rollout • November 2005 Budget crisis revealed
schools (3-9) implemented begins SchoolStat
• December 2001 district-wide implemented district-wide
State Takeover

Figure 2.1

Core Curriculum
A uniform curriculum for grades K-8 in math and literacy was implemented system-wide in September 2003.
A uniform curriculum in science was implemented for grades 7 and 8 in September 2004 and implemented for
grades K-6 in September 2005.
A uniform curriculum in social studies was implemented for grade 8 in September 2004 and grades K-7 in
September 2005.

SchoolStat
A performance management system developed by the Fels Institute; includes
1) data on student performance, attendance, and school climate; and
2) monthly data review meetings intended to help school leaders actualize what they are learning from the data.
The SchoolStat contract was cancelled in summer 2007 in the wake of budget cuts.

Benchmarks
Interim assessments administered every six weeks to inform instruction
(administered less frequently in high schools); aligned with the Core Curriculum;
implemented in grades 3-9 in September 2003, and grades 10-11 in September 2004.

SchoolNet
Web-based instructional management system; includes student performance data, curricular materials,
professional development materials, and online communities; users include school staff, parents, and students;
about 50 schools were equipped each semester, with all schools equipped by March 2006.

14

Chapter Two
Philadelphia’s Managed Instruction System20
I tell my teachers, ‘The Core Curriculum is your Bible.’
Principal

Benchmarks replace religion around here.
Teacher Leader

In response to accountability pressures from No Child Left Behind, School
District of Philadelphia leaders instituted a Managed Instruction System
that represented a more prescriptive approach to curriculum, instruction,
and assessment than the district had taken in previous reform eras. For this
chapter, we address two sets of questions: First, what were district leaders’
expectations for how school staff would use Benchmark data, and what sup-
ports did they provide to help practitioners become proficient in using data
to guide instruction? Second, were teachers responsive to the Managed
Instruction System, particularly the Benchmark assessments? Did they use
them? Did they find them helpful?

Leaders expected that data from the Benchmark assessments would be used
by school practitioners in the context of a more broad-based focus on data-
driven decision-making and that the data would inform planning and action
at the classroom, grade, and school levels. In this chapter, we provide a
description of Philadelphia’s Managed Instruction System, district leaders’
expectations for the use of the MIS, and the supports that were provided to
help practitioners use its components. Drawing on data from the district-
wide teacher survey and data from our interviews in schools, we also report
teachers’ responses to the MIS.

The Philadelphia Context

District-wide curriculum and student assessment has been an integral part
of the School District of Philadelphia’s efforts to improve education and stu-
dent achievement for more than 25 years. Over this time, assessment results
have been used for both instructional and accountability purposes. The cen-
terpiece of Superintendent Constance Clayton’s 12-year administration
(1980-1992) was the K-12 Standardized Curriculum with a week-by-week
schedule for instruction. A criterion-referenced test for each subject area
administered annually measured students’ mastery of the Standardized
Curriculum.

20
This chapter is based on a presentation by Research for Action and the Consortium for Policy
Research in Education, Building with Benchmarks: The Role of the District in Philadelphia’s
Benchmark Assessment System, presented at the Annual Meeting of the American Educational
Research Association, New York, NY, March 2008.
15

David Hornbeck, who became superintendent in 1994, brought standards-
based reform to Philadelphia. The School District of Philadelphia abandoned
Vallas had become

the Standardized Curriculum of the Clayton era, shifting its emphasis from
convinced of the
teachers covering a prescribed curriculum to all students meeting rigorous
performance standards. In Philadelphia’s first move towards accountability
efficacy of a standard

based on student achievement, the district adopted the Stanford
district-wide

Achievement Test (SAT9), an off-the-shelf, nationally-normed test, as an curriculum during his
important part of the Performance Responsibility Index (PRI). Principals’
performance reviews and salaries were tied to their schools’ meeting district-
tenure as CEO of the

established PRI targets.21 The School District of Philadelphia issued curricu-
Chicago Public
lum frameworks that provided teachers an overall approach to curriculum
and instruction and sample lessons for different subjects and grade levels.
Schools.

However, the frameworks did not offer a scope and sequence, and many
teachers, as well as the Philadelphia Federation of Teachers (PFT),
expressed frustration with what they saw as a lack of curricular guidance.22

Since a state takeover of the Philadelphia school district in 2001, the district
has served as a laboratory for fundamental changes in school governance
and management. The most publicized of these changes was a complex pri-
vatization scheme that includes market solutions such as a “diverse
provider” model of school management,23 expansion of charter schools, and
until 2007, extensive outsourcing of additional core district functions, includ-
ing Benchmark assessments.24 However, at the same time, the district insti-
tuted strong centralizing measures for schools that were not part of the
diverse provider model.

When he came to Philadelphia in 2002, CEO Paul Vallas, with the support of
the PFT, began plans for a Managed Instruction System. As shown in Figure
2.1 one of Vallas’ first initiatives was to institute a district-wide Core
Curriculum in four academic subjects for grades K-8. Benchmark

21
Porter, A. C., Chester, M. D., & Schlesinger, M. D. (2002, June). Framework for an effective
assessment and accountability program: The Philadelphia example. Teachers College Record,
106(6), 1358-1400.

22
Corcoran, T. B. & Christman, J. B. (2002, November). The limits and contradictions of sys-
temic reform: The Philadelphia story. Philadelphia: Consortium for Policy Research in
Education.

23
In total, seven different organizations (three for-profit educational management organizations
(EMOs), two locally based non-profits, and two universities) were hired and given additional
funds to provide some level of management services in 46 of the district’s 264 schools (Bulkley
et al., 2004). The SRC also created a separate Office of Restructured Schools (ORS) as its own
internal “provider” to oversee 21 additional low-performing schools, granted additional funding
to 16 low-performing schools that were making progress (the “sweet sixteen,” and converted
three additional schools to charter schools (Useem, 2005).

24
For example, the School District of Philadelphia contracted with Kaplan to develop the Core
Curriculum for grades nine through twelve and hired outside vendors such as Princeton Review
to run extensive after-school programming for students who were struggling.
16

assessments accompanied the Core Curriculum. Vallas had become con-
vinced of the efficacy of a standard district-wide curriculum during his
tenure as CEO of the Chicago Public Schools. Philadelphia central office staff
who had served during the Hornbeck years also saw the value in this
approach. They, along with staff from the Philadelphia Education Fund,
developed the district’s Core Curriculum for grades K-8.

Vallas made the Core Curriculum and Benchmarks mandatory for district
schools that were not managed by private providers and voluntary for those
managed by private providers. However, all of the providers (with the excep-
tion of Edison Schools, Inc.) adopted parts or all of the district’s Core
Curriculum and the Benchmark assessments.25

District-Wide Teacher Survey Data Used for Analysis in this Chapter

In June 2006 and June 2007, the school district distributed a pencil-and-paper survey to all of its
approximately 10,500 teachers. A total of 6,680 teachers (65 percent of all teachers) from 204 of
280 schools responded to the spring 2006 survey. A total of 6,007 teachers (60 percent of all
teachers) responded to the spring 2007 survey. These response rates are comparable to that for
large-scale teacher surveys in other major cities; for example, teacher surveys fielded by the
Consortium on Chicago School Research typically produce a response rate of about 60 percent.

District leaders had particular expectations and theories about how teachers would use the Managed
Instruction System. But how did teachers respond to it? For this chapter, we examined survey
responses from elementary and middle grade teachers who said that: (a) they were teaching in a
grade span in which Benchmark assessments were offered and (b) they taught either in a self-con-
tained elementary classroom or were assigned to teach math, English, language arts, and/or reading
in grade three or above. There are 1,754 teachers in the data set for 2006 and 1,941 teachers in
2007 who meet these criteria. In this report, we use the most recent data unless a particular ques-
tion was not on the survey in 2007.

The Core Curriculum

In grades K-8, the Core Curriculum includes performance goals that specify
what students must know and be able to do by the end of the school year,
while indicating the intermediate levels of proficiency students should attain
to be on track to meet state standards. The curriculum includes a specific

25
Edison, Inc. was the only outside provider that came to Philadelphia with a fully-developed cur-
riculum. It also quickly developed its own interim assessments that were designed to predict stu-
dents’ performance on the PSSA. When CEO Vallas heard about Edison’s assessments, he decided
that they were a good idea. However, curriculum and assessment staff became convinced that
aligning them with the Core Curriculum was more important than having them serve a strictly
predictive function.
17

pacing schedule that is organized by six-week instructional cycles. It indi-
cates how many days should be spent on topics covered in the Core
It was a rare teacher

Curriculum and identifies the relevant textbook pages (specific textbook
who reported that he
series are mandated for literacy, mathematics, and science). The district or she did not
requires that all elementary students have 120 minutes of literacy and 90
minutes of math per day.26 The Core Curriculum provides teachers with
“always” or “often”

suggested “best practices” and multicultural connections that can be inte- use the Core
grated into daily lessons. Supplemental resources for enrichment are provid-
ed, as well as strategies for working with special student populations.
Curriculum to guide
instruction.
Despite these supports, the Core Curriculum poses considerable challenges
for Philadelphia teachers. The district’s research-based “balanced approach”
to literacy requires that teachers use guided reading groups and reading cen-
ters – instructional strategies that are new to many teachers and that test
teachers’ classroom management skills. Teachers are also required to use
Everyday Math (grades 1-5) and Math in Context (grades 6-8), research-
based curricula developed in the 1990s and promoted by the National
Science Foundation. Both math curricula emphasize problem solving and
conceptual learning, an approach that challenges elementary and middle
grades teachers who often do not have sufficient mathematical knowledge to
choose instructional strategies that will help students scaffold from misun-
derstanding to understanding. These curricula also “spiral,” returning over
and over again to concepts previously taught, each time developing the con-
cept more deeply. The spiraling approach creates conflicts for teachers
because, as a district administrator explained, teachers “feel uncomfortable
going on [to new material] before the kids have mastered certain things .”
Comments made by teachers echo this statement. For example, a third grade
teacher remarked about the Everyday Math curriculum,
I just don’t believe that the children can grasp concepts in two days
and then be introduced to them again three weeks later. You know, in
some skills, all skills, you need consistent practice, practice with it.
And I don’t believe that program gives it to them (2006).

Teachers’ Use and Perceptions of the Core Curriculum

Results from the teacher survey indicated that teachers’ responses to the
Core Curriculum were generally strong and positive. By the time the dis-
trict-wide teacher survey was conducted in June 2007, four years after the
district-wide rollout of the Core Curriculum, it was a rare teacher (9 percent)
who reported that he or she did not “always” or “often” use the Core
Curriculum to guide instruction (other response choices were “occasionally”

26
Travers, E. (2003, September). Philadelphia school reform: Historical roots and reflections on
the 2002-2003 school year under state takeover. Penn GSE Perspectives on Urban Education, 2(2).
18

and “never”). Eighty-six percent of the teachers said that they often or
always used the Core Curriculum to organize and develop course units and
classroom activities. Seven out of ten teachers reported that they often or
always used the Core Curriculum to “redesign assessment strategies.”

These findings are consistent with our qualitative research as many teachers
were positive overall about the Core Curriculum and its ability to engage
students. For example, a fifth grade teacher explained that the goal of her
school was to follow the Core Curriculum with “fidelity” because it helped
teachers stay on track and helped students achieve proficiency. She stated,
This year that just passed, [our goal was] to follow the Core
Curriculum because we began to believe that if we followed that
grade through grade that kids would be proficient. If I’m doing my
own thing, you’re doing your own thing, we’re not really following one
thing, the kids are not going to reach their fullest potential.
(May 2007)

Furthermore, some teachers reported making instructional changes in their
classroom based on specific strategies highlighted in the Core Curriculum.
They expressed confidence that using these strategies would result in
increased student achievement.

As shown in Figure 2.2, substantial majorities of teachers reported that their
school placed a strong emphasis on achieving the standards outlined in the
Core Curriculum, that the Core Curriculum was clear, that they believed
that they were engaging their students when implementing the Core
Curriculum, and that they had received adequate support to implement the
Core Curriculum. Given the teachers’ generally positive reports about the
clarity of the curriculum, its capacity to engage students, and the support

Figure 2.2 Teacher Survey Responses on Core Curriculum:
Percent reporting agreement
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

School has emphasized achieving proficiency standards in the Core Curriculum (n=1510) 90%.

The core curriculum is clear (n=1525) 82%.

Teacher believes he/she can engage students with Core Curriculum (n=1505) 76%.

Teacher reports adequate support to implement Core Curriculum (n=1515 ) 73%.

Most students will meet standards (n=1515) 44%.

*Number of respondents to each question appears in parentheses.

19

they had received for implementation, however, it is notable that fewer than
half of the teachers thought that most of their students would be able to
Each cycle of instruction

meet the academic proficiency standards outlined in the Core Curriculum.
and assessment consists
of six weeks: five weeks
SchoolNet of instruction, followed

SchoolNet is a district-wide instructional management system for the
by administration of
Benchmark assessments and other student data. It is intended to make
assessment data immediately accessible to every classroom teacher and build-
Benchmark assessments

ing principal and to provide analysis and instructional tools for educators’
and a sixth week of
use.27 Student information available on SchoolNet includes: PSSA and Terra review and/or extended
Nova results (by individual, class, grade, and school), Benchmark results, stu-
dent reading levels, student report card data, attendance data, and discipli-
development of topics.

nary data. (See Table 2.1 for a description of the major assessments used in
Philadelphia K-8 schools.) SchoolNet provides a number of other online fea-
tures to assist teachers with data analysis and re-teaching, including links to
the actual Benchmark items, information about how to re-teach the particular
standards, and additional practice worksheets for students. To facilitate
teachers’ use of SchoolNet, the School District of Philadelphia planned to
issue laptop computers to all teachers in district-managed schools (but not
schools managed by outside providers) thus reinforcing the expectation that
teachers’ classroom instruction would be “data-driven.”28

The district expected all teachers to receive training on the use of SchoolNet
and used a school-based, turnkey training approach. Generally, principals
and a technology support person received professional development from the
central office and were expected to return to their schools and train their
staff. As one administrator described, “The principals got trained in a day
during the summer. The teachers got trained on the first half day in October.
The principals got the PowerPoint and the principals trained the staff. We
wrote a script for them.” Our research indicated that, while training did
occur in the schools, there was considerable variation in whether principals’
expected teachers to use SchoolNet. Several principals echoed the sentiment
expressed by one, “I don’t necessarily think that going on the computer to
look at the data is a good use of teachers’ time. We print the data for them.”

27
Students’ families also have limited access to SchoolNet data through the system’s FamilyNet
tool to obtain up-to-date information on their children’s test scores (including Benchmark assess-
ments), report card grades, and attendance.

28
A fourth component of the Managed Instruction System was SchoolStat, a data management
system that compiled and compared school level data on student performance and behavior and
student and teacher attendance. Developed in partnership with the Fels Institute of Government
of the University of Pennsylvania, SchoolStat was used at regular meetings of regional superin-
tendents with their principals to discuss the status of, and ways to improve, climate and achieve-
ment in their schools. SchoolStat was discontinued in 2007, due to budget cutbacks.

20

Benchmark Assessments

Benchmark assessments were implemented district-wide in grades 3-8 in
Philadelphia in October 2004. In the preceding two years, they had been used
in the set of schools managed by the district’s Office of Restructured Schools
(ORS). Each cycle of instruction and assessment consists of six weeks: five
weeks of instruction, followed by administration of Benchmark assessments
and a sixth week of review and/or extended development of topics.29

At the time of the study, the district administered Benchmarks in Reading
and Mathematics to students in grades 3-8. Each Benchmark assessment
was designed to test only those concepts and objectives taught since the most
recent assessment was given. District leaders reported that the assessments
were also aligned to Pennsylvania’s assessment anchors (and, therefore, to
the content of the state test) and state standards. All of the items in the
Benchmark assessments are multiple choice and come directly from the con-
cepts and skills in the district’s pacing guide (called the “Planning and
Scheduling Timeline”). When the Benchmarks were first implemented, stu-
dents took paper and pencil tests. As schools came online with SchoolNet,
students took the assessments on computers.

On the district’s website, the Office of Curriculum identified multiple purpos-
es for the Benchmark assessments (School District of Philadelphia, 2007):

• To provide PSSA practice for students by simulating rigor, types
of questions and building test-taking stamina;
To provide teachers, administrators, students, and parents with a
quick snapshot of student progress;
•

To determine if what is taught is what is learned;
To help teachers reflect on instructional practices; and
•

To provide data to assist in instructional decision-making.
•
•

While the district’s website formally identified these purposes for the
Benchmarks, analysis of interviews with central office staff suggests two
central goals. First, the Benchmarks would provide feedback to teachers
about their students’ success in mastering concepts and skills covered in the
Core Curriculum during the five-week instructional period. One district
leader explained the limitations of past reliance on the state assessment
PSSA for formative information,

29
Journalistic accounts of the use of interim assessments (largely in Education Week) led us to the
conclusion that in most school districts using interim assessments, the tests are given between
three times a year and monthly. Aside from Philadelphia, we did not identify any other districts
where time was set aside explicitly for addressing weaknesses identified from analysis of interim
assessment data.
21

Table 2.1 District-Wide Assessments
Assessment Description

District Benchmark Assessments
Not required in schools managed by Administered at the end of the 5th week in a 6 week instructional
outside providers but used in all cycle to give teachers feedback about students’ mastery of topics
schools in the district except schools and skills in the Core Curriculum. Reading and mathematics in
managed by Edison Schools, Inc. grades 3-8; science in grade 3, 7 and 8.
Multiple choice questions.

Literacy Assessments
Informal reading assessments used in Administered at least two times a years for the purpose of
grades K-8. Developmental Reading establishing students’ instructional level in reading. In the early
Assessemnt (DRA) and the Dynamic grades these assessments are administered individually and assess
Indicators of Basic Early Literacy phonetic awareness, fluency, and re-telling. In grades 4-8 they
Skills (DIBELS) used in K-3. Gates- are administered in a group setting and assess word recognition
McGinitie used in grades 4-8 and comprehension.

Standardized Summative Assessments
Pennsylvania System of Standards-based test in literacy, math and science used to meas-
School Assessment (PSSA) ure achievement at district, school, grade, classroom and student
level. Multiple choice and open-ended response questions aligned
with Pennsylvania standards. Math and literacy in grades 3-8 and
11; science in grades 4, 8 and 11. The PSSA Writing Assessment
assesses students’ ability to write a five paragraph essay in
response to prompt. Scored for focus, content, organization, style
and conventions. Given in grades 5, 8, and 11. Not used for
accountability purposes.

Used in calculating whether a school makes Annual Yearly
Progress under NCLB.

22

We started with Benchmarks because that’s the only formative piece
we have. That became the one big thing that teachers had where they
could change directions if they needed to make mid-course correc-
tions. Before, you waited every year for return of the PSSA results.
(2005)

Second, the six-week cycle of teaching and assessment would, as one district
leader noted, “create some kind of a pacing and sequence program.”(2005)
Principals and teachers confirmed that the Benchmarks provided a curriculum
roadmap with specific destinations demarcated along the way. One principal
described the reaction of teachers at her school: “When teachers saw kids’
results on the Benchmarks, they really knew ‘I didn’t cover this. I should have
covered this.’” At another school, a fourth grade teacher remarked,

The other tests, like the tests that I give in the classroom are maybe
targeting one story or one particular skill, whereas [Benchmarks] give
you the big picture of what you have done in the last 6 weeks and
whether you achieved what you were supposed to teach them in the
last 6 weeks (2007).

Similarly, a sixth grade teacher described the Benchmarks as “checkpoints”
that help him to see exactly where he is with the Core Curriculum and how
well the students understand what he is teaching (2007).

Teachers’ Use and Perceptions of Benchmark Assessments

Results from the teacher survey indicated that teachers’ use of the
Benchmark assessments was widespread and frequent. In 2007, fewer than
three percent of teachers reported that they had never examined their stu-
dents’ Benchmark assessment scores during the year. Almost half of the
teachers (45 percent) said that they had examined these scores more than
five times during the year, and an additional 44 percent said they had exam-
ined them three to five times. This high use held across both elementary and
middle grades teachers.

The survey data indicated that a majority of teachers believe that the
Benchmark assessments were a source of useful information about students’
learning. In 2006, 86 percent of the teachers reported that Benchmark assess-
ments were useful for identifying particular curriculum topics where students
still needed to improve. Likewise, in 2006, 67 percent agreed with the state-
ment that “The Benchmark tests are a useful tool for identifying students’ mis-
understandings and errors in their reasoning.” Figure 2.3 presents teachers’
responses to questions about Benchmarks on the 2007 survey. Almost three
quarters of the teachers said that they agreed or strongly agreed that the
Benchmarks gave them a good indication of what the students were learning
in their classroom (2007 data). Smaller percentages of teachers expressed posi-
23

tive views of the instructional consequences and pacing of Benchmarks. Sixty-
one percent of the teachers felt that the Benchmark assessments had
“When teachers saw

improved instruction for students with skills gaps (one of their key stated pur-
kids’ results on the
poses), 58 percent thought that Benchmarks set an appropriate pace for teach-
ing the curriculum, and 57 percent said that Benchmark assessments provided
Benchmarks, they

information about their students’ learning that they would not otherwise have
really knew ‘I didn’t

known – a remarkable admission for teachers to make. cover this. I should

These findings are consistent with our qualitative research. In our inter-
have covered this.’”
views with teachers, the majority reported that the Benchmarks helped
them identify student weaknesses that they would have missed if they had
- A Principal

not had Benchmark data. For example, a third grade teacher commented,
I think it really helps me to see what I need to review and go over.
Okay, nobody got their fraction question right; let’s go back and
review fractions. It just helps me see that. (2006)

A sixth grade teacher described how she learned from the Benchmarks that
her students were having difficulty following directions and needed to be
shown the steps for how to complete a particular assignment.
I have to model for them how I’m thinking . . . because they weren’t
reading the directions and they weren’t working through all the steps.
(2007).

Figure 2.3 Teacher Reports on Benchmarks:
Percentage of respondents reporting agreement

0% 10% 20% 30% 40% 50% 60% 70% 80%

Give me a good indication of what students are learning in my classroom (n=1496) 73%.

Have improved instruction for students at my school with skills gaps (n=1481) 61%.

Give me information about my students that I didn’t already know (n=1490) 57%.

Set an appropriate pace for teaching the curriculum to my students (n=1490 ) 58%.

*Number of respondents to each question appears in parentheses.

24

District Supports for Use of the Benchmark Data

The district provided a set of supports to all schools in the district: access to District leaders expected
online data, resources, and reports through SchoolNet, structured tools for
analyzing and reflecting on Benchmark data, and professional development.
teachers to use the sixth

The district provided additional supports to low-performing schools. week of instruction not

District leaders expected individual teachers to access and use a variety of
just to re-teach in the

analyses of Benchmark data available on SchoolNet and to take advantage of
same old way but to
instructional features of SchoolNet such as information about how to re- find new instructional
teach particular skills and concepts. strategies that would

The district also developed several tools that support teachers’ use of the prove more successful.
Benchmark data: the Item Analysis Report, the Data Analysis Protocol, and
the Teacher Reflection Protocol. (See boxed text on page 26 for a description
of each of these tools.) The purpose of the Item Analysis Report is to give
teachers a user-friendly way to access and manage data from Benchmark
assessments. The Data Analysis Protocol, which teachers are required to
hand in to principals, reinforces the expectation that Benchmarks, as a form-
ative assessment, will be used for instructional purposes by helping teachers
to think through the steps of analysis and action as they review the Item
Analysis Report. District leaders expected the analysis of Benchmarks to cre-
ate an opportunity for teachers to reflect on their instruction. The district
leaders reasoned that, in analyzing the Benchmarks, teachers could begin to
examine their own content knowledge and instructional repertoire with an
eye on identifying what professional development and support would be ben-
eficial to them. They expected teachers to use the sixth week of instruction
not just to re-teach in the same old way but to find new instructional strate-
gies that would prove more successful. One district administrator described
what she hoped would be a teacher’s thought process as she reviewed the
Benchmark data for her class:
I think the Benchmarks give you information about your class, which
then will say to you, “Okay, I’ve taught inference, and the
Benchmarks are showing me over and over again the kids aren’t get-
ting inference. I need to do something about trying to find a resource
for inference.” (2005)

To encourage teachers’ reflective use of the Benchmarks, the district created
a single-page Teacher’s Reflection Protocol intended to be completed by indi-
vidual teachers following each administration of the assessment.
While the primary focus of central office staff members was on the use of
Benchmark results by individual teachers, they also anticipated that various
groups in the school – especially grade groups – would examine the data. The
focus on groups of teachers was consistent with an emphasis on Benchmarks

25

Tools to support teachers’ use of Benchmark data

Item Analysis Report
The Item Analysis Report is generated by SchoolNet and provides teachers with an
item-by-item analysis of the test at the individual student level. The Item Analysis Report
provides data spreadsheets for every teacher that includes, for every student, the correct
and wrong answers selected; how many and exactly which items each student answered
correctly; the average percentage correct for each class for each item by state standard
statement; and the state standard statement tested for each item. (A mock-up of the
report can be found in Appendix B.)

Data Analysis Protocol
The Data Analysis Protocol poses the following tasks and questions:
• Using the Item Analysis Report, identify the weakest skills/concepts for your class for
this Benchmark period.
• How will you group or regroup students based on the information in the necessary item
analysis and optional standards mastery reports? (Think about the strongest data and
how those concepts were taught.)
• What changes in teaching strategies (and resources) are indicated by your analysis of
Benchmark reports?
• How will you test for mastery?

The Teacher Reflection Protocol
The Teacher Reflection Protocol includes the following writing prompts:
• In order to effectively differentiate (remediate and enrich), I need to…
• Based on patterns in my classes’ results, I might need some professional development
or support in…

serving instructional purposes. This expectation that teachers would talk with
one another regularly was explained by a district leader who commented:
The expectation is that the 3rd grade teachers will sit at a table with
each other and say, “Here’s how my kids did on Item 1. How did your
kids do? Whoa! My kids didn’t do well. Your kids all nailed it. Tell me
how you taught that? Alright, I’ll go back and I’ll try that.” That’s sup-
posed to happen item by item. (2005)

However, it did not provide a set of tools to guide group discussions of
Benchmark data. And the district professional development for principals
focused on the technical aspects of accessing and organizing data, not on lead-
ing staff through conversations about the data. District leaders also expected
that principals would use the Benchmark data to assess the successes and
gaps in a school’s instructional program. For example, the district directed
principals to use Benchmark results as they developed their School
Improvement Plans, a yearly exercise in which school staff assesses areas of
weakness that should be a focus for improvement in the following year.
26 The survey results shed light on where teachers received the most help with

how to use Benchmark results. Many schools had school-based literacy teacher
leaders and, less frequently, math teacher leaders. The number and mix of
Professional

teacher leaders depended on availability of funding. The greatest sources of
development for
help in interpreting Benchmarks and other data and using them to make principals focused
instructional decisions, according to the teachers, were the school-based literacy
and math teacher leaders. One-third of the teachers reported that the literacy
on the technical

or math teacher leaders provided “a great deal of help,” and 76 percent said aspects of accessing
that they provided at least “some help” (possible responses were; no help, some
help, and a great deal of help). Approximately two-thirds of the teachers report-
and organizing

ed that principals were at least “some help.” Clearly, school-based leaders made
data, not on leading
use of data a priority for their work with teachers. However, 69 percent of the staff through
teachers reported that regional office or central office personnel were “no help,”
an indication that regional staff do not often reach classroom teachers.
conversations about
the data.

In Summary

Historically, although education reformers have had considerable success
convincing districts to undertake organizational reforms, substantial instruc-
tional change in the classroom has been more difficult to achieve. This histo-
ry would give good reason to suggest that teachers would look at the institu-
tion of a Core Curriculum and Benchmarks and other assessments with
skepticism. However, our data from a district-wide teacher survey and quali-
tative research in ten schools indicated a more positive response. The
Managed Instruction System is, in fact, exerting considerable influence on
classroom instruction. Almost all teachers in grades 3-8 reported that they
used the Core Curriculum and data from the Benchmark assessments and
most found them useful. Our visits to ten schools between September 2005
and June 2007 corroborated findings from the teacher survey: use of the MIS
– the Core Curriculum and Benchmarks – had permeated schools, as the
quotes at the beginning of this chapter indicate.

It is likely that the historical context of the School District of Philadelphia,
the district’s design of the MIS, and the supports that it implemented to help
teachers use the Core Curriculum and Benchmarks contributed to teachers’
acceptance of the MIS. Philadelphia teachers were ready for the Core
Curriculum and Benchmarks; they saw the value of strong curricular guid-
ance in an era of high-stakes accountability.

The design of Philadelphia’s Benchmark assessments had two notable
advantages: alignment with the Core Curriculum and the provision of anoth-
er week of instruction after teachers received their students’ Benchmark
results. Alignment with the Core Curriculum made Benchmark results very
relevant to teachers’ instructional planning. Eighty-six percent of the teach-
27

ers said that they often or always used the Core Curriculum to organize and
develop course units and classroom activities. Thus, alignment likely con-
Philadelphia’s

tributed to instructional coherence throughout the school, a key feature of
Benchmark assess-
schools shown to make student learning gains in Chicago and elsewhere.30
Instructional coherence requires a common instructional framework that
ments had two notable

“guides curriculum, teaching, assessment, and learning climate” and
advantages: alignment

includes expectations for student learning and teaching materials.31 The with the Core
sixth week for remediation and extension of topics offered the opportunity
for Benchmarks to serve instructional purposes by providing teachers with
Curriculum and the

formative information that could guide their follow-up with students. School
provision of another
leaders and teachers appreciated these strengths. week of instruction

Finally, the district’s infrastructure for supporting the MIS likely con-
after teachers

tributed to teachers’ acceptance of the Core Curriculum and Benchmarks. received their
Our research showed that this infrastructure was in place by the time of this
study. Most teachers reported that their school emphasized the proficiency
students’ Benchmark

standards in the Core Curriculum and that they received adequate support
results.
for using the Core Curriculum. Most reported that they received the
Benchmark data in a timely way and that they had participated in profes-
sional development on how to access data. Additionally, from teachers’ per-
spective at least, school leaders had begun to organize school infrastructure
to support teachers’ use of Benchmark data. Teachers reported that they had
opportunities to review data with colleagues, and had received help from
math and literacy teacher leaders in using data.

Our research also suggests limitations of Benchmark assessments. Districts
may look to interim assessments, such as Philadelphia’s Benchmarks, for
three distinct purposes – instructional, evaluative, and predictive.32 Although
Perie and her colleagues note that a single assessment can serve multiple
purposes, they also comment that “one of the truisms in educational meas-
urement is that when an assessment system is designed to fulfill too many
purposes – especially disparate purposes – it rarely fulfills any purpose
well.”33 Certainly, Philadelphia’s district leaders and school practitioners
looked to Benchmarks for many things.

30
Newmann, F. M., Smith, B., Allensworth, E., & Bryk, A. S. (2001, January). Improving Chicago's
schools: School instructional program coherence benefits and challenges. Chicago: Consortium on
Chicago School Research.; Newmann, F. M., Smith, B., Allensworth, E., & Bryk, A. S. (2001).
Instructional program coherence: What it is and why it should guide school improvement policy.
Educational Evaluation and Policy Analysis, 23, 297-321.
31
Newmann, F. M. et al., 2001.
32
Perie, M. et al., 2007.
33
Perie, M. et al., 2007, p. 6
28

They intended for Benchmarks to serve instructional purposes by providing
“results that enable educators to adapt instruction and curriculum to better
The predictive use of

meet student needs.”34 As noted, the six week instructional cycle supported
Benchmark results
this intention. District leaders expected teachers to test for mastery again at can distract school
the end of the re-teaching week. However, our qualitative research suggests
that such teacher-developed assessment often did not often occur at the end
leaders and teachers

of the sixth week. It should be noted that the lack of such retesting repre- from the instructional
sents a disjuncture in the steps of the feedback system described in Chapter
One. Assessing the results of re-teaching is an essential part of determining
and evaluative

whether interventions have been successful.
purposes that offer
the most potential for
Other conditions, related to the assessments themselves, are also necessary
in order for interim assessments to meet instructional purposes. The assessment items
strengthening

must not only show teachers (as well as students) what students don’t understand, but instructional capacity.
also give adequate indications of why the confusion exists, what the missteps are. The
lack of open-ended questions on the Benchmark assessment was a limitation in this
regard. Further, if the distracter items on a multiple-choice test are not designed well,
they do not offer good clues to students’ misunderstanding. Finally, if the items operate
at only the lower levels of cognition (e.g., knowledge and comprehension), and do not
tap into analytical thinking, they are not good tests of conceptual proficiency.

Evaluative purposes include information about the fidelity of implementation
of curriculum and instructional programs and “enforce some minimal quality
through standardization of curriculum and pacing guides.”35 This appears to
be the greatest strength of the Philadelphia’s Benchmarks as they are cur-
rently designed.

Philadelphia’s Benchmark assessments were not designed to be predictive of
a students’ performance on end-of-year tests. Yet, as we will show in
Chapter Four, school practitioners believed that Benchmark results would
predict students’ performance (and were encouraged to believe this by
regional and central office staff and provider staff who worked with them).
The predictive use of Benchmark results can distract school leaders and
teachers from the instructional and evaluative purposes that offer the most
potential for strengthening instructional capacity.

The Managed Instruction System assumed strong leadership capacity at the
school level. One district leader described the principal’s complex role with
regards to the professional climate that would need to be established:

34
Perie, M. et al., 2007, p. 4.
35
Perie, M. et al., 2007, p. 5
29

To give teachers the time to have the conversation to plan instruction School leaders needed
and to support the teachers in doing what they need to do as far as
giving them the resources, the professional development, the climate to ensure that the school
to feel safe to talk about what they know and what they still need to schedule accommodated
learn themselves.

School leaders needed to ensure that the school schedule accommodated
grade group meetings,

grade group meetings, that these meetings were worthwhile, and that the that these meetings
allotted time was used to analyze and discuss student Benchmark results
and to learn about new instructional techniques. It was also up to principals
were worthwhile, and

to help with identifying the professional development needs of their faculty,
that the allotted time
as a whole and as individual teachers, based on the results of the was used to analyze and
Benchmarks; for example, what else did teachers need to understand about
the Core Curriculum? They needed to create a professional climate that
discuss student

encouraged professional learning through inquiry, reflection, and informed Benchmark results and
action. In Chapter Four, we delve into whether these expectations of school
leaders were realistic.
to learn about new
instructional techniques.
In this chapter, we have established the broad acceptance of the Core
Curriculum and Benchmarks by teachers and the formation of the basic
infrastructure to support implementation. The next question becomes
whether the Managed Instruction System, and its use of Benchmarks, had a
positive impact on student achievement. We take up that question in the
following chapter.

30

Chapter Three
The Impact of Benchmarks on Student
Achievement
An ultimate goal of systematically tracking student progress is to increase
student learning. However, whether the use of Benchmark data has an actu-
al – rather than theoretical – impact on achievement is a question that itself
needs to be examined empirically. This chapter builds on analyses presented
in Chapter Two, which showed that the basic infrastructure for a Managed
Instruction System was firmly in place and accepted by teachers. The wide-
spread use and acceptance of the Managed Instruction System by teachers
across the school district presents an important opportunity to assess the
impact of such a system on student achievement, since an essential precondi-
tion – widespread use by teachers – is met.

We asked whether students experienced greater learning gains at schools
where the conditions were supportive of data use: that is, where the
Managed Instruction System was more widely accepted and used and where
analysis of student data was more extensive? We address this question
using two types of data: student scores on standardized tests, measured over
time, and data from two teacher surveys fielded by the School District of
Philadelphia in the spring of 2006 and the spring of 2007.

The Organizational Learning Framework and Key Research Questions

As described in Chapter One and depicted again in Figure 3.1 on page 32,
the model of data use in schools posits that the organizational learning
framework involves analysis of data on student learning, followed by deci-
sions about instructional practices. When these instructional decisions are,
in turn, reflected in the instruction that teachers actually deliver, increased
student performance may result. In this model, then, four activities by teach-
ers are essential to using data to increase student learning:

1) organization of data,
2) thoughtful analysis of student data and informed decisions about
how instruction should be modified in response to the data,
3) faithful implementation of the instructional decisions, and
4) assessment of the effectiveness of instructional strategies.

The model implies that the links in the chain and the quality of the activities
can affect how much students learn. The model also highlights the human,
social, and material conditions – for example, the quality of leadership and
relationships among staff, access to technology, professional development –
that increase the likelihood of teachers being able to make good use of stu-
dent data.
31

Documenting the skill with which teachers carry out the data analysis and
subsequent instructional decisions requires a close examination of the
strength of feedback systems within a school. Chapters Four and Five draw
on in-depth qualitative research to explore the quality of the conversations,
strategies, and decisions that arose from examining student data. Using the
teacher survey data, however, we can make a broad assessment of the links
between student achievement and school conditions that are fundamental for
good data use in a Managed Instruction System.

Figure 3.1 depicts the organizational learning model that we incorporate into
the quantitative analysis presented in this chapter. Specifically, we can
examine whether teachers embraced the MIS; the availability of certain
material resources for, and expertise in, examining data (human capital); the
professional climate at the school (social capital and professional communi-
ty); and gains in student achievement. We cannot observe the faithfulness
with which teachers followed the feedback loop or the quality of their discus-

Figure 3.1 Conceptual Framework

Context School Capacity Outcome
Hu

l
ta

No
m

pi
an

Ca

Child
Ca

al
ci
p

Left Gains in
ita

Accessing
So
l

Behind and organizing Student
data
Policy Achievement
Feedback systems Sense-
Assessing and making
in Instructional to identify
modifying
solutions Community problems and
solutions

School
District Trying
solutions
es

Managed
St
rc

ru
ou

Instruction
ct
es

ur

System
lR

al
ia

Ca
er

p
at

ac
M

ity

32

sions, decisions, and follow-up in their classrooms. However, if we observe
that student learning growth is greater at schools where conditions are more
supportive of the use of a Managed Instruction System and examination of
student data, then – even if we cannot examine each part of the organiza-
tional learning model – we will have preliminary quantitative evidence that
examination of student data can result in greater student learning.

Analytic Approach

Our analysis relies on measurement of student academic growth, obtained
from longitudinal data on student achievement made available by the School
District of Philadelphia. (See the boxed text on page 34 for a description of
how we created a measure of student academic growth.)

Data on whether conditions at schools were conducive to organizational
learning that used analysis of student performance data as a driver were
obtained from surveys of teachers conducted by the School District of
Philadelphia during the spring of 2006 and 2007. These surveys included
questions about school leadership, climate, and collegiality, developed and
documented by the Consortium on Chicago School Research, as well as sev-
eral sets of questions on teacher satisfaction with the Core Curriculum and
Benchmark assessments, the amount of professional development for analy-
sis of student data, access to technology that could enable viewing student
data online, and collective examination of data with fellow teachers and
school leaders. The scales are described briefly in the data section, below,
and in more detail in Appendix E.

Our first analytic step was to examine the extent to which teachers’ reports
about each school condition were correlated with their reports about other
school conditions. We assessed these correlations by using data at the
teacher level. This descriptive work was intended to clarify whether and how
school conditions tended to occur together in “packages.”

Our second step was regression analysis to examine associations between
student achievement and each school condition separately, controlling for
individual student characteristics and the percentage of low-income students
at the school. We used a two-level hierarchical linear model to analyze the
relationship between student test score gains and teacher survey measures,
aggregated to the school level. At Level One (the student level), we used
individual-level student information to adjust for student gender, special
education status, race/ethnicity, grade when taking pre-test, and grade when
taking post-test. At Level Two (the school level), we controlled for the per-
centage of students receiving free or reduced-price lunch, using a categorical
33

Measure of Student Academic Growth
Measure of Student Academic Growth

To create a measure of student academic growth, we examined changes in
students’ performance on standardized tests given at the end of successive
school years. This strategy sometimes is known as a value-added approach
because it examines the “value added” to learning by attending school in a
given year. By comparing the score in the first year to the score in the sec-
ond year, we obtained an estimate of how much new learning students
experienced during a school year of interest. In this chapter, we examine
improvement in student academic growth in two school years (2005-2006
and 2006-2007), for students in 4th through 8th grades.

To obtain a true value-added estimate, students must have taken two tests
that are vertically scaled, meaning that the tests have been created to
measure the growth in the same kinds of skills and knowledge in the same
way. These vertically scaled tests become part of a family of assessments,
such as the Terra Nova, Stanford Achievement Test, or, potentially, a state-
developed assessment. A complicating factor for this analysis was that some
of the tests students took in different years were not vertically scaled – in
other words, they were part of different families of tests. To address this
incompatibility between tests, we converted the student’s score on each test
to a ranking within the district. Students who made learning gains relative
to other students in the district in a given year received a positive value for
their learning during that year; those whose learning did not keep up with
other students in the district received a negative value for the year’s learn-
ing. For example, a student who scored at the 50th percentile in the district
at the end of grade three and in the 52nd percentile at the end of grade
four would have “moved ahead” of his peers by experiencing greater learn-
ing gains. Students who had a test score at only one point in time were
excluded from the analysis.

It is essential to understand that the measure of learning that we examined
is explicitly comparative. While all students could have learned something
(and likely did learn) during a given school year, only students who
improved their standing in the ranking of students within the School District
of Philadelphia received positive scores. (For a technical description of this
method, see Appendix D).

34

variable with four categories. More detail on the model is presented in
Appendix D.

In our third step, we used multiple regression to determine the school vari-
ables that were most strongly associated with student achievement. We con-
ducted this regression knowing from steps one and two that many of the
school variables were strongly related to each other and to student achieve-
ment. What we looked for in the multiple regression were “points of lever-
age” – that is, school characteristics associated with higher achievement that
districts could focus on in efforts to improve instruction.

Since the teacher survey was confidential, we could not link teachers’ survey
responses to achievement outcomes for the specific students they taught.
Therefore, for the regression analyses, we aggregated teachers’ responses to
the school level, which allows us to observe the mean (average) score on par-
ticular items for each school. For example, schools with a higher mean value
on an item about the quality of school leadership are interpreted as having
stronger school leadership. In order to be sure that a school’s mean response
was not determined by just a few staff members, we included schools in the
analysis only if at least 30 percent of the teachers responded to that item.
Since we could not determine the exact number of teachers in the school who
taught in Benchmark subjects and Benchmark grades, we looked to see
whether 30 percent of all teachers at the school responded to the survey.
For this reason, we created the score for the school by using data from all
teachers-respondents, rather than just those who were teaching
Benchmarks.

Student Test Score Data

Student test score data from spring 2005, 2006, and 2007 were incorporated
into the analysis for students who were in grades 4 through 8 during 2005-
2006 and/or 2006-2007. The tests were either the Terra Nova or assessments
from the PSSAs, depending on the grade and year. Raw scores for each stu-
dent were converted to their percentile score within the district during the
year, and these scores then were converted to standardized scores with a
mean of zero and a standard deviation of one.

35

Teacher Survey Data

In June 2006 and June 2007, the school district distributed a pencil-and-
paper survey to all of its approximately 10,500 teachers. The survey asked
teachers to report on their instructional practices and use of data to inform
instruction, as well as the quality of leadership, the amount of teacher colle-
giality, and the general climate in their school. In addition, teachers were
asked about the subject(s) they taught and the grade span in which they
were teaching.

A number of the survey questions were borrowed from the indicators of
school leadership and climate developed by the Consortium on Chicago
School Research and field-tested in surveys of teachers in the Chicago Public
Schools. The indicators are described briefly below. More detail on the indi-
cators appears in Appendix E.

Instructional Leadership
Instructional Leadership.
This indicator measures the quality of school leadership in the areas of
use of student data, monitoring of instructional quality, and setting clear
goals and high expectations for teachers. Since this indicator is refer-
enced frequently throughout the rest of this chapter, it is important to
note that it incorporates a number of items about the emphasis of the
school leadership on using data to track student progress.

Professional Climate
Commitment to the School.
This indicator measures the extent to which teachers would prefer to work
at their school than at any other school and would recommend the school
to parents.

Instructional Innovation and Improvement.
This indicator summarizes teachers’ reports about whether their
colleagues try to improve their teaching and are willing to try new
strategies.

Teacher Collective Responsibility.
This indicator measures teachers’ sense of responsibility for their
students’ academic progress and for the overall climate of the school.

36

In addition, a number of survey items measured satisfaction with, and use of,
elements of the Managed Instruction System. Brief descriptions follow below
and detailed descriptions are provided in Appendix E.

Managed Instruction
Use of the Core Curriculum.
This measure is created from teacher reports about how much the
Core Curriculum guides their topic coverage, instructional activities,
and assessment strategies.

Satisfaction with Benchmarks.
This indicator measures teachers’ beliefs and attitudes about
whether the Benchmark assessments provide useful information
about student progress in a timely and clear manner.

Collegial Instructional Responses to Student Data.
This indicator measures how often during the year teachers met
with colleagues at their school to discuss re-teaching a subject or
re-grouping students, based on examination of Benchmark scores.

Technology Access and Support.
This indicator measures classroom Internet access, working
computers, and technology support for teachers. The indicator is
not specific to the Managed Instruction System. However,
student scores on Benchmarks and suggestions for instructional
modifications are available on the web. Technology in good
working order and support for its use would make it easier for
teachers to make full use of the Managed Instruction System.

Professional Development on Data Use.
This indicator measures whether, during the school year, the school
offered professional development on how to access and interpret
student performance data.

37

Findings

Associations Among School Characteristics

Our first analytic step was to examine the correlations among three sets of
variables: the measure of instructional leadership, measures of positive pro-
fessional climate among teachers (teacher commitment to the school, colle-
gial climate, and innovation), and measures of managed instruction (use of
the Core Curriculum, satisfaction with the Benchmark assessments, access
to technology, collegial discussions of instructional responses to student data,
and professional development). These correlations, presented in Table 3.1,
are from the 2007 teacher survey. Only teachers who were teaching subjects
and grades that used Benchmark exams are included in this correlation
matrix, but the values are very similar when all teachers are included.

Table 3.1 Pearson Correlation Matrix for Key Teacher Survey Variables (2007 Survey)

Instructional leadership 1.00
Professional Climate

Commitment to the school .58 1.00 Managed Instruction

Innovation
Context .38 .31 1.00

Teacher collective responsibility .41 .41 .82 1.00
e rs h i p a nd
L e ad
Use of Core Curriculum .21 .17 .14 .17 1.00

Satisfaction with Benchmarks .20 .18 .15 .21 .29 1.00

Collegial instructional responses .41 .18 .14 .18 .23 .33 1.00

Technology access and support .31 .32 .25 .26 .12 .15 .14 1.00

Professional development on data use .28 .18 .10 .10 .16 .09 .23 .10 1.00
e
a us
at
es
y
lit

t

nd
or
ns
ks
ibi

pp
po

to
ar
ns
ol

su
es
hm

en
po
ho
ip

lr
m

nd

pm
nc
es
sh

sc

ulu

na

sa
er
er

Be

elo
he

Ins
tio
ric

y
ad

es

nit
tiv
ot

th

ev
uc
ur

tr u c
cc
l le

ec

wi

t i on al C o mm u
ld
tt

str
eC

ya
oll
na

en

ion

na
l in
n

or

log
rc
tio

itm

sio
tio

ct
fC

gia
uc

he

no
fa
va

es
mm

eo

lle
tis
str

ac

ch
no

of
Co

Co
Us

Sa
Te

Pr
Te
In

In

38

There are moderate-to-strong positive associations within the group of vari-
ables that speak to instructional leadership and positive professional climate
The school characteris-

among teachers (teacher commitment to the school, collegial climate, and
tics of strong instruc-
innovation). For example, the correlations between instructional leadership, tional leadership, a pos-
on the one hand, and the professional climate variables, on the other, range
from .38 to .58. Further, the correlations among the three variables that
itive professional cli-

address professional climate are particularly strong, ranging from .41 to .82. mate, investment in the
Finally, and importantly, the correlation matrix also shows that strong
instructional leadership and a positive professional climate are positively
Managed Instructional

associated with the five “managed instruction” variables.
System, and use of
student data to inform
A reasonable conclusion from these correlations is that the school character-
istics of strong instructional leadership, a positive professional climate,
instruction tend to be

investment in the Managed Instructional System, and use of student data to found together. They co-
inform instruction tend to be found together. That is, they co-occur as “pack-
ages” because schools that are “good” in one respect tend to be “good” in
occur as “packages”

other respects; schools with strong instructional leadership are often schools
because schools that
where teachers trust each other and encourage their colleagues to innovate are “good” in one
and grow professionally. From a research perspective, these characteristics
of schools can be difficult to separate analytically, requiring us to choose one
respect tend to be

variable to serve as a proxy for a range of favorable conditions at the school. “good” in other

That said, it is notable that of the four variables that describe school leader-
respects.

ship and professional climate, instructional leadership has the strongest
relationship with the five variables related to the Managed Instruction
System. For example, the correlation for instructional leadership and the fre-
quency with which teachers met to discuss instructional responses to student
data is .41, while the correlation between innovation and discussion of
instructional responses to data is just .14. It is worth recalling that, in this
study, instructional leadership refers to the extent to which the school lead-
ership emphasizes data-driven decision-making, tracks student progress,
knows what kind of instruction is occurring in classrooms, and encourages
teachers to use what they learn from professional development. It makes
sense, then, that instructional leadership, defined in this way, would be a
good predictor of how often teachers met to discuss instructional responses to
student data (the collective examination variable) as well as the amount of
professional development provided on topics related to student data.

Our model of organizational learning posits that the quality of school leader-
ship is an important factor that supports “take-up” of the Managed
Instruction System and collective examination of student data. It is not diffi-
cult to imagine that instructional leadership would be an important condi-
tion that would allow innovation and collegial learning – including analysis
39

of student data – to operate. The moderate or strong relationship between
instructional leadership and every other variable presented in Table 3.1 sup-
Learning from data is

ports this argument. Further, the centrality of the instructional leadership
a social activity.
variable to effective data use by faculty is shown in subsequent analyses in
this chapter.
Benchmark data are
useful to teachers
Also of note is that among the five MIS variables, the highest correlations
are between perceptions of the usefulness of Benchmark assessments and
when they have oppor-

frequency of examination of student data with colleagues (r=.33) and useful-
tunities to discuss
ness of Benchmarks and use of the Core Curriculum (r=.29). The first corre- them with colleagues.
lation supports the idea that learning from data is a social activity.
Benchmark data are useful to teachers when they have opportunities to dis-
cuss them with colleagues. The second correlation indicates the mutually
reinforcing relationship between the Core Curriculum and the Benchmarks
that the district intended. The more teachers invest in the Core Curriculum
by adhering to it, the more useful Benchmark assessments are likely to seem
as a tool to guide instruction, since the Benchmarks are aligned with the
Core Curriculum. The reverse is also likely to be true: the more a teacher
finds results from Benchmark assessments to be informative, the more will-
ing he or she is likely to adhere to the Core Curriculum.

Relationships between School Characteristics and Achievement

The preceding section emphasized the positive relationships among instruc-
tional leadership, a positive professional climate, use of key elements of the
Managed Instruction System, and support for teachers’ use of the student
data. In this section, we use a multilevel model to examine the relationships
between each of these variables (aggregated to the school level) and growth
in student learning. Since the instructional leadership, professional climate,
and MIS variables are so inter-related, we examine separately the associa-
tion between each variable and student achievement growth. Beginning on
page 42, we identify and discuss the school variables that are the strongest
and most consistent predictors.

Table 3.2 presents the coefficients from separate multilevel regressions pre-
dicting mathematics and reading growth in 2005-2006 and 2006-2007.
Thirty-six separate regressions are represented in the table. The variables
are standardized so that the magnitude of the effects can be compared.

There are several important patterns to note in Table 3.2. First, almost
every variable is a statistically significant predictor of learning growth.
Second, there is a positive relationship between all of the school variables
and student learning growth. Schools where teachers reported stronger

40

Table 3.2 Relationships between Student Learning Growth and School Variables

Reading 2005-06 Math 2005-06 Reading 2006-07 Math 2006-07
Estimate p* Estimate p Estimate p Estimate p

Instructional Leadership 0.11** 0.000 0.12 0.000 0.17 0.000 0.15 0.000

Commitment to the School 0.18 0.000 0.18 0.000 0.17 0.000 0.14 0.000

Instructional Innovation & Improvement 0.20 0.000 0.20 0.000 0.15 0.000 0.16 0.000

Collective Responsibility 0.19 0.000 0.18 0.000 0.14 0.000 0.15 0.000

Use of the Core Curriculum 0.18 0.000 0.14 0.001 0.13 0.002 0.09 0.040

Collegial Instructional Responses 0.13 0.000 0.11 0.001 0.03 0.510 0.03 0.530
School Capacity
Technology Access and Support 0.15 0.000 0.14 0.000 0.10 0.000 0.08 0.001

Professional Development on Data Use 0.13 0.010 0.14 0.007 0.14 0.001 0.13 0.006

Satisfaction with Benchmarks 0.04 0.380 0.02 0.650 0.07 0.078 0.07 0.140

*The p-value is the probability that the estimate is simply the result of chance.
** Statistical significance is indicated in bold type.

instructional leadership, a more positive professional climate, greater use of
the Core Curriculum, and more supports for data use by teachers experi-
enced greater learning gains than schools without the same positive fea-
tures. The effects of the school variables are observed even after controlling
for individual student characteristics (demographics, special education or
English Language Learner status, and grade in school) and the percentage of
students at the school who were from low-income families.

In Table 3.2, the coefficients range approximately from .10 to .20 for each year
and each subject. Generally speaking, the instructional leadership and profes-
sional climate variables have slightly larger impacts on achievement than the
MIS variables, although the magnitudes of the effects are quite close. For
example, for reading growth during the 2006-2007 school year, the magnitude
of the effect for instructional leadership was .17, in contrast to .10 for techno-
logical access and support and .13 for use of the Core Curriculum. An effect of
.17 is considered to be of moderate size in education research.36 That is, for
each one standard deviation increase in the mean reported quality of the
school’s instructional leadership, the school’s achievement ranking in the dis-
trict was predicted to increase by .17 of a standard deviation.

36
Lipsey, M. W., and Wilson, D. B. (1993). The efficacy of psychological, educational, and behav-
ioral treatment: Confirmation from meta-analysis. American Psychologist, 48, 1181-1209.
41

There are two variables that, at least in some years, do not have statistically
significant associations with achievement growth. A measure of satisfaction
A measure of satisfaction

with Benchmarks was not significantly associated with either reading or math
with Benchmarks was not
achievement growth, for either 2005-2006 or 2006-2007 (although it
approached statistical significance at α=.05 in 2006-2007). Likewise, a measure
significantly associated

of collegial instructional responses to student data was not a significant predic-
with either reading or

tor in 2006-2007. The direction of the coefficients was positive in all cases. math achievement

The framework that informs this study may provide some insight on the weak
growth.

relationship between satisfaction with Benchmarks and achievement. The
framework hypothesizes that the link between the data itself and student
achievement is moderated by interpretation, subsequent instructional deci-
sions, implementation of those decisions, and assessment of those decisions.
The measure of satisfaction with Benchmarks tells us about only a small piece
of that process: whether the teachers felt that Benchmarks provided useful,
clear, and timely information about student progress. It does not tell us
whether teachers had good ideas about how to respond to the data. Although
accessing clear data in a timely way is important, it is insufficient for produc-
ing student achievement. As the case studies of the next chapter show, the
ability of teachers to make sense of the data and plan appropriate instruction-
al responses is heavily contingent on school resources, especially the quality of
leadership and support provided by the principal and content area teacher
leaders. It is also possible that there were inadequacies in the quality of the
Benchmark assessments that lead to a weak relationship between teachers’
satisfaction with the Benchmarks and gains in student achievements. As stat-
ed in the Introduction, a review of the technical quality of the assessments was
beyond the scope of this study.

Identifying the Strongest Predictors of Achievement

In our final step, we used multivariate regression to identify school charac-
teristics that had an especially strong relationship with achievement. Our
purpose in so doing was to assess whether there were particular organiza-
tional characteristics on which education leaders could focus in order to help
teachers make the most of student data.

When the relative strength of the four instructional leadership and school
climate variables was tested in multiple regressions, the two variables that
had the strongest and most consistent relationships with student achieve-
ment across years and subjects were instructional leadership and teacher col-
lective responsibility. We then added each of the five MIS variables to a
regression with either the instructional leadership or collective responsibility

42

measures. One of these MIS variables – use of the Core Curriculum – was a
statistically significant predictor of student achievement growth in some
Schools with stronger

years and for some subjects.
instructional leader-
ship, a stronger sense
Table 3.3 presents the results of two regressions that include use of the Core
Curriculum along with instructional leadership and collective responsibility,
of collective responsi-

respectively. When instructional leadership and use of the Core Curriculum bility among teachers,
are included together as predictors of achievement, the magnitude of the
leadership effect ranges from .08 to .15; the Core Curriculum effect is signifi-
and/or greater use of

cant for reading and mathematics in the 2005-2006 school year; and the
the Core Curriculum to
r-squared ranges from .06 to .12. The magnitudes of the effects and the r- inform content, instruc-
squared are similar for a regression that includes collective responsibility
and use of the Core Curriculum. Substantively, these regressions suggest
tion, and assessment

that schools with stronger instructional leadership, a stronger sense of col- produced greater
lective responsibility among teachers, and/or greater use of the Core
Curriculum to inform content, instruction, and assessment produced greater
student learning gains

student learning gains than other schools.
than other schools.

None of the other Managed Instruction System (MIS) variables was a signifi-
cant predictor of achievement growth when entered into a regression with
instructional leadership or teacher collective responsibility.

Table 3.3 Key School Variables Predicting Growth in Student Learning

Reading 2005-06 Math 2005-06 Reading 2006-07 Math 2006-07
estimate p estimate p estimate p estimate p

Instructional Leadership 0.08* 0.010 0.10 0.002 0.15 0.000 0.15 0.000

Use of the Core Curriculum 0.15 0.002 0.10 0.030 0.04 0.300 .00 0.976

R-squared at Level 2 (school level) .08 .06 .12 .09

Collective Responsibility 0.17 0.000 0.17 0.000 0.13 0.000 .14 0.000

Use of the Core Curriculum 0.12 0.004 0.08 0.060 0.08 0.053 .03 0.476

R-squared at Level 2 (school level) 0.13 .10 .09 .07

* Statistical significance is indicated in bold type.

43

In Summary

In this chapter, we discussed the results of our efforts to disentangle the
impact of various factors on growth in student achievement. Importantly,
While Benchmarks may

we found that some factors were stronger and more consistent predictors of
be helpful, they are not
achievement gains than others. In particular, we found that instructional
leadership and collective responsibility were strong predictors of learning
in themselves sufficient

growth. Use of the Core Curriculum was also a robust predictor, showing
to bring about increases

more power in 2005-06 and in reading than in math. The implications of in achievement without
these findings, we suggest, are powerful. In particular, we suggest that
translating student data into student achievement requires a strong learning
a community of school

community at the school. The instructional leadership and collective respon-
leaders and faculty who
sibility measures imply that school leaders and faculty feel accountable to
one another, that they are diligent in monitoring student progress, and that
are willing and able to

they are willing to use data as a starting point for inquiry.
be both teachers and
learners.
It is notable that these measures of school leadership and school community
are stronger predictors of student learning growth than satisfaction with the
usefulness of Benchmark data. While Benchmarks may be helpful, they are
not in themselves sufficient to bring about increases in achievement without
a community of school leaders and faculty who are willing and able to be
both teachers and learners.

44

Chapter Four
Making Sense of Benchmark Data
The quantitative analysis presented in Chapter Three established that
strong instructional leadership and collective responsibility were the most
Few studies of schools

robust predictors of growth in student achievement, with use of the Core
have looked closely

Curriculum being slightly less robust. It also highlighted the difficulty of enough at how school
analytically separating individual characteristics of schools such as instruc-
tional leadership, professional climate, use of the Core Curriculum, and use
leaders facilitate

of student data to inform instruction. These characteristics tended to co-
collective interpretation
occur as “packages.” of data in instructional

In this chapter we use our qualitative data to uncover what school leaders –
communities – what do

principals and teacher leaders – actually do as they work with teachers in practitioners talk about
instructional communities to make sense of Benchmark results and plan
instructional actions. We wanted to determine, what can school leaders do to
and how do they talk

ensure that the use of Benchmark data contributes to organizational learn-
ing and ongoing instructional improvement within and across instructional
communities?

In theory, instructional communities, such as grade groups, provide “an ideal
organizational structure” for school staff to learn from data and use data to
improve student learning.37 “Organized talk”38 in instructional communities
is foundational for building shared understanding of issues and concerted
efforts to remedy problems. In the four-step feedback system described in
Chapter One, organized talk is represented in the second step, “sense-mak-
ing with data to identify problems and solutions.” (See Figure 4.1) School
leaders have a key role to play in facilitating interpretation of data to create
actionable knowledge.39 But few studies of schools have looked closely
enough at how school leaders facilitate collective interpretation of data in
instructional communities – what do practitioners talk about and how do
they talk about it. We use our observations of grade group meetings to exam-
ine and assess the quality of interpretation processes and the factors that
influenced that quality.

37
Mason, S. A. & Watson, J. G. 2003.
38
Rusch, E. A. (2005). Institutional barriers to organizational learning in school systems: The
power of silence. Educational Administration Quarterly, 41, 83 – 120. Retrieved on May 8, 2007,
from SAGE Full-Text Collections.
39
Daft, R. L. & Weick, K. E. (1984). Towards a model of organizations as interpretation systems.
Academy of Management Review, 9(2), 284-295.
45

Figure 4.1 Feedback Loop for Engaging with Data
Strategic sense-making
focuses on the
Accessing identification of short-
and organizing
data term tactics that help
a school reach its
Feedback Sense-
Assessing systems making Adequate Yearly
and modifying in Instructional to identify
to
solutions Community problems and
problems and Progress (AYP) targets.
solutions
solutions

Trying
solutions

Three Kinds of Sense-Making: Strategic, Affective, and Reflective

Our observations of grade groups suggest that practitioners engaged in three
major types of sense-making as they sat together to discuss and interpret
Benchmark data: strategic, affective, and reflective. Not surprisingly, the
pressures of the accountability environment strongly influenced their sense-
making. However, our observations also showed that the actions of school
leaders could mediate these policy forces to create instances of substantive
professional learning for school staff. Disappointingly, such instances were
infrequent. There is an important opportunity for the district to strengthen
the impact of Benchmark data on teacher and student learning. Below, we
discuss the three kinds of sense-making.

Strategic sense-making focused on the identification of short-term tactics
that help a school reach its Adequate Yearly Progress (AYP) targets.
Strategic sense-making included conversations about “bubble students” who
have the highest likelihood of moving to the next level of performance (from
Below Basic to Basic or from Basic to Proficient) thereby increasing the prob-
ability that the school would meet its AYP goal. These conversations related
to the predictive purpose of interim assessments in the framework offered by
Perie et al.,40 described in the Introduction. Strategic conversations also
focused on improving test-taking conditions and test-taking strategies.

40
Perie, M. et al., 2007.

46

Three Kinds of Sense-Making: Strategic, Affective, and Reflective

Strategic Sense-Making: Most Common

Focuses on short-term tactics that help a school reach its Adequate Yearly
Progress targets, including having conversations about students who have
the highest likelihood of moving to the next performance level.

Affective Sense-Making: Common

Focuses on teachers’ professional agency and responsibility, beliefs about their
students, desire to encourage one another, and motivate their students.

Reflective Sense-Making: Least Common

Focuses on questioning and evaluating the instructional practices used in the
school and what teachers need to learn in order to help students succeed.

Finally, in strategic conversations, practitioners used Benchmarks for evalu-
ative purposes as they worked to identify strengths and weaknesses that cut
across grades and classrooms so that they could allocate resources (staff,
materials, and time) in ways that increased the odds that the school would
meet its AYP goal (e.g., assigning “strong” teachers to the accountability
grades, purchasing calculators, lengthening instructional time for literacy
and mathematics). In our observations, strategic sense-making dominated
the talk about Benchmark data.

Affective sense-making included instances in which leaders and classroom
teachers addressed their professional agency, their beliefs about their stu-
dents, their moral purpose, and their collective responsibility for students’
learning. During affective talk, school leaders and teachers offered one
another encouragement. They expressed a “can do” attitude, often relating
this sense of professional agency back to the pressures that they felt from
the accountability environment. In affective talk, practitioners also affirmed
their belief that their students “can do it.” They discussed how to motivate
their students to put forth their best effort on standardized exams and in
general. Affective sense-making was the second most prevalent kind of dis-
course that we observed.

Reflective sense-making occured when teachers and leaders questioned and
evaluated the instructional practices that they employed in their classrooms
and their school. They connected what they were learning about what their

47

students knew and did not know to key concepts in the Core Curriculum and
they identified resources that would help them strengthen instruction of
Reflective sense-

those concepts. Researchers have pointed out the importance of reflective
making offers the
discourse as “a springboard for focused conversations about academic content
that the faculty believes is important for students to know.”41 These conver-
most promise for

sations helped teachers focus on what they needed to learn in order to help
building instructional

their students succeed. Such discourse about the curriculum served to shift capacity because it
teachers’ attention away from students’ failures and towards analyzing and
strategizing about their own practices.
focuses on teachers’
learning.
In summary, reflective conversations helped practitioners plan the kinds of
professional development that would strengthen teachers’ understanding
and use of the Core Curriculum. They generated consideration of what other
kinds of data they needed to take into account as they made sense of the
Benchmark results. They offered the most promise for building increased
school and classroom instructional capacity.

Making Sense of Benchmark Data: Four Examples

Below, we use fieldnotes from observations of grade group meetings in four
schools to construct descriptions of the typical processes of school leaders
and grade groups as they made sense of Benchmark data. These grade group
meetings were consistent with what teachers and school leaders told us
about their use of Benchmark data in interviews and with other types of
meetings that we observed. The examples provide windows into why
instances of strategic and affective talk were so prevalent. They also shed
light on why the survey variable, teacher satisfaction with Benchmarks, was
not associated with gains in student achievement. Finally, they suggest
opportunities for increasing instances of reflective conversations about
Benchmark results as a springboard for staff to learn more about their stu-
dents, the curriculum, and pedagogy.

Attendance at each of the four meetings that we describe below consisted of
the school’s principal, at least one teacher leader (usually a reading or math
coach), and between two to four classroom teachers.42 In the four schools,

41
Mintz, E., Fiarman, S. E., & Buffett, T. (2006). Digging into data. In K. P. Boudett, E. A. City,
& R. J. Murname (Eds.), Data wise: A step-by-step guide to using assessment results to improve
teaching and learning (81-96). Cambridge, MA: Harvard Education Press, p. 94.
42
In order to minimize some aspects of variation and to focus on different types of sense-making
relative to Benchmark data, these examples are drawn from a small subset of observations con-
ducted between January 2005 and December 2006 in which the organizational context of the
observations (grade group meetings) and the tools (the Benchmark Item Analysis Report) were
held constant.
48

grade group meetings generally occurred every week or every other week
and involved teachers from the same grade or from consecutive grades (K-2,
“The re-teaching

3-5). In each of the examples, school leaders and teachers were using the dis-
opportunity can be
trict’s Item Analysis Report available on SchoolNet. (See page 26 for a powerful, especially
description of the Item Analysis Report.) In some grade groups, principals
played particularly prominent roles, but in every grade group, teacher lead-
if it’s done right after

ers, and to a lesser extent, classroom teachers, also were active participants. students take the test

Sense Making Example 1: Encouraging re-teaching to emphasize procedures
and it is fresh in their
for multi-step math problems minds.”

The principal opened the discussion of the Benchmark data by ask- - A Principal
ing: “How many students are Proficient or Advanced? How many are
close to Proficient or Advanced? What are the questions that gave the
students the most problems?” Teachers took time to use colored high-
lighters to note students’ different status and to make decisions
about tutoring assignments.

A 4th grade teacher pointed out that most of her students missed a
question about the length of a paper clip because they didn’t notice
that the paper clip was placed at the 2 cm mark on the ruler in the
picture, not at 0: “They needed to subtract 2 to get the right
answer.” The math teacher leader reassured the 4th grade teacher
that “It’s the evil test makers at work. Nobody ever starts measuring
something from 2 cm.”

The principal chimed in with sympathetic comments about test ques-
tions that defy common sense. She also reminded the teachers that
re-teaching can be an opportunity to point out what students must
keep in mind as they approach test items on the Benchmark and
PSSA tests. “The re-teaching opportunity can be powerful, especially
if it’s done right after students take the test and it is fresh in their
minds. Sometimes it’s two or three steps (in a math problem) that
you need to get to in order to get the right answer.”

Later in the meeting, the principal offered to teach a lesson about
fractions and decimals to the 4th graders, another concept that had
stumped many students.

Many of the meetings we observed began in the same way that this one did,
with the principal or a teacher leader asking: “How many students are
Proficient or Advanced? How many are close to Proficient or Advanced?”

Even though the Benchmark data are meant to provide diagnostic informa-
tion about what students have learned in the previous five weeks, conversa-
tions about results often assumed that they were predictive of performance
on the PSSA – evidence of how the state’s accountability measure pervaded
practitioners’ thinking about what they could learn from the Benchmark 49

data. Practitioners from all of the schools in our qualitative sample reported
that the identification of bubble students – students on the cusp of scoring
The principal encour-

Proficient or moving from Below Basic to Basic – was a common practice in
aged teachers to help
their analysis of Benchmark data. their students believe
they “can do it” – an
The teachers put stars next to those kids that they’re going to target.
And we made sure that those kids had interventions, from Saturday
example of affective
school to extended day, to Read 180. And then we followed their sense-making in
Benchmark data. Those were the kids that the teachers were really
going to focus on, making sure that those kids become Proficient, or which school-based
move that 10 percent out of the lower level so that we can make Safe
practitioners focus on
Harbor next year. (Teacher, 2006)

School leaders reported that they were encouraged by the district and
how to motivate their

provider staff who worked with their schools to pay attention to proficiency students.
levels and to track the progress of students who would be most likely to
score proficient with additional supports.

The principal in this example implored teachers to strike while the iron was
hot and take advantage of the re-teaching opportunity immediately so that
students could see where they went awry – a strategy that research on form-
ative assessment recommends.43 And, in fact, all of the teachers at this school
made a practice of going over responses to assessment items with their class
right after they finished the test. In this example, however, the principal
focused on re-teaching the procedural aspects of the math problem (“some-
times it’s two or three steps that you need”), rather than returning to the
concepts under study – a point that we will take up again in Example 2.

Sense Making Example 2: Identifying motivational strategies and tutoring
resources

At this school, the 5th grade teachers said that their students were
having a lot of difficulty with Benchmark items related to fractions,
particularly reducing improper fractions. One teacher noted that she
had connected fractions to a lesson that she had done earlier and
that, “A lot of light bulbs went off [when students saw how to draw
on what they already knew].” Building on this, the principal said that
she loved the image of students “tapping into prior knowledge” and
suggested that everyone make posters of light bulbs for their class-
room to motivate students during the Benchmarks and other tests.
“Tell students to hang up a light bulb, put on your thinking caps and
say ‘I can do it.’” The principal also pointed out that their volunteer
tutors might be a good resource to help students who were having
trouble with fractions.

43
Black, P. & Wiliam,D. 1998.

50

In this example, the principal diverted the conversation to address how to
motivate students. She encouraged teachers to help their students believe
Discussions about

they “can do it” – an example of affective sense-making in which school-
Benchmark data most
based practitioners focus on how to motivate their students. often did not focus on

As in the previous example, no one in the meeting addressed conceptual
building teachers’

issues related to mathematical content. Students were challenged by items “pedagogical content
related to fractions, but the conversation did not explore the intended pur-
pose of these questions. As Spillane and Zeuli (1999)44 found in their study of
knowledge.” Deep

mathematics reform, our research indicates that discussions about
understanding of con-
Benchmark data most often did not focus on building teachers’ “pedagogical tent makes it possible
content knowledge.”45 for teachers to explain

Pedagogical content knowledge couples knowledge about content to knowl- disciplinary concepts
edge about pedagogy. Teachers with strong pedagogical content knowledge
understand what teaching approaches fit the content being taught; their
to students and to

deep understanding of content makes it possible for them to explain discipli-
craft learning tasks
nary concepts to students and to craft learning tasks that build students’ that build students’
conceptual understanding; their broad repertoire of instructional strategies
provide them with options to help students with different learning needs.
conceptual

The alignment of Benchmark assessments with the Core Curriculum offers understanding.
the opportunity for teachers to look at results with an eye towards strength-
ening their pedagogical content knowledge. Our observations of grade group
meetings and our interviews with school leaders indicate that this was
rarely a focus of practitioners’ analysis.

Sense Making Example 3: Revamping classroom routines to support student
independence

The math teacher leader suggested that middle grade students need
more independence during regular classes in order to improve their
performance on tests. “One of the reasons that people say the kids
know the material, but don’t test well, is that the conditions are so
different. During instructional periods, you need to let the kids do
more on their own, so it’s more like a testing situation where they
have to interpret the instructions on their own.”

He suggested that the teachers should tell students the objective for
the lesson, then have them work in small groups to figure out what is
being asked of them in the directions for the math activity. Teachers

44
Spillane, J. P. & Zeuli, J. S. (1999). Reform and teaching: Exploring patterns of practice in the
context of national and state mathematics reforms. Educational Evaluation and Policy Analysis,
21(1), 1-27.

Shulman, L. S. (1987). Knowledge and teaching: Foundations of the new reform. Harvard
45

Educational Review, 57(1), 1-22.
51

should circulate during this time, noting where students are on the Follow-up by principals
right track and where they are not. They should ask questions that
will help students improve their interpretations. He concluded, “Our and teacher leaders in
students need to learn to be more independent. After they’ve finished classrooms was much
the task, then you can review and reflect with the small groups about
how it went.” less likely to occur in
most schools than one
Like the principal in the first example, this math teacher leader
offered to come into classes and help teachers if they were ready to might hope, a gap that
try out some of the new instructional practices discussed.

The math leader in this example made the broad point that students need to
weakens the kinds of

learn to work more independently and then offered specific ideas for doing
feedback systems nec-
this. Although these suggestions were meant to address problems students essary for organiza-
encounter in the testing situation, they are also good instructional practice. tional learning.

Offers of support from school leaders are prominent in Examples 1, 2, and 3,
as are teaching tips. Principals and teacher leaders offered to conduct
demonstration lessons and to consult about classroom management of small
groups. They also suggested steps that teachers might themselves take – re-
teaching, a change in classroom routines that would encourage more student
independence, ways to motivate students. We read many of these offers of
support and recommendations as ways for school leaders to demonstrate
their investment in teachers’ struggles and to encourage teachers in the con-
text of the larger accountability policy context that often stigmatizes schools,
educators, and students for low student achievement rather than supporting
and rewarding them.

Our interviews of staff suggest that follow-up by principals and teacher lead-
ers in classrooms was much less likely to occur in most schools than one
might hope, a gap that weakens the kinds of feedback systems necessary for
organizational learning. When leaders do not visit classrooms to see whether
teachers are trying the strategies discussed in grade group meetings and
whether they use the strategies well, an important evaluative function of
Benchmark assessments is lost. Leaders do not have good information to
judge the efficacy of the solutions.

Sense Making Example 4: Understanding the standards and learning how to
teach standards-based content

At a fourth school, teachers brought the Item Analysis Report for their
classrooms as well as copies of the Core Curriculum, having already
made notes to themselves about student strengths and weaknesses.
When teachers brought up the difficulty their students were having with
reading the math problems on the Benchmark assessment, the principal
reminded them that they could read the math questions to students.
52

The principal directed these fourth grade teachers to think about the The principal pushed
relationship between the Benchmark assessments and the Core
Curriculum standards in order to figure out why some questions were teachers towards the
presenting more difficulty for students than others. “Look at questions standards of the Core
that test the same standard. Are they written the same way or a differ-
ent way? Is one harder than the other?” Curriculum and raised
interesting questions for
The math teacher leader chimed in to give a specific example of how to
do this. She pointed out how two of the Benchmark items assessed stu- teacher reflection.
dents’ knowledge of scientific notation, but in different ways. She fol-
lowed up by saying that she would work with a small group of students
that were having problems with scientific notation at a time that the
classroom teachers could observe this as a demonstration lesson.

In this example, the principal pushed teachers towards the standards of the
Core Curriculum and raised interesting questions for teacher reflection. The
principal and the math teacher leader worked as a tag team; the principal
raised a broad point about noticing differences in questions about the same
standard and the math leader follows up with specific examples. In this
meeting, teachers were expected to bring the Core Curriculum and their
Benchmark data and to be prepared to discuss their preliminary analysis of
results and what they intended to do.

In Summary

It is notable that school leaders in all four schools established key organiza-
tional structures to support use of the Benchmarks – structures that were
not necessarily present in all of the other schools in our sample or across the
district. School schedules accommodated regular grade group meetings. In
addition, school leaders – the principal and teacher leaders – consistently
attended grade group meetings, ensuring that grade teachers actually gath-
ered together and sending a message that the meeting was important. The
presence of these leaders provided at least the opportunity for school leaders
to learn about teachers’ perspectives on the data, teachers’ understanding of
the Core Curriculum, and what instructional strategies teachers were using.
Their presence also provided the opportunity for school leaders to signal
instructional priorities and draw connections between what was being
learned from data in other grades that was relevant to this group of grade
teachers. Opportunities for cross-school knowledge were increased, as princi-
pals and teacher leaders shared ideas learned in one grade group with others
throughout the school. As the examples illustrate, whether and how leaders
capitalized on these opportunities varied considerably.

Across the four observations, practitioners used the Item Analysis Report to
identify student weaknesses. It is noteworthy that much of the conversation 53

about remediating gaps focused on a single test item, rather than on curricu-
lar standards or instructional approaches that would address these stan-
It is noteworthy that

dards. The format of the Item Analysis Report itself may drive practitioners
much of the conversa-
to focus on individual items. This particular report does not group together
items testing the same standard and it identifies the standard only by num-
tion about remediating

ber – thereby requiring that an educator be sitting with the Core Curriculum
gaps focus on a single

Standards in order to identify the actual content with which students are test item, rather than
struggling. The emphasis on individual items also may contribute to the
inordinate amount of time school leaders and teachers spent in discussions
on curricular standards

about test questions that were poorly worded or otherwise framed in a way
or instructional
that did not make sense or whose content had not been covered in the Core
Curriculum yet. In such cases, school leaders need to direct attention back to
approaches.

the curriculum and the standards, as the principal in Example 4 does.

It is important that school leaders have sufficient knowledge about the
Benchmarks, the curriculum, and the PSSA so that they can help teachers
stay focused on what useful information they can garner from the
Benchmarks. For example, understanding the relationship between a frac-
tion and a decimal is one of the “big ideas” in upper elementary mathematics
that has the potential to open up a discussion of what is, or is not, in the cur-
riculum for addressing this important concept. The image of an instructional
community ready to engage deeply with a content area represents quite a
different picture than most discussions about Benchmark data that we
observed or heard about.

As a consequence of reviewing Benchmark data, practitioners in the four
examples above planned actions that included:

1.Identifying students who were likely to move from Basic to Proficient or from
Below Basic to Basic and targeting them for special interventions in order to
increase the likelihood that the school will make AYP. Across the schools,
these interventions varied considerably – extended day programs, Saturday
school, work with volunteer tutors, special attention from the math or reading
specialist, computer assisted programs. It is likely that their quality varied as
well, but formal or informal assessment of the interventions was rare. As one
principal told us, “You know, we’ve never really looked to see if those tutors
are doing a good job.” (2007)

2.Identifying skills and concepts to be re-taught in the sixth week of the
instructional cycle or in subsequent units. From our data, we surmise that re-
teaching was one of the actions most frequently taken as a result of reviewing
the Benchmark results. District leaders and principals reported that there
were too many instances of teachers simply returning to the content material,
using the same instructional strategies. But some teachers reported that it

54

was important to try different instructional strategies for re-teaching an area
of weakness. As one explained,

I can see how my whole class is doing. And they [members of my
grade group] can say, “This one question, only four of your twenty
kids got it right.” So, I know that if only four kids got it right,
that’s something I need to go back and re-teach, or get a fresh
idea about how to give them that information. (Teacher, 2006)

3.Identifying students who shared similar weaknesses (or, in some cases,
strengths) for re-grouping to provide differentiated instruction. Our data indi-
cate that re-grouping was another one of the actions most frequently taken as
a consequence of reviewing the Benchmark results. Often referred to as “flex-
ible groupings,” teachers and school leaders explained that they grouped stu-
dents around shared weaknesses identified through examination of the
Benchmark data. One teacher described how “the groups constantly
changed” so that she could “target specific kids and their specific needs and
group kids according to where they were lacking.” When she felt it was appro-
priate, she would also assign different homework to different students based
on their needs. In other schools, teachers described how they had begun cre-
ating groups that cut across classrooms based on shared student weaknesses.

4.Re-thinking classroom routines that emphasized greater student independ-
ence, motivation, and responsibility for their own learning. This kind of action
was not mentioned frequently. However, one example is a fifth grade teacher
who described how she regrouped students, putting stronger students with
weaker students as a way to encourage and facilitate peer teaching.

I put the item analysis report on the overhead [for the whole
class to see]. It’s because of that relationship I have with my
students. It’s that community. So [I want my students thinking
about] why our class average is 60% when I scored 100%. I
didn’t get any wrong. We need to help our classmate that had
difficulty, that may have received 40%. That’s where I go into my
grouping. How can I pool my strong students [to work with
students who are struggling? (May 2007).

5.Identifying content and pedagogical needs of teachers to inform opportunities
for continued professional learning and other supports that addressed those
needs. Formal professional development sessions and less formal on-the-spot
coaching were also planned based on results from the Benchmarks, especially
when those data corroborated data from the PSSA. One teacher described a
particularly strong approach to supporting teachers’ learning:

We actually had a professional development about it,
where [the principal] did a lesson to show us, and then
we went to two other teachers' rooms and saw them do
a lesson. And then pretty much that whole week that
followed, [the principal] came around to see how we were
using it, if we needed any help, what other support we
needed to get this going and into play. (June 2006)

55

Each of these planned actions makes sense. Each emerged from paying
attention to data.” However, the quality of the actions varied considerably.
Spillane et al., (2002) argue that educators’ interpretations of policy man-
dates are critical to their implementation of these mandates.46 In the exam-
ples above, we note the influence of the accountability environment on edu-
cators’ interpretation of the mandate for data-driven decision-making.
Clearly, this policy context and the fact that these schools had been identi-
fied as “low performing,” influenced practitioners’ perceptions of why exam-
ining data is important. They needed to address the primary problem that
they felt compelled to solve: how to make AYP. They brought the imperative
to “do something” – some might say “do anything” – to their discussion and
interpretation of Benchmark data.

However, school leaders can mediate the high stakes accountability environ-
ment by creating opportunities for teachers to learn from Benchmark data.
Beer and Eisenstat (1996) lay out the significance of organzied talk to organi-
zational learning:

Lacking the capacity for open discussion, [practitioners] cannot arrive
at a shared diagnosis. Lacking a shared diagnosis, they cannot craft a
common vision of the future state or a coherent intervention strategy
that successfully negotiates the difficult problems organizational
change poses. In short, the low level of competence in most organiza-
tions in fashioning an inquiring dialogue inhibits identifying root
causes and developing fundamental systemic solutions.47

Our data indicate that the quality of practitioners’ sense-making determines
the quality of the actions that they take based on the data. This finding offers
insight into why the survey measure – teacher satisfaction with Benchmarks –
was not a predictor of gains in student achievement. If practitioners focus only
on superficial problems – described as “the low-hanging fruit” by principals in
our study – their intervention strategies are likely to be mundane.48

46
Spillane, J. P., Reiser, B. J., & Reimer, T. (2002). Policy implementation and cognition:
Reframing and refocusing implementation research. Review of Educational Research, 72(3), 387-
431.
47
Beer, M. & Eisenstat, R. A. (1996). Developing an organization capable of implementing strat-
egy and learning. Human Relations, 49(5), 597-619, p. 599-600.
48
Sarason, S. B. (1982). The culture of the school and the problem of change. Boston: Allyn &
Bacon, Inc.
56

Chapter Five:
Making the Most of Benchmark Data:
The Case of Mahoney Elementary School
In this chapter, we use our qualitative data to examine how the multiple fac-
tors, that were so difficult to disentangle quantitatively, interact within a
The Benchmarks were a

school context. While research has emphasized that school leaders are in a
powerful vehicle for
position to encourage and support school staff to use data to transform prac-
tice,49 there remains much to be done in offering detailed examinations of
reinforcing the use of

school leaders’ work in this area.50 Spillane and his colleagues distinguish
the curriculum, for

between “macro functions” (e.g., encouraging data-driven decision-making) focusing teachers’
and “micro tasks” (e.g., displaying the data, formulating substantive and
provocative questions about the data). They urge researchers to analyze how
attention on the

educators “define, present, and carry out these micro tasks” and how the
standards, and for
micro-actions interact with one another and with other contextual factors.51
Our goal was to understand how school leaders build the strong feedback
organizing conversations

systems that we discussed in Chapter One.
about student achieve-
ment in which teachers
Below, we focus on the Mahoney Elementary School, briefly described in
53

Example 4 of Chapter Four. Here, we look in more detail at how school lead-
were expected to talk

ers – particularly the principal and subject area teacher leaders – established
about ways to improve
strong processes for collective learning from Benchmark data within and
across instructional communities at Mahoney.52 For Mahoney, the
their teaching.

Benchmarks were a powerful vehicle for reinforcing the use of the curriculum,
for focusing teachers’ attention on the standards, and for organizing conversa-
tions about student achievement in which teachers were expected to talk
about ways to improve their teaching. In effect, these school-based discus-
sions around the Benchmark assessments helped nurture the “instructional
coherence” cited in Chapter Two and identified by the Consortium for Chicago
School Research (CCSR) as showing a positive impact on student learnings.54

49
Choppin, J. (2002, April 2). Data use in practice: Examples from the school level. Paper pre-
sented at the Annual Meeting of the American Educational Research Association, New Orleans,
LA. ; Wohlsetter, P., Datnow, A., & Park, V. (2007, April). Creating a system for data-driven
decision-making: Applying the principal - agent framework. Paper presented at the Annual
Meeting of the American Educational Research Association, Chicago, IL.
50
Spillane, J. P., Halverson, R. R., & Diamond, J. B. (2001, April). Investigating school leader-
ship practice: A distributed perspective. Educational Researcher, 30(3), 23-28.
51
Spillane, J.P. et al., 2001, p. 24.
52
Brown, J. S. & Duguid, P. (2000). Organizational learning and communities of practice:
Toward a unified vision of working, learning, and innovation. In Lesser, E. L., Fontaine, M., and
Slusher, J. A., Knowledge and communities (99-121). Boston: Butterworth Heinemann.; Wenger,
E., McDermott, R., & Snyder, W. M. (2002). Cultivating communities of practice. Boston:
Harvard Business School Press.
53
Pseudonyms are used in this case study for the school and its principal.
54
Newmann, F. M. et al., 2001.
57

Table 5.1 Interviews and Observations Conducted at Mahoney
Elementary School 2005-06 through 2006-07

Researchers conducted intensive fieldwork at Mahoney Elementary School in 2005-06
and 2006-07. During that time, we conducted a total of six observations of leadership
team meetings, grade group meetings, CSAP meetings and a school-wide professional
development session. We interviewed a total of 11 school staff including the principal,
math and literacy leaders, a school secretary and classroom teachers. We interviewed
Table ? School Based Interviews and Observations
some individuals multiple times.

Staff Position 2005-06 Interviews 2006-07 Interviews

Principal 2 2

Math leader 2 2

Literacy leader 1 2

Third grade teacher 1

Fourth grade teacher A 1 1

Fourth grade teacher B 1

Fifth grade teacher A 1

Fifth grade teacher B 1

Fifth grade teacher C 1

Sixth grade teacher 1

Secretary 1

Setting 2005-06 Observations 2006-07 Observations

Leadership Team 1

Grade Group 1 1

Comprehensive Student 1 1
Assistance Process

Professional Development 1

58

Figure 5.1 Feedback Loop for Engaging with Data

Accessing
and organizing
data

Sense-
Feedback making
Assessing Systems
and modifying to identify
in Instructional problems and
solutions Community solutions

Trying
solutions

School Leaders and Effective Feedback Systems

At Mahoney, the principal, Ms. Bannon, established high expectations and a
high level of structure to classroom instruction. She participated actively in
the school’s weekly grade group meetings and worked closely with teacher
leaders and classroom teachers to improve instruction. Her high expectations
for teachers and students created discomfort for some staff members; however,
her commitment to children was respected. Ms. Bannon and the math and lit-
eracy teacher leaders orchestrated grade group discussions of Benchmark and
other assessment data that built a shared set of goals for teaching and learn-
ing and provided an ongoing context for professional learning.

Mahoney’s teacher leaders were both fully released from regular classroom
instruction. Not only did they work with Ms. Bannon to identify short-term
interventions based on Benchmark data at meetings together, they also col-
laborated with the principal on developing long-term strategies for meeting
the school’s goals. The principal explained why she had prioritized putting
limited resources into full-time teacher leaders when she became the princi-
pal a few years before our study began,
“It was a hard decision since it meant larger class sizes. But I wanted
to begin with a strong leadership team. It’s a choice between having a
great teacher reach 25 students or having a great teacher reach other
teachers”(2007).

59

The multiple contributions of the teacher leaders at Mahoney were apparent
in both interviews and observations. For example, in our complete fieldnotes
“[Allocating the

for the grade group meeting briefly in Example 4 in Chapter Four, the math
resources for full-time
teacher leader: content area teacher
leaders] was a hard
• pointed out that using calculators would improve student scores on a significant
number of Benchmark and PSSA (state-wide accountability test) questions;
decision since it meant
• offered to conduct a workshop for teachers about how to use their classroom larger class sizes. But I
sets of calculators as part of the upcoming professional development day;
wanted to begin with a
• explained that “matrix multiplication” showed up on the Benchmarks, but was
a technique that is specific to a particular curriculum and wouldn’t strong leadership team.
be on the PSSA; and It’s a choice between
• provided strategies for teaching the mathematical concept of “expanded nota-
tion” and offered to come into the 4th grade classrooms and to model lessons
having a great teacher
on expanded mathematical notation for small groups of students. reach 25 students or

At this meeting the math teacher leader used her knowledge of the Core
having a great teacher
Curriculum, the Benchmark assessments, and the state’s accountability reach other teachers.”
assessment to help teachers set instructional priorities. She offered sugges-
tions about instructional materials (e.g., calculators). She pointed out the - Mahoney Principal
kinds of professional development that the school ought to offer. Perhaps, most
importantly, she established why it was important that teachers open their
classroom doors and allow her to provide support and guidance through
demonstration lessons. Many teachers interviewed, especially in the lower
grades, articulated the value of the teacher leaders’ ongoing support. One said,
“Knowing that my literacy leader is there [is important], and if I say to her,
‘You know, I’m not really sure how I’m going to do this lesson,’ she’s always
there and very helpful.” (2006).

In Chapter One, we posited a four-step feedback cycle as a central element
within a school’s overall capacity for data-driven organizational learning and
student achievement gains. These steps included school leaders and teachers:

1 Accessing and organizing data about students’ understanding of the Core
Curriculum (the Benchmark assessments);
2 Making sense of the data – both individually and collectively (grade group
meetings) – to identify problems and potential solutions;
3 Trying the solutions back in their classrooms; and
4 Assessing and modifying their solutions based on classroom assessments.

As discussed in Chapter Two, the school district intended for the Benchmark
assessments to provide the kind of formative feedback that allows teachers
to make mid-course corrections in their instructional strategies. Teacher

60

leaders at Mahoney were critical to the school’s success in implementing
systems and an organizational culture that enabled these kinds of feedback
“Knowing that my

systems across the school. In any cycle, the “linkages” that connect the steps
literacy leader is there
are crucial and are often the weak points in a system (See Figure 5.1). [is important], and if
Teacher leaders helped support those links, and in many cases served as
links themselves, sharing knowledge from grade group meetings across the
I say to her, ‘You know,

school. I’m not really sure how

Additionally, review of Benchmark data at Mahoney was integrated into the
I’m going to do this

kinds of feedback systems discussed in Chapter One. Teachers experimented
lesson,’ she’s always
with new practices that had been identified in grade group meetings. School there and very helpful.”
leaders followed up in classrooms to help teachers with new instructional
strategies and to modify these practices where appropriate. These steps - Mahoney Teacher
became routine at Mahoney, thus ensuring that feedback systems were
strong and coherent during the period of our research.

Grade Group Meetings and Benchmark Discussions

Grade group meetings were a key opportunity for looking at and learning
from Benchmark data at Mahoney. These meetings were held weekly and
included the principal, the math teacher leader, the literacy teacher leader,
and the two or three classroom teachers for each grade. Grade group meet-
ings were described by the principal and teacher leaders as the most impor-
tant site in the school for teacher learning. In fact, during the second year of
our research, Ms. Bannon reported that they had decided to call the meet-
ings “Professional Learning Communities” instead of grade groups, to high-
light their contribution to teachers’ professional learning.

Grade group meetings at Mahoney were highly structured and consistently
focused on instructional issues. Each meeting began with a member of the lead-
ership team handing out a typed agenda with a guiding question at the top,
ended with the principal summarizing next steps, and was followed up with
typed notes distributed to all participants. According to teachers and school lead-
ers, grade group meetings always focused on analysis of data or reflection on
instruction. As one teacher told us, “Everything begins by talking about data.”

The Benchmark Item Analysis Reports were important tools in grade group
meetings, as they were in other schools. At Mahoney, however, the Core
Curriculum Standards was another key tool in grade group meetings.
Teachers were expected to bring the curriculum framework to grade group
meetings so they could refer to it as they discussed the standards in which
their students showed weaknesses. In addition, teachers were expected to
prepare for grade group meetings by filling out the district’s Benchmark
61

Data Analysis Protocol, which asked them to assess students’ weaknesses
and identify strategies for improving the areas of weakness. They used these
[The principal] com-

protocols in conversations with their colleagues. The structure of the meetings
mented that the cut-off
themselves supported the continuity of the feedback system. Use of the same for-
mats and reports created a common framework and language. Clear follow-up
points for identifying

about next steps ensured that the momentum of the meeting was not lost.
individual students as
Advanced and
The heart of the grade group meetings was the discussion of Benchmark and
other assessment data. As in other schools, Mahoney’s grade group discus-
Proficient were too

sions of Benchmarks encompassed what we identified earlier in Chapter
low, saying that “we
Four as three interconnected types of sense-making: strategic (e.g., short-
term tactics to help the school reach AYP), affective (teachers’ beliefs about
have to set our own

their students and their collective responsibility for student learning), and
goal as higher than

reflective (evaluating their own instructional practices and connecting that.”
Benchmark data to with key curriculum concepts).

Analysis and discussion of Benchmark data not only focused on instruction,
but also highlighted the interim assessments’ connection to other accounta-
bility tests, an example of strategic sense making. Teachers and leaders dis-
cussed how many and which students were close to Proficient or Advanced –
performance categories on the PSSA. Talk about Benchmarks and the PSSA
also led to talk about the school’s moral purpose and the leaders’ belief in the
capabilities of their staff and students. In one grade group meeting, Ms.
Bannon commented that the cut-off points for identifying individual students
as Advanced and Proficient were too low, saying that “we have to set our
own goal as higher than that”(2005). The expectation that all students would
be Proficient was accompanied by a consistent focus in grade group meetings
on the Core Curriculum, the standards, and what teachers could do to
improve their own teaching. As one teacher said:
The school has been focused on using the data to help the kids and
push the instruction. Every kind of thing that we do, every assess-
ment we give, we look at it; we see what we need to change, and how
we can differentiate our instruction so that it’s helping them do more.
(2006)

Teachers at Mahoney were pushed to question their own past practices and
they both sought and shared new ways to approach content that needed to be
taught and new ways to help their students learn. The re-naming of the grade
group meetings as “Professional Learning Communities” was appropriate.

62

Organizational Learning and Instructional Coherence

In summary, the principal and teachers leaders at Mahoney had a clear Alongside principals,
understanding of the powerful connection between the Benchmarks and the
Core Curriculum and their importance to establishing instructional coher-
teacher leaders can

ence across the school. The principal allocated resources for knowledgeable assume important
teacher leaders who were expert in the content and assessment issues in
their own curricular areas. Together, the principal and teacher leaders
leadership functions

established a set of structures and practices that ensured that Benchmark
relative to data use.
data were used as part of a process for ensuring high quality instruction
within and across grade groups, as well as other settings in the school. At
Mahoney, the principal and the teacher leaders were “learning leaders,” who
created a climate in which adult learning was central to school
improvement.55 They took the lead in helping teachers sift through reams of
data and make sense of competing priorities. Leadership around the use
Benchmark data was distributed across the roles of principal and teacher
leaders.56 Alongside principals, teacher leaders can assume important leader-
ship functions relative to data use.

Elmore, R. F. (2000, December). Building a new structure for school leadership. Washington,
55

DC: The Albert Shanker Institute.; DuFour, R. (2002, May). The learning-centered principal.
Educational Leadership, 59(8), 12-15.; Spiri, M. H. (2001, May). School leadership and reform:
Case studies of Philadelphia principals. Philadelphia, PA: Consortium for Policy Research in
Education.
56
Spillane, J.P. et al., 2001.

63

Making the Most of Benchmark Data at Mahoney Elementary School
Engaged Principal:

• Built strong leadership team by allocating full time teacher leaders in math and reading

• Worked with teacher leaders to develop long-term instructional improvement strategies and
shorter-term priorities for their work with classroom teachers

• Emphasized data-driven decision-making

• Actively attended grade group meetings

• Established meeting routines that were used across the school

• Set high expectations for teachers’ preparation for and participation in grade group meetings

• Used discussions of Benchmark data in grade groups to reinforce importance of proficiency
standards of Core Curriculum

• Encouraged strategic, affective, and reflective sense-making, with the strongest emphasis on
reflective sense-making

• Worked with teacher leaders to spread insights and knowledge about instruction across the school

Full-time Math and Reading Teacher Leaders:

• Well-versed in the Core Curriculum, the Benchmark assessments, and the PSSA exams and
understood the connections and disconnections among the three

• Continuously enhanced their knowledge of research-based instructional strategies that
supported effective use of the Core Curriculum

• Helped teachers interpret Benchmark data

• Recommended specific instructional strategies based on the Benchmark results

• Moved in and out of classrooms to see if teachers were implementing curriculum well and
provided coaching and demonstration where needed

• Gathered resources to supplement the curriculum

• Collaborated with principal on long and shorter-term instructional strategies to meet school's
goals

Effective Grade Group Meetings:

• Held weekly
• Principal, teacher leaders, and classroom teachers came prepared to participate
• Discussions included strategic, affective, and reflective sense-making
• Highly structured meeting routines, focused on instructional issues and ongoing professional
learning of staff
• Began with an agenda and guiding question
• Ended with school leader summarizing next steps
• Follow-up notes distributed across the school

64

Conclusion
Making the Most of Interim Assessment Data:
Implications for Philadelphia and Beyond
Federal, state, and district policies that use standardized tests as the central
metric for accountability have fueled the fervor for student achievement data,
Data can make

especially in districts with large numbers of academically failing students. The
problems more
rise of interim assessments is inextricably tied to the policy environment of No visible, but only
Child Left Behind. Controversy notwithstanding, the use of interim assess-
ments by large urban school districts to improve instruction and student
people can solve

achievement is on the rise. The findings from our research on the use and
them.
impact of these assessments in Philadelphia’s K-8 schools will not end the
debate. They do, however, offer formative lessons to Philadelphia and beyond
about the design, implementation, and impact of interim assessments. Below,
we discuss the implications of this research for policy makers and district and
school leaders. The research also has important implications for the higher
education community that educates and certifies district and school leaders.

Investing in School Leaders

The most important message from this research is that the success of even a
well-designed system of interim assessments is dependent on the knowledge
and skills of the school leaders and teachers who are responsible for bringing
the system to life in schools. Stringent accountability measures, strong cur-
ricular guidance, and periodic assessments are not substitutes for skilled
and knowledgeable practitioners. Data can make problems more visible, but
only people can solve them.

In addition, mandated accountability measures, in and of themselves, are an
inadequate foundation for building the kinds of collegial relationships that result
in shared responsibility for school improvement and improved student learning.

In Philadelphia, the very federal and state policies that persuaded district lead-
ers and school practitioners to pay careful attention to data, also constrained
their ability to make the most of Benchmark results for improving instruction
and student achievement. Immediate needs for improved testing outcomes
often worked against practitioners learning more about how to help all students
master the concepts and skills of the Core Curriculum.

However, our research also indicates that the use of Benchmark data is not
always a narrow exercise in preparing to “teach to the test.” We witnessed how
school leaders were able to mediate the often counter-productive environment of
high stakes accountability. In the language of organizational learning, these
leaders enacted organizational practices that contributed to individual teacher
learning and professional growth, while at the same time fortifying a collective
understanding of the challenges, goals, and path ahead for the school. 65

Data-driven decision-making represents a new way of thinking for most educa-
tors. And, as this report has demonstrated, the logic of data use is built on
School leaders need

numerous assumptions that cannot be taken for granted, especially the ability
to be able to lead the
of school leaders to help teachers make the most of Benchmark results.
Organizational learning offers a robust framework for understanding what
kinds of deliberative

school leaders need to know and be able to do in order to make the most of
conversations that

interim assessment results and other kinds of data about student achievement. create opportunities
for teacher learning.
• As learning leaders, principals and teacher leaders need to know how to facilitate
“learning” discussions about data. School leaders can make a real difference in
helping staff move beyond data use as a narrow exercise in preparing to
“teach to the test.” But to do so, they must know how to frame conversations
about assessment data so that teachers understand the connections to larger
school improvement priorities and to the curriculum. They need to know how
to pose questions that invite teachers to talk openly about: curriculum con-
cepts, how their students learn best, what instructional practices have worked
and those that haven’t, what additional curricular resources they need, what
they need to learn about content, and where they might seek evidence-based
instructional strategies that would address the learning weaknesses of their
students. They also need to be able to steer teachers away from inappropriate
use of Benchmark data, such as predicting performance on the PSSA. School
leaders need opportunities to practice these skills and receive feedback.
Understanding the value and purposes of the different types of sense-making
identified in our research – affective, strategic, and reflective – and how to
use them offer a framework for such training.

• As learning leaders, principals and teacher leaders need to know how to allocate
resources and establish school organizational structures and routines that support
the work of instructional communities and assure that the use of Benchmark data
is embedded in the feedback systems necessary for organizational learning.
School schedules need to accommodate regular meetings of grade groups.
Principals and teacher leaders need to be at these meetings and, with teach-
ers, establish meeting routines that include agendas, discussion protocols
with guiding questions, and documentation of proceedings. Follow up to the
meetings is crucial. School leaders need to visit classrooms to see if and
how teachers are using instructional strategies and to offer resources and
coaching so that teachers can deepen their understanding of curriculum con-
tent and pedagogy. Assessing the impact of interventions is also crucial.
Important steps include helping teachers to design classroom based assess-
ments for use during the sixth week of instruction and examining the quality
of common interventions such as tutoring and after-school remediation pro-
grams. School leaders must recognize their role in the creation and diffusion
of knowledge across the school.

66

Designing Interim Assessments and Supports for Their Use

This research also offers lessons about designing interim assessments and
the resources that will encourage and support the use of data from those
assessments. Philadelphia’s Benchmark assessments have a number of clear
design strengths that may offer guidance to other districts considering adop-
tion of interim assessments. The alignment of the Benchmarks with the Core
Curriculum reinforced expectations for what teachers should teach and at
what pace; it made the Benchmark results highly relevant to teachers’
instructional planning. The timely return of the results and the allocation of
a sixth week for re-teaching after review of the data buttressed the instruc-
tional intention of the Benchmarks. District supports in the form of technolo-
gy, tools for data analysis and interpretation, and professional development
were largely appreciated by school staff. All of these elements likely con-
tributed to broad acceptance and use of the Core Curriculum and
Benchmark assessments by Philadelphia K-8 teachers.

• As districts and schools develop organizational structures, processes and tools to
support the use of interim assessment data, they need to ask themselves these
questions:
Do the structures, processes, and tools support the review of data as a collec-
tive learning activity of instructional communities? Are they supporting the
review of data as an activity which helps teachers deepen their pedagogical con-
tent knowledge and understand what their students know and how they learn?
Do they support the multiple steps of feedback loops? Do they encourage
leaders’ follow-up work with teachers in classroom? Do they promote the
assessment of interventions and modifications where necessary?
• In Philadelphia, district leaders should revisit their purposes for the Benchmark
assessments with the goal of prioritizing one or two purposes. To achieve the
instructional purposes that district leaders intended, it is likely that the
Benchmark assessments are in need of modifications.
In order to capitalize on Benchmarks to fulfill instructional purposes, the
district leaders should: review Benchmark items to make certain that they:
test for a range of thinking skills – knowledge, comprehension, application,
synthesis and evaluation – and that they offer distractor answers that provide
insight into what students don’t understand. Continued efforts should be
made by the district and testing industry to include open-ended items.

67

Implications for Future Research

We believe that the use of a multi-method design and organizational learn-
ing as an analytic framework were two strengths of this study. Used in con-
cert, they offer considerable promise in unraveling the connections among
many factors related to the use of data in schools and gains in student
achievement. There are numerous refinements to our approach that
researchers might make that would make significant contributions to both
theory and practice. These include more direct survey measures of data use
and analyses at the classroom and instructional community levels.

We also realize that we only scratched the surface in terms of the three
kinds of sense-making and the relationships between the kinds of sense-
making and the resulting instructional plans. We suggest that discourse
analysis offers a robust methodology for research on data use and instruc-
tional improvement.

One of the controversies surrounding interim assessments is whether they
actually serve formative purposes for teachers and students. While we, as
well as other researchers, have begun to build a knowledge base about the
impact of interim assessments on teachers’ instructional practice, there
remains much work to do on whether interim assessment results help stu-
dents understand their mistakes and make appropriate adjustments in their
thinking.

68

Reference List

A curriculum audit of the Philadelphia public schools, Philadelphia, PA.
International Curriculum Management Audit Center. Phi Delta Kappa,
International. May 16-21, 2005.

Argyris, C. & Schon, D. A. (1978). Organizational learning: A theory of action perspec-
tive. Reading, MA: Addison-Wesley.

Beer, M. & Eisenstat, R. A. (1996). Developing an organization capable of implement-
ing strategy and learning. Human Relations, 49(5), 597-619, p. 599-600.

Black, P. & Wiliam, D. (1998, October). Inside the black box: Raising standards
through classroom assessment. Phi Delta Kappan.

Blume, H. (2009, January 28). L.A. teachers' union calls for boycott of testing. Los
Angeles Times [On-line]. Retrieved on February 11, 2009 from
http://www.latimes.com/news/education/la-me-lausd28-2009jan28,0,4533508.story.

Brown, J. S. & Duguid, P. (1998). Organizing knowledge. California Management
Review, 40(3), 28-44, p. 28.

Brown, J. S. & Duguid, P. (2000). Organizational learning and communities of prac-
tice: Toward a unified vision of working, learning, and innovation. In Lesser, E. L.,
Fontaine, M., and Slusher, J. A., Knowledge and Communities (99-121). Butterworth
Heinemann.

Bulkley, K. E., Mundell, L., & Riffer, M. (2004). Contracting out schools: The first
year of the Philadelphia Diverse Provider Model. Philadelphia: Research for Action.

Burch, P. (2005, December 15). The new education privatization: Educational con-
tracting and high stakes accountability. Teachers College Record.

Cech, S. J. (2008, September 17). Test industry split over ‘formative’ assessments.
Education Week, 28(4), 1, 15, p. 1.

Century, J. R. (2000). Capacity. In N. L. Webb, J. R. Century, N. Davila, D. Heck, &
E. Osthoff (Eds.), Evaluation of systemic change in mathematics and science educa-
tion. Unpublished manuscript, University of Wisconsin-Madison, Wisconsin Center
for Education Research.

Choppin, J. (2002, April 2). Data use in practice: Examples from the school level.
Paper presented at the Annual Meeting of the American Educational Research
Association, New Orleans, LA.

Clune, W. H. & White, P. A. (2008, October). Policy effectiveness of interim assess-
ments in Providence Public Schools. WCER Working Paper No. 2008-10. Wisconsin
Center for Education Research, School of Education, University of Wisconsin-
Madison http://www.wcer.wisc.edu/. p. 5.

Corcoran, T. B. & Christman, J. B. (2002, November). The limits and contradictions
of systemic reform: The Philadelphia story. Philadelphia: Consortium for Policy
Research in Education.

Daft, R. L. & Weick, K. E. (1984). Towards a model of organizations as interpretation
systems. Academy of Management Review, 9(2), 284-295.

DuFour, R. (2002, May). The learning-centered principal. Educational Leadership,
59(8), 12-15.

69

Elmore, R. F. (2000, December). Building a new structure for school leadership.
Washington, DC: The Albert Shanker Institute.

Halverson, R. R., Prichett, R. B., & Watson, J. G. (2007). Formative feedback systems
and the new instructional leadership (WCER Working Paper No. 2007-3). [On-line].
Retrieved on July 16, 2007, from
http://www.wcer.wisc.edu/publications/workingPapers/index.php.

Knapp, M. S. (1997). Between systemic reforms and the mathematics and science
classroom: The dynamics of innovation, implementation, and professional learning.
Review of Educational Research, 67(2), 227-266.

Leithwood, K., Aitken, R., & Jantzi, D. (2001). Making schools smarter: A system for
monitoring school and district progress. Thousand Oaks, CA: Corwin Press.

Lipsey, M.W., and D.B. Wilson (1993). The efficacy of psychological, educational, and
behavioral treatment: Confirmation from meta-analysis. American Psychologist, 48,
1181-1209.

Little, J. W. (1999). Teachers' professional development in the context of high school
reform: Findings from a three-year study of restructuring high schools. Paper pre-
sented at the Annual Meeting of the American Educational Research Association,
Montreal, Quebec.

Mason, S. A. & Watson, J. G. (2003). Understanding schools' capacity to use data.
Paper presented at the Annual Meeting of the American Educational Research
Association, Chicago, IL.

Mintz, E., Fiarman, S. E., & Buffett, T. (2006). Digging into data. In K. P. Boudett, E.
A. City, & R. J. Murname (Eds.), Data wise: A step-by-step guide to using assessment
results to improve teaching and learning (81-96). Cambridge, MA: Harvard
Education Press, p. 94.

Newmann, F. M., Smith, B., Allensworth, E., & Bryk, A. S. (2001, January).
Improving Chicago's schools: School instructional program coherence benefits and
challenges. Chicago: Consortium on Chicago School Research.

Newmann, F. M., Smith, B., Allensworth, E., & Bryk, A. S. (2001). Instructional pro-
gram coherence: What it is and why it should guide school improvement policy.
Educational Evaluation and Policy Analysis, 23, 297-321.

Perie, M., Marion, S., Gong, B., & Wurtzel, J. (2007, November). The role of interim
assessments in a comprehensive assessment system. Washington, DC: The Aspen
Institute.

Porter, A. C., Chester, M. D., & Schlesinger, M. D. (2004, June). Framework for an
effective assessment and accountability program: The Philadelphia example.
Teachers College Record, 106(6), 1358-1400.

Resnick, L. B. & Hall, M. W. (1998). Learning organizations for sustainable education
reform. Journal of the American Academy of Arts and Sciences, 127(4), 89-118, p. 108.

Rusch, E. A. (2005). Institutional barriers to organizational learning in school sys-
tems: The power of silence. Educational Administration Quarterly, 41, 83 – 120. [On-
line]. Retrieved on May 8, 2007, from SAGE Full-Text Collections.

Sarason, S. B. (1982). The culture of the school and the problem of change. Boston:
Allyn & Bacon, Inc.

70

Senge, P. (1990). The fifth discipline: The art & practice of the learning organization.
NY: Doubleday.

Shulman, L. S. (1987). Knowledge and teaching: Foundations of the new reform.
Harvard Educational Review, 57(1), 1-22.

Spillane, J. P., Halverson, R. R., & Diamond, J. B. (2001, April). Investigating school
leadership practice: A distributed perspective. Educational Researcher, 30(3), 23-28.

Spillane, J. P., Reiser, B. J., & Reimer, T. (2002). Policy implementation and cogni-
tion: Reframing and refocusing implementation research. Review of Educational
Research, 72(3), 387-431.

Spillane, J. P. & Thompson, C. L. (1997, June). Reconstructing conceptions of local
capacity: The local education agency's capacity for ambitious instructional reform.
Education Evaluation and Policy Analysis, 19(2), 185-203.

Spillane, J. P. & Zeuli, J. S. (1999). Reform and teaching: Exploring patterns of prac-
tice in the context of national and state mathematics reforms. Educational
Evaluation and Policy Analysis, 21(1), 1-27.

Spiri, M. H. (2001, May). School leadership and reform: Case studies of Philadelphia
principals. Philadelphia, PA: Consortium for Policy Research in Education.

Travers, E. (2003, September). Philadelphia school reform: Historical roots and reflec-
tions on the 2002-2003 school year under state takeover. Penn GSE Perspectives on
Urban Education, 2(2).

Useem, E. (2005, August). Learning from Philadelphia's school reform: What do the
research findings show so far? Paper presented at the No Child Left Behind
Conference, Sociology of Education Section of the American Sociological Association,
Philadelphia, PA.

Wagner, T. (1998). Change as collaborative inquiry: A 'constructivist' methodology for
reinventing schools. Phi Delta Kappan, 80(7), 378-383.

Wenger, E., McDermott, R., & Snyder, W. M. (2002). Cultivating communities of prac-
tice. Boston: Harvard Business School Press.

Wohlsetter, P., Datnow, A., & Park, V. (2007, April). Creating a system for data-driv-
en decision-making: Applying the principal-agent framework. Paper presented at the
Annual Meeting of the American Educational Research Association, Chicago, IL.

71

Appendix A Phase One Qualitative Research — School Characteristics
2006-07 Data

Achievement
% from % Advanced
Number of Low-Income & Proficient
School Provider Grades Students Families Racial/Ethnic Make-up Reading/Math
School A* University of K-8 425 85.7 91.1% African American 5th Grade
Pennsylvania .05% White 27.3/42.0
5.4% Asian
1.4% Latino 8th Grade
1.6% Other 42.6/52.7
School B Edison K-5 465 80.8 97.8% African American 5th Grade
Schools, Inc. 1.3% White 22.4/46.9
0.6% Latino
0.2% Other
School C Victory K-6 390 86.5 97.9% African American 5th Grade
0.5% Latino
0.5% Other
School D Office of K-8 399 86.9 23.1% African American 5th Grade
Restructured 1.0% White 28.6/60.0
Schools 75.4% Latino
8th Grade
0.5% Other 68.1/44.7
School E Universal K-6 193 85.7 93.3% African American 5th Grade
Companies 1.6% White 70.0/75.0
4.7% Latino
0.5% Asian
School F* Office of K-6 412 90.4 98.1% African American 5th Grade
Restructured 0.2% White 49.2/77.1
Schools 1.7% Latino
School G* “Sweet 16” K-8 635 85.4 83.9% African American 5th Grade
0.5% White 8.7/30.4
8.2% Latino
7.2% Asian 8th Grade
0.2% Other 43.8/35.3

School H* Foundations, K-6 391 90.0 95.4% African American 5th Grade
Inc. 2.0% White 4.3/27.7
2.3% Latino
0.3% Other
School I* Edison K-8 311 86.3 59.8% African American 5th Grade
36.0% Latino
2.9% Other 8th Grade
27.3/39.4
School J Temple K-8 463 91.7 99.4% African American 5th Grade
University 0.6% Latino 29.5/37.8

8th Grade
41.7/36.2

72 * Case Study schools 2006-2007.

Appendix B Benchmark Item Analysis Form

73

Appendix C List of Topics Covered in Interviews

The following are lists of topics covered in interviews with principals, teacher leaders and
classroom teachers. Each round of interviews (Fall 2005, Spring 2006, Fall 2006 and
Spring 2007) covered a different, though sometimes overlapping, set of topics.

2005-06 Interview Topics

• School context
School’s history with reform
Current reform initiatives
Principal’s leadership style
• Changes in and rationale for instructional priorities
Identify and explain classroom changes and previous practices
Staff and other influences that led to instructional changes
Resources necessary for instructional changes
• Leadership team and other instructional communities (grade groups, SLCs)
Composition of the leadership team and instructional communities
Members’ roles, settings for meetings
Relationships with the provider
Examples of instructional decisions and use of data
• Roles and responsibilities around data
Principal’s and leadership team’s role in using data
Provider’s role and expectations
Responsibilities around organizing and analyzing the data
• Benchmarks and other formative assessment
Importance and use of formative assessments
Provider and others’ role in using formative assessments
• Professional development about data
Settings and topics of professional development sessions
• Staff capacity for data
Examples of sophisticated and unsophisticated data use
• Resources necessary to use data effectively
Technology
Human support
• Professional development around data use
• Data analysis tools
Identify and describe data analysis tools
People and processes involved in implementing the tools
• Useful/helpful data
Data used to inform classroom instruction or identify broad problems
How were benchmarks used?
Useful tools and formats for data analysis
• Settings for discussions and analysis of data

74

2006-07 Interview Topics

• Context surrounding school leadership
Leadership styles and influences on classroom instruction
Leadership actions that have influenced instruction
Background and self-assessment of effectiveness in school role
Sources of support and guidance for teachers and leaders
Thoughts on leading in a high stakes environment
Role of formal and informal teacher leadership
• School Improvement Planning (SIP)
Progress on improvement goals and future priorities
Process for planning the goals and priorities
• Instructional changes
Changes that school leaders have encouraged and the role of data in promot
ing those changes
Instructional communities and grade groups
Structure and roles of the groups
Groups’ roles in encouraging and guiding teachers,
Challenges the groups face
• Data use
Instructional changes made because of data
Data that teachers have used and found helpful
Settings for examining data
Tools teachers used to examine data
Benchmarks and PSSA writing rubric
Where and when do teachers use these tools?
What do they learn from each kind of assessment?
• Professional development
Types of professional development
Impact of the professional development
School leaders’ roles in professional development sessions
• Impact of high stakes accountability environment
Guidance and support from colleagues and leaders

75

APPENDIX D Technical Details on Data and Methods

Survey Data

The teacher survey was distributed through the schools, and completed sur-
veys were collected and returned by the schools to the district’s research office.
The survey did not ask teachers to provide their names or other information
that could identify them as individuals. Still, some teachers, especially those
who work in schools where social trust is low, are wary of completing surveys.
It is also notoriously difficult to compel a busy teacher to complete a long sur-
vey, which, in this case, involved hundreds of questions spread over 16 pages.
Given these challenges, the response rates for the surveys are respectable. A
total of 6,680 teachers (65 percent of all teachers) from 204 of 280 schools
responded to the spring 2006 survey. A total of 6,007 teachers (60 percent of
all teachers) responded to the spring 2007 survey. These response rates are
comparable to that for large-scale teacher surveys in other major cities; for
example, teacher surveys fielded by the Consortium on Chicago School
Research typically produce a response rate of about 60 percent.

To make the school-level predictor variables used in the multilevel models,
data from all teachers who responded to the survey (not just teachers in
Benchmarks grades and subjects) was aggregated. Schools at which fewer
than 30 percent of the teachers responded were excluded from the analysis.

Assessment of Student Learning: The Rank-Based Z-Score Method

During the school years 2004-2005, 2005-2006, and 2006-2007, Philadelphia
students in grades three through eight took standardized tests of achieve-
ment in reading and mathematics at the end of the school year. However, in
some grades, students took the Terra Nova test, a commercially available
assessment developed by CTB McGraw Hill. In other grades, students took
an assessment developed by Commonwealth of Pennsylvania (PSSA). The
different assessments taken in different years necessitate a special strategy
to examine learning gains.

To create a comparable indicator of achievement, we placed student scores
on the rank-based z-score scale. The rank-based z-score converts a student’s
percentile (in the Philadelphia distribution of scores) to their position in the
normal distribution, so a student at the 50th percentile would have a rank-
based z-score of 0, while one at the 95th percentile would have a rank-based
z-score of 1.64, and one at the 5th percentile would have a score of -1.64. The
indicator of learning growth was created by subtracting the z-score at the
end of Year 1 from the z-score at the end of Year 2.

76

This method is the same used by RAND in its recent reports on the impact
on student achievement of privatization of schools in Philadelphia (Gill,
Zimmer, Christman, & Blanc, 2007) and on Philadelphia’s charter schools
(Zimmer, Blanc, Gill, & Christman, 2008).

Technical Description of the Multilevel Models

The dependent variable was the student’s rank-based z-score on
reading comprehension or mathematics at Time 2 (that is, either the score
from spring 2006 or spring 2007). The equations are as follows:

Level 1

Yij = 0j + 1j(Race/Ethnicity) ij + 2j(Gender)ij + 3j(Special
Education)ij + 4j(Grade at Test 1)ij + 5j(Grade at Test 2)ij

+ 6j(Rank-based z-score on Test at Time 1)ij + rij

Level 2

0j = 00 + 01(Percent Low Income)j + 02(Additional School-Level
Variables)j + u0j

All predictor variables were grand-mean centered.

77

Appendix E Technical Detail on Scales Used in Chapter 3

The first four scales presented here – Instructional Leadership, Teacher-
Teacher Trust, Instructional Innovation and Improvement, and Teacher
Collective Responsibility – incorporate most of the specific items that make
up the indicators with those names developed by the Consortium on Chicago
School Research (CCSR). Information on the CCSR scales can be accessed at
http://ccsr.uchicago.edu/content/page.php?cat=4. The specific items that comprise
the scales used in this chapter are shown below. Likewise, the values for
Cronbach’s alpha were created for these scales from the Philadelphia teacher
survey data.

(Eight items; Cronbach’s alpha: .94)

To what extent do you disagree or agree with the following statements?
(Response categories: Strongly Disagree, Disagree, Agree, Strongly Agree)

The leadership at this school:
• Makes clear to the staff the expectations for meeting instructional goals.
• Communicates a clear vision for our school.
• Sets high standards for student learning.
• Carefully tracks student academic progress.
• Encourages teachers to implement what they have learned
in professional development.
• Knows what’s going on in my classroom.
• Actively monitors the quality of teaching in this school.
• Has made data-driven decision-making a priority at the school.

Teacher Commitment to the School
(Four items; Cronbach’s alpha: .84)


• I usually look forward to each working day at this school.
• I wouldn’t want to work in any other school.
• I would recommend this school to parents seeking a place for their child.
• Teachers at this school respect other colleagues who are experts at their craft.

78

Instructional Innovation and Improvement
(Three items; Cronbach’s alpha: .90)

How many teachers in this school:
(Response categories: None, Some, About Half, Most, All)

• Set high standards for themselves?
• Are willing to try new ideas?
• Are really trying to improve their teaching?

Teacher Collective Responsibility

How many teachers in this school:
(Response categories: Some, About Half, Most, All, None)

• Help maintain discipline in the entire school, not just their classroom?
• Take responsibility for improving the school?
• Feel responsible for helping each other do their best?
• Feel responsible when students in this school fail?

Use of the Core Curriculum (Spring 2006)

I use the Core Curriculum:
(Response categories: Never, Occasionally, Often, Always)

• To guide subject/topic coverage
• To organize and develop instructional units and classroom activities
• To redesign assessment strategies

Use of the Core Curriculum (Spring 2007)

During the past twelve months, how often did you use the following
components of the District’s Core Curriculum?
(Response categories: Never, Occasionally, Often, Always)

• The Planning and Scheduling Timeline
• The Writing Plan
• The Course of Study and Prerequisite Skills
• The Coordinating Documents

79

Usefulness of Benchmarks to Inform Instruction
(Seven items; Cronbach’s Alpha:.92)

To what extent do you disagree or agree with the following questions?
(Response categories: Strongly Agree, Agree, Disagree, Strongly Disagree)

• Benchmark test scores give me information about my students
that I didn’t already know.
• The Benchmarks set an appropriate pace for teaching the curriculum
to my students.
• Results on the Benchmark tests give me a good indication
of what students are learning in my classroom.
• At my school, the use of Benchmark tests has improved
instruction for students with skill gaps.
• The Benchmark tests are a useful tool for identifying the content
descriptors that students do and do not understand.
• The Benchmark tests are a useful tool for identifying students’
misunderstandings and errors in their reasoning.
• The Benchmark tests are a useful tool for helping students
identify what they know and what they still need to learn.

Collective Examination of Benchmarks

• During the past 12 months, how often did the following occur in your school?
(Response categories: Never, 1-2 times, 3-5 times, More than 5 times)
• Your grade group, field coordinators, or coaches met to discuss ideas for re-
teaching a skill that students were lacking, according to the Benchmark test.
• Your grade group, field coordinators, or coaches met to discuss re-grouping
students for instruction on the basis of Benchmarks scores.

80

Access to and Support for Technology Use

Does the following exist in your classroom or school?
(Response categories: Yes, No)

• Internet in the classroom

• Our school’s technology coordinator helps teachers
integrate computing technology into lessons.
• I can find help in my school when I have trouble using computing technology.
• The computing technology in my school is in good working order.

Professional Development on Using Data
(Four items; Cronbach’s Alpha: .84)

Over the past 12 months, which of the following have been the focus of a professional
development session, faculty meeting, grade group meeting, or subject area
meeting?
(Response categories: Check all that apply)

• Accessing your students’ performance data on the computer
• Principal and/or school leadership team presentation about
your school’s performance data
• Using student performance data to develop an action plan
• Using student performance data to assess the effectiveness
of teaching practice

81

Authors

Jolley Bruce Christman

Jolley Bruce Christman, Ph.D. served as the Principal Investigator on this proj-
ect. She is a Founder and Principal of Research for Action. Most recently, her
research has focused on the topics of instructional communities, school leader-
ship, organizational learning, and privatization in public education. Another
important focus of her work has been on the use of research to inform policy
and practice. She has worked extensively with teachers, principals, parents,
students and other public school activists to incorporate research and reflection
into their efforts to improve urban public schools.

Ruth Curran Neild

Ruth Curran Neild, Ph.D. served as a Co-Principal Investigator on this project.
She is a Research Scientist at the Johns Hopkins University. Her scholarly
interests, broadly speaking, focus on improving educational outcomes for urban
youth through transforming their school experiences. She has published in the
areas of high school choice, teacher quality, the ninth grade transition, high
school reform, and high school graduation and dropout. She is committed to
communicating clearly about research findings to practitioners and policy-
makers and is a frequent presenter at conferences and workshops.

Katrina Bulkley

Katrina Bulkley, Ph.D. served as Co-Principal Investigator on this project. She
is an Associate Professor of Educational Leadership at Montclair State
University. Her work explores the role of governance changes in educational
reform. Her recent studies have focused on the role of for-profit and non-profit
management organizations in the operations of public schools nationally and in
Philadelphia. She is the editor (with Priscilla Wohlstetter) of Taking Account of
Charter Schools: What’s Happened and What’s Next? (2004, Teachers College
Press) and (with Lance Fusarelli) of “The politics of privatization in education:
The 2007 Yearbook of the Politics of Education Association.”

82

Suzanne Blanc

Suzanne (Sukey) Blanc, Ph.D. is an educational anthropologist and a former
middle school math teacher. She is a senior research consultant at Research for
Action and is the founder of Creative Research and Evaluation Services. Her
work centers on program evaluation and participatory research in urban
schools and communities. She has conducted numerous evaluations of National
Science Foundation projects in science, technology, and engineering and also
has a long-standing interest in the connection between education and other
aspects of urban life such as community arts, community, revitalization, and
community organizing.

Roseann Liu

Roseann Liu is a Ph.D. student at the University of Pennsylvania's Graduate
School of Education pursuing a dual degree in anthropology and education.
She is interested in the cultural productions of youth in transnational and dias-
poric communities. Prior to beginning graduate school, she was a Research
Associate at Research for Action.

Cecily Mitchell

Cecily Mitchell is especially interested in school-based interventions to improve
the educational experiences and outcomes for students who have been margin-
alized within the educational system. Her undergraduate thesis was based on
a participatory research project that examined how student academic engage-
ment is mediated by school rules and norms together with race and gender in
a 2nd grade classroom. Prior to coming to RFA, she worked in a school-based
behavioral health program to develop effective classroom interventions for
students with emotional/behavioral disabilities.

Eva Travers

Eva Travers, Ph.D. is Professor Emeritus at Swarthmore College where she
taught urban education and education policy. She is involved in ongoing
research by RFA on system-wide school reforms in Philadelphia. She held a
number of administrative positions at Swarthmore College, including Director
of the Program in Education, and Associate Dean. She has served on a variety
of national working groups and task forces looking at issues of teacher
preparation and teacher education.

RESEARCH FOR ACTION
3701 Chestnut Street
Philadelphia, PA 19104
ph 215.823.2500
fx 215.823.2510
www.researchforaction.org

Making the Most of Interim Assessment Data

More Related Content

What's hot (20)

Similar to Making the Most of Interim Assessment Data (20)

Recently uploaded (20)

Making the Most of Interim Assessment Data