Private Management of Public Schools: Early Experiences in Four School
Districts (Chapter Report, 04/19/96, GAO/HEHS-96-3).

Pursuant to a congressional request, GAO reviewed the experiences of
four school districts that contracted for public school management,
focusing on the effects of private management contract arrangements on
students.

GAO found that: (1) in Baltimore, Maryland, one contractor used its
average per pupil allocation to manage 9 of the district's 183 schools,
implement an instructional approach, and pay employees' salaries and
benefits, utilities, and leasing and maintenance costs; (2) the Dade
County, Florida, contractor raised $2 million to implement its
instructional approach in one school and hire principals and teachers,
but the district retained its responsibilities over the school's budget,
food service, and maintenance and repair; (3) in Hartford, Connecticut,
the contractor received district, state, and federal funds to manage the
district's 32 schools and made recommendations for improving
instruction; (4) the Minneapolis, Minnesota, contractor served as the
district's superintendent and was paid for achieving specific goals,
objectives, and assignments; (5) while the school boards and teachers
unions supported implementation of private management in Dade County and
Minneapolis, there was opposition to private management in Baltimore and
Hartford; (6) in Baltimore, teaching assistants with college degrees
were placed in each class, students had access to more computers, and
schools were cleaner and better maintained; (7) in Dade County,
attendance rates improved, teaching assistants with college degrees
increased, and students had access to more computers; (8) in Hartford,
students had access to more computers and the contractor was in the
process of repairing school buildings and installing computer labs in
five schools; (9) in Minneapolis, attendance rates improved and
suspension rates declined; and (10) in each of the four districts,
standardized achievement test scores did not significantly increase or
were not available.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  HEHS-96-3
     TITLE:  Private Management of Public Schools: Early Experiences in 
             Four School Districts
      DATE:  04/19/96
   SUBJECT:  Management consultants
             Public schools
             Privatization
             Service contracts
             Education or training costs
             Facility maintenance
             Students
             State and local procurement
IDENTIFIER:  Baltimore (MD)
             Dade County (FL)
             Hartford (CT)
             Minneapolis (MN)
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


Cover
================================================================ COVER


Report to Congressional Committees

April 1996

PRIVATE MANAGEMENT OF PUBLIC
SCHOOLS - EARLY EXPERIENCES IN
FOUR SCHOOL DISTRICTS

GAO/HEHS-96-3

Private Management of Public Schools

(104800)


Abbreviations
=============================================================== ABBREV

  ANOVA - analysis of variance
  CTBS - Comprehensive Test of Basic Skills
  EAI - Education Alternatives, Inc. 
  MSPAP - Maryland State Performance Assessment Program
  NCE - normal curve equivalent
  PSG - Public Strategies Group

Letter
=============================================================== LETTER


B-259870

April 19, 1996

The Honorable Nancy L.  Kassebaum
Chairman, Committee on Labor
 and Human Resources
United States Senate

The Honorable Arlen Specter
Chairman, Subcommittee on Labor,
 Health and Human Services,
 Education, and Related Agencies
Committee on Appropriations
United States Senate

As you requested, we are providing our report about the early
experiences of four school districts that contracted with private
companies for management of their public schools. 

As arranged with your office, unless you publicly announce the
contents of this report earlier, we plan no further distribution
until 30 days after its issue date.  At that time, we will send
copies of this report to the Secretary of Education, appropriate
congressional committees, and other interested parties.  If you have
any questions about this report, please call me at (202) 512-7014 or
Fred Yohey, Assistant Director, at (202) 512-7218.  Other GAO
contacts and acknowledgments are listed in appendixes IX and X. 

Cornelia M.  Blanchette
Associate Director, Education and
 Employment Issues


EXECUTIVE SUMMARY
============================================================ Chapter 0


   PURPOSE
---------------------------------------------------------- Chapter 0:1

Troubled by children's low test scores, as well as their low
attendance, promotion, and graduation rates, educators and parents
have searched for ways to improve public schools.  School districts
nationwide are experimenting with a range of reform options, one of
which is private management of public schools. 

Because of their interest in school reform, the Chairman, Senate
Committee on Labor and Human Resources; and the Chairman,
Subcommittee on Labor, Health and Human Services, Education and
Related Agencies, Senate Committee on Appropriations, asked GAO to
provide information on the early experiences of school districts that
contracted with private, for-profit companies for management of the
public schools.  Specifically, GAO was asked to describe (1) what the
private companies and school districts were required to do under the
contracts, (2) what happened in the school districts as the contracts
were being implemented, and (3) the impact of private management
efforts on school students. 

To respond to this request, GAO (1) visited the four school
districts--Baltimore City Public School District, Baltimore,
Maryland; Dade County Public Schools, Dade County, Florida; Hartford
School District, Hartford, Connecticut; and Minneapolis School
District, Minneapolis, Minnesota--that had contracts with private,
for-profit companies for school management in school year 1994-95 and
(2) reviewed the contracts and reported school operating expenses and
analyzed student outcome data. 


   BACKGROUND
---------------------------------------------------------- Chapter 0:2

School districts have historically contracted with private companies
for noninstructional services, such as student transportation,
building maintenance, and food provision.  Some school districts have
also contracted with private companies for limited instructional
services to specific student groups, such as at-risk children.  To
date, however, only a few districts have signed contracts with
private companies for managing their public schools.  More
specifically, only four school districts had contracts with two
private companies--Education Alternatives, Inc.  (EAI) and Public
Strategies Group (PSG)--to provide such services in the 1994-95
school year. 


   RESULTS IN BRIEF
---------------------------------------------------------- Chapter 0:3

The four school districts, which GAO visited, gave their private
management companies varying authority and responsibilities.  For
example, EAI had a 5-year contract with the Baltimore City Public
School District to manage 9\1 of the district's 183 schools and a
5-year contract with the Hartford School District to manage all of
the district's 32 schools.  In contrast, EAI's 5-year contract with
the Dade County Public School District gave the company authority to
implement its instructional approach in only one of the district's
schools.  PSG, on the other hand, has a 3-year contract with the
Minneapolis School District to provide leadership to the district,
including a superintendent, and to achieve certain specific goals,
such as improve students' test scores and attendance and reduce
suspensions. 

The districts' experiences during the implementation of the contracts
also differed substantially.  In Baltimore and Hartford, where
opposition in the community was considerable, implementation was
difficult, and the contracts were terminated or will be terminated
before the expiration of their 5-year terms.  In Baltimore, both the
district and EAI cited the district's budgetary constraints as the
reason for contract termination.  Regarding the Hartford contract,
EAI said it ceased services due to the district's failure to pay for
services rendered in accordance with the contract.  Hartford, on the
other hand, said that the relationship broke down because EAI
concluded that it would not operate under the contract as written. 
In contrast, in Dade County and Minneapolis, where opposition was
virtually nonexistent, implementation was easier, and EAI and PSG,
respectively, were generally able to implement their contracts.  The
Dade County contract expired in June 1995, and the Minneapolis
contract remains in effect until December 1996. 

In Baltimore, EAI implemented its instructional approach and changed
spending patterns for the nine schools by spending more than the
district had spent on facilities and computer hardware and software. 
In Hartford, EAI focused its management efforts on 6 schools and
provided fax machines and copiers for the district's 32 schools.  In
Dade County, EAI also implemented its instructional approach, which
included the use of computer technology and teacher training, in one
school.  In Minneapolis, PSG's president served as the
superintendent, and the company partially achieved some of the goals
specified by its contract. 

Regardless of the status of the contract, these private management
efforts yielded mixed results.  Although scores on standardized
achievement tests did not substantially improve in the three
districts where test scores were available for analysis in all four
districts, the private management companies made changes that
benefited students.  For example, student attendance improved in Dade
County and Minneapolis.  In addition, EAI's instructional approach,
implemented in Baltimore and Dade County, placed teaching
assistants--most of them with college degrees--in every classroom. 
Moreover, EAI also enhanced school building repair and maintenance
and increased the number of computers in Baltimore and Hartford,
giving students greater access to computers. 


--------------------
\1 Baltimore and EAI subsequently entered into three additional
contracts, each for one school.  However, GAO included only the first
contract, which covered nine schools, in its study because this
contract was the largest and was signed much earlier than the
additional three. 


   PRINCIPAL FINDINGS
---------------------------------------------------------- Chapter 0:4


      RESPONSIBILITY AND AUTHORITY
      VARIED UNDER CONTRACTS
-------------------------------------------------------- Chapter 0:4.1

Under the contracts, private companies and districts had varying
authority and responsibilities for managing schools.  In Baltimore,
EAI received the average per pupil allocation to manage 9 of the
district's 183 schools and implement an instructional approach.  With
the money, allocated to EAI in monthly allotments, the company was to
pay the costs of operating the nine schools, including employees'
salaries and benefits, utilities, leasing costs, and repairs and
maintenance.  EAI was to keep any money left over after all school
expenses had been paid as compensation for its services. 

In Dade County, EAI agreed to raise over $2 million to implement its
instructional approach in one of the district's schools.  This money,
which was controlled and disbursed by the district, was used to pay
the cost of implementing EAI's instructional approach, including its
consulting fee.  EAI also helped the district in hiring a newly built
school's principal and teachers.  The district was responsible,
however, for managing all other aspects of the school, such as its
budget, food service, and building maintenance and repair. 

Under the Hartford contract, EAI was to make recommendations for
improving instruction.  EAI was to also manage the operations of the
32-school district with the funds available under the district's
school appropriation and grant money, including federal and state
grants.  EAI and Hartford, however, ultimately disagreed on the
interpretation of many key contract provisions relating to the
payment and control of funds. 

Minneapolis' contract requires that PSG provide leadership services
to the district.  Specifically, PSG serves as the school district
superintendent for a monthly fee of $5,000 and is compensated for
achieving certain goals, objectives, and assignments in the
district's improvement agenda.  For example, the agenda specified
that PSG improve student test scores and attendance, reduce
suspensions, enhance district leadership, and improve community trust
in the district's schools.  Except for the superintendent's salary,
PSG is only paid when it completes assignments and meets the contract
goals and objectives.  PSG characterizes its contract as a
"pay-for-performance" contract. 


      SCHOOL DISTRICTS'
      EXPERIENCES IN IMPLEMENTING
      PRIVATE MANAGEMENT DIFFERED
-------------------------------------------------------- Chapter 0:4.2

In implementing private management, Dade County's and Minneapolis'
efforts were generally supported by such stakeholders as the school
board and teachers union.  In Dade County, where the contract expired
in June 1995, EAI helped the district hire the principal and teachers
for the school and implemented its instructional approach.  In
Minneapolis, which has a contract in effect until December 1996,
PSG's president serves as the superintendent, and, during the first
18 months of the 3-year contract, PSG partially achieved some of the
goals it was required to accomplish under the contract. 

Both the Baltimore and Hartford districts faced opposition in the
community to private management companies.  For example, in
Baltimore, the teachers union opposed EAI's requirement that teaching
assistants have 90 college credits or be transferred to nonprivately
managed schools.  Citing budget constraints, the district terminated
its contract with EAI in March 1996.  In Hartford, where the district
has announced that it will terminate its contract, resistance reached
its peak when EAI submitted a budget proposal to eliminate a
substantial number of teaching jobs. 

Even with the opposition in the community, EAI was generally able to
implement the Baltimore contract.  The company reported spending
about the same amount each year on direct operations for the nine
schools as the district had reported spending during the year
preceding the contract.  However, EAI allocated funds differently by
spending more on facilities and computers and less, in the first year
of private management, on general and special education instruction. 

In Hartford, the district and EAI agreed that EAI would prioritize
its efforts in 6 of the 32 schools.  Hartford, in responding to a
draft of this report, stated that EAI suggested this six-school focus
to achieve "showcase results quickly as a strategy to build community
support." Hartford explained, however, that the district could not
come to an agreement with EAI on new payment terms, and EAI was
unwilling to continue to perform under the existing contract. 
According to EAI, however, it ceased services to the district due to
the failure of the Hartford district to pay it for services rendered
in accordance with the contract. 


      BENEFITS FROM PRIVATE
      MANAGEMENT EFFORTS
-------------------------------------------------------- Chapter 0:4.3

School students enjoyed a number of benefits as a result of private
management.  In Dade County and Minneapolis, district officials
reported that attendance rates improved.  In addition, in
Minneapolis, officials reported that suspension rates for all
Minneapolis students in school year 1994-95 generally decreased from
school year 1993-94.  In Baltimore and Dade County, students received
individualized instruction because EAI placed a teaching assistant,
who usually had a college degree, in each class.  In addition, in
Baltimore, Dade County, and Hartford, students had access to more
computers.  In Baltimore, schools were cleaner and better maintained
and repaired; EAI also painted the schools, improved heating and air
conditioning, repaired bathrooms, and improved landscaping.  To
reduce energy costs, EAI also retrofitted lighting in the Baltimore
schools.  In Hartford, EAI was in the process of repairing school
buildings and installing computers and computer labs in 5 of the
district's 32 schools. 

Despite the positive impact of private management efforts, one
measure of student outcome--scores on standardized achievement
tests--did not substantially improve in Baltimore, Dade County, and
Minneapolis.  Because the Hartford schools were in the early stages
of contract implementation, student outcomes were not available for
assessment. 


   RECOMMENDATIONS
---------------------------------------------------------- Chapter 0:5

GAO is making no recommendations. 


   SCHOOL DISTRICT AND COMPANY
   COMMENTS
---------------------------------------------------------- Chapter 0:6

Comments on a draft of this report were requested from the Department
of Education and the four school districts and two companies that
were the primary focus of GAO's work.  Only the Department of
Education declined to provide comments.  In general, the comments,
which are addressed in chapter 5, clarified or provided additional
information pertinent to issues discussed in the report.  Changes
have been made where appropriate.  Appendixes IV through VIII contain
the text of the comments. 


INTRODUCTION
============================================================ Chapter 1

School districts' contracting with private, for-profit companies
reflects an increasing trend toward private management of public
services and concerns about the quality of U.S.  education.  As a
result, school districts nationwide are experimenting with a range of
reform options, including school-based management,\2 charter
schools,\3 and contracting with private companies for management of
their public schools. 

School districts have historically contracted with private companies
for noninstructional services, such as student transportation,
building and vehicle maintenance, and food provision.  School
districts have also contracted with private companies for limited
instructional services to specific student groups, such as at-risk
children.  To date, however, despite considerable discussion, only a
handful of districts have signed contracts with private companies to
manage schools or school districts. 

In some school districts, the decision to contract with a private
company for school management followed years of frustration with
school performance.  School districts began contracting with private
companies in the hope that they would (1) bring new ideas and a bold
new direction to public schools and (2) do a better job of educating
children. 

Federal and state laws support innovative efforts to improve
education.  For example, under the Goals 2000:  Educate America Act,
enacted by the Congress in 1994, state education agencies may use
federal funds for state planning and evaluation activities involving
local efforts to contract with private management organizations to
reform public schools.  Some states have laws authorizing charter
schools that permit private companies to enter the public education
arena.  Under such laws, private companies, teachers, school
administrators, and other members of the community may propose,
create, or manage a publicly funded school. 


--------------------
\2 For more information on school-based management, see Education
Reform:  School-Based Management Results in Changes in Instruction
and Budgeting (GAO/HEHS-94-135, Aug.  23, 1994). 

\3 Charter schools, which are authorized by state legislatures, are
public schools that are created by parents, teachers, school
administrators, and other members of the community.  These schools
have considerable autonomy and often operate free from such external
controls as teachers' unions and district and state requirements. 
For more information on charter schools, see Charter Schools:  New
Model for Public Schools Provides Opportunities and Challenges
(GAO/HEHS-95-42, Jan.  18, 1995). 


   CHARACTERISTICS OF SCHOOL
   DISTRICTS SELECTING PRIVATE
   MANAGEMENT
---------------------------------------------------------- Chapter 1:1

In school year 1994-95, four school districts--Baltimore City Public
School District, Baltimore, Maryland; Dade County Public Schools,
Dade County, Florida; Hartford School District, Hartford,
Connecticut; and Minneapolis School District, Minneapolis,
Minnesota--had contracts with private companies for school
management.\4


--------------------
\4 In the 1995-96 school year, five additional public schools in
Boston, Massachusetts; Mt.  Clemens, Michigan; Sherman, Texas;
Wichita, Kansas; and Wilkinsburg, Pennsylvania are receiving
management services from private companies.  This report does not
discuss these districts' experiences. 


      BALTIMORE CITY (MARYLAND)
      PUBLIC SCHOOL DISTRICT
-------------------------------------------------------- Chapter 1:1.1

In Baltimore, 1 in 4 citizens is functionally illiterate.  The
average student is a poor, African American child living in a
female-headed, single-parent household; he or she has not met state
standards for test scores, has missed 1 in 5 days of school, and has
only 1 chance in 2 of earning a high school diploma. 

For its 183 schools, the Baltimore City Public School District spent
an average of $5,948 per student in school year 1994-95.  However,
Baltimore spent $40,000 less per classroom than districts in the
nearby Maryland suburbs.  School officials believed that Baltimore's
lower classroom funding level made it difficult for the district to
address the low test scores, low attendance rates, and extremely high
secondary school dropout rate. 


      DADE COUNTY (FLORIDA) PUBLIC
      SCHOOL DISTRICT
-------------------------------------------------------- Chapter 1:1.2

The Dade County Public School District is an inner city district with
a multicultural population.  Foreign-born students enter the schools
at a rate of 11,000 to 15,000 per year.  Nearly 60 percent of the
district's students have a native language other than English.  In
school year 1994-95, over half of the students received a free or
reduced-priced lunch, an indicator of poverty. 

For its 286 schools, the Dade County Public School District spent an
average of $4,773 per student in fiscal year 1994-95.  About 18
percent of the students drop out of school each year.  In school year
1993-94, almost 40 percent of Dade County's students scored in the
bottom quartile in reading on standardized tests. 


      HARTFORD (CONNECTICUT)
      SCHOOL DISTRICT
-------------------------------------------------------- Chapter 1:1.3

Hartford is the poorest city in Connecticut.  Nevertheless, the
Hartford School District spent $8,688 per pupil in school year
1994-95, well over the national average of $5,170.  Despite such
funding, district officials believed the district lacked the
resources to operate its schools in a "proper and efficient" manner. 
Teacher salaries, which averaged about $58,000 a year, consume a
significant portion of the district's budget. 

In the 32-school district, test scores are well below the state
average.  About 16 percent of students drop out each year; about 93
percent of students are minority.  The district has 40 percent of the
state's bilingual students and 16 percent of its students are special
education students.  In school year 1994-95, approximately 78 percent
of Hartford students received free or reduced-price lunches. 


      MINNEAPOLIS (MINNESOTA)
      SCHOOL DISTRICT
-------------------------------------------------------- Chapter 1:1.4

The Minneapolis School District is in a large, multicultural city,
which has undergone dramatic changes in ethnicity, race, and income
during the past 20 years.  In school year 1994-95, 36 percent of the
families received benefits from Aid to Families With Dependent
Children; 65 percent of children in first grade were eligible to
receive free or reduced-price lunches. 

In the Minneapolis School District, where the average per pupil
expenditure was $6,408 in school year 1994-95, minority children
account for 61 percent of the students.  That number is projected to
reach 70 percent by 1998.  In school year 1994-95, 39 percent were
African American children; 12 percent, Asian American; 4 percent,
Hispanic American; 7 percent, Indian American; and 39 percent, white
American.  These children come from families in which over 70
different languages and dialects are spoken.  Languages include
Cambodian, Hmong, Lao, Russian, Spanish, Ukrainian, and Vietnamese,
among others.  The gap in test scores between minority and white
students has persisted and is growing yearly. 


   PRIVATE COMPANIES BRING
   DIFFERENT APPROACHES TO SCHOOL
   MANAGEMENT
---------------------------------------------------------- Chapter 1:2

Two private companies--Education Alternatives, Inc.  (EAI) and Public
Strategies Group (PSG)--had contracts with school districts for
school management in the 1994-95 school year.  The companies differ
in their approaches to providing services. 


      EAI'S APPROACH
-------------------------------------------------------- Chapter 1:2.1

EAI's approach focuses on three aspects of the schools:  education,
facilities, and financial management.  EAI has joined with three
other companies--KPMG Peat Marwick, Johnson Controls World Services,
Inc., and Computer Curriculum Corporation--to provide services to
schools.  Collectively, they refer to themselves as the Alliance for
Schools That Work.  As the lead company of the alliance, EAI is
responsible for overall school management and provides education
management through its Tesseract\5 instructional approach.  Major
components of Tesseract include a personal education plan for each
child, low adult-to-student ratios in the classroom, continuous
training for teachers, active parental involvement, the use of
computer technology in instruction, and special education
inclusion.\6

Other companies in the alliance provide additional services.  KPMG
Peat Marwick, the world's largest public accounting and consulting
company, offers financial management assistance.  Johnson Controls
World Services, Inc., which has operated, maintained, and managed
facilities for over a century, provides supervision and maintenance
of buildings, energy use, transportation, and other noninstructional
services for schools.  Computer Curriculum Corporation, owned by
Paramount Communications, the world's largest entertainment and
educational publishing corporation, provides computer-based,
instructional software. 


--------------------
\5 The word Tesseract, a registered EAI trademark, comes from a
children's book, A Wrinkle in Time, by Madeleine L'Engle.  In the
book, the word is defined as a fifth-dimensional corridor leading to
destinations otherwise beyond reach. 

\6 Inclusion, as used by EAI, describes placing special education
children in regular classrooms instead of self-contained classrooms. 
Classroom teachers are responsible for educating all students in the
classroom. 


      PSG'S APPROACH
-------------------------------------------------------- Chapter 1:2.2

PSG provides superintendent services for the Minneapolis School
District.  Its president functions as the Minneapolis School
District's superintendent and leads a team of employees from PSG who
help manage the school district.  According to PSG, it focuses on
helping schools in several ways, including improving performance. 
The company's stated goal is to develop current Minneapolis district
administrators to take over the management of the schools when the
company's contract expires. 


   OBJECTIVES, SCOPE, AND
   METHODOLOGY
---------------------------------------------------------- Chapter 1:3

Because of their interest in school reform, the Chairman, Senate
Committee on Labor and Human Resources, and the Chairman,
Subcommittee on Labor, Health and Human Services, Education and
Related Agencies, Senate Committee on Appropriations, asked that we
provide information about the experiences of school districts that
have contracted with private companies for management of their public
schools.  Specifically, we were asked to describe (1) what the
private companies and school districts were required to do under the
contracts, (2) what happened in the school districts as the contracts
were being implemented, and (3) the impact of private management
efforts on students. 

To obtain this information, we used several methodologies:  contract
reviews, site visits, and review and analysis of reported school
operating expenses and student outcomes.  We reviewed the contracts
for and visited schools in the four districts that had contracts for
school management with private, for-profit companies in school year
1994-95.  Our study excluded the five additional districts that had
signed contracts for private company management of schools in the
1995-96 school year because the districts either had not begun
operating under the contracts or had only been doing so for a few
months.  Our study also excluded school districts with contracts for
only highly specialized instructional services, such as teaching
foreign languages or tutoring.  Nor did we study districts that
contracted solely for noninstructional services. 


      SITE VISITS
-------------------------------------------------------- Chapter 1:3.1

We visited the Baltimore City, Dade County, Hartford, and Minneapolis
school districts.  Three of these districts had contracts with EAI
and one with PSG.  By visiting only districts with contracts in
effect for the 1994-95 school year, we were able to observe program
implementation. 

In each district, we talked to stakeholders--superintendents, school
board members, private company officials, school administrators, city
council members, teachers, principals, teachers union
representatives, and parents.  We asked about the changes that had
taken place under the private company's direction, including services
provided, difficulties in program implementation, and results to
date.  Except for the Minneapolis School District, where primarily
superintendent services were provided, we visited privately managed
schools. 


      ANALYSIS OF STUDENT OUTCOMES
-------------------------------------------------------- Chapter 1:3.2

In the Baltimore City Public School District, which had the longest
running and most comprehensive private management effort under way,
we analyzed 4 years of data on student outcomes for seven of the nine
privately managed public schools.\7 We reviewed data for 1 year
before private management as well as for 3 years during private
management.  We matched each privately managed school with a
similar\8 nonprivately managed school and then compared standardized
student test scores and attendance.  (See table III.1 in app.  III
for a list of schools in the analysis.)

We looked at only two areas for test scores, reading and math.  For
both areas, scores were available from routinely administered
standardized tests.  We used normal curve equivalent (NCE) scores
from the Comprehensive Test of Basic Skills (CTBS) and, when
available, outcome scale scores from the Maryland State Performance
Assessment Program (MSPAP).  (See app.  I for a discussion of our
selection of statistical procedures and technical details about the
analyses.)

We used two types of analyses--longitudinal and cross-sectional--to
compare student outcomes.  In both types of analyses, we compared
test scores and attendance measures of students attending the
privately managed schools with those of students attending matched,
nonprivately managed schools.  In the cross-sectional analysis, we
performed the comparison at the school level, analyzing each year
separately, disregarding changes in the student population from year
to year.  In the longitudinal analysis, we limited the analysis to
those students who were enrolled in a privately managed school (or
its matched, comparison school) for the entire 3-year period. 
Students meeting this criterion were aggregated into a privately
managed or nonprivately managed group for the analysis. 

We did not conduct our own analyses of student outcomes for test
scores and attendance in the Dade County or Minneapolis school
districts.  Instead, we used findings from evaluations conducted by
the districts themselves, which sometimes included measures other
than test scores and attendance (for example, suspensions).  We did
not analyze student outcome data for the Hartford School District,
which was in the relatively early stages of implementation. 


--------------------
\7 Although the company managed nine schools, we limited our analysis
to the seven elementary schools that included kindergarten through
fifth grades because (1) more test score data were available for this
group and (2) the pool from which to select a comparison school was
larger. 

\8 Schools were matched on the basis of demographics and test scores
before private management. 


      ANALYSIS OF REPORTED SCHOOL
      OPERATING EXPENSES
-------------------------------------------------------- Chapter 1:3.3

Of the four school districts we visited, Baltimore City and Hartford
were the only districts in which the company was to manage school or
district budgets.  Although Hartford was not far enough along in its
budget implementation for a financial analysis, we did compare
reported operating expenses for Baltimore for 1 year before private
management with reported operating expenses for 2 years during
private management. 

To carry out our analysis, we obtained and reviewed reported
operating expense data provided by the district and EAI.  We could
not, however, determine the reliability of the expenses provided by
the Baltimore district because audits were not done for individual
schools.  When we attempted to verify the EAI data, EAI would not
provide the supporting documentation we needed.  As a result, we
based explanations for differences between reported operating
expenses before and during private management solely on oral or
written explanations from EAI and Baltimore City Public School
District officials.  However, for the nine Baltimore schools, we were
able to reconcile EAI's fiscal year 1993 and 1994 expenses with its
audited statements of cash receipts and disbursements. 

We did our study between December 1994 and December 1995 in
accordance with generally accepted government auditing standards. 


AUTHORITY AND RESPONSIBILITIES
VARIED UNDER PRIVATE MANAGEMENT
CONTRACTS
============================================================ Chapter 2

The four school districts we visited gave private companies varying
authority and responsibilities in managing their schools.  Under the
contracts, school districts also retained varying powers and
responsibilities and compensated the companies in different ways. 
Except for the Minneapolis contract, however, the contracts that
school districts had with private companies were similar in their
lack of requirements for improving student outcomes, which are
traditionally measured by test scores and attendance and graduation
rates. 


   EAI GIVEN BROAD AUTHORITY TO
   MANAGE NINE BALTIMORE PUBLIC
   SCHOOLS
---------------------------------------------------------- Chapter 2:1

In July 1992, the Baltimore City Public School District signed a
5-year contract with EAI that gave EAI broad authority and
responsibility for managing 9\9 of the district's 183 schools.  Under
the contract, EAI was to (1) implement its Tesseract instructional
approach, which included supplying computers for use as instructional
tools and college-educated teaching assistants; (2) provide building
maintenance and other noninstructional services; (3) manage the nine
schools' budgets; and (4) determine school staffing levels for the
nine schools with the approval of the school district superintendent. 
The district could terminate the contract upon 90 days' written
notice. 

The company had authority to discuss matters concerning the nine
schools with union representatives and to participate in evaluating
employees at the schools after obtaining the required approval. 
However, EAI could not reprimand or terminate any school system
personnel.  EAI also had authority to hire its own personnel to
implement its Tesseract approach. 

In managing the nine schools, EAI could either pay a subcontractor to
provide noninstructional services or purchase them from the district;
however, EAI could not subcontract for instructional services without
the district's expressed agreement. 

Annually, the Baltimore City Public School District was to pay EAI in
12 monthly installments, with the amount of each installment based on
the district's historical expenditure patterns.  The total payment to
EAI was to be the districtwide average per pupil expenditure
multiplied by the total number of students in the nine schools.  With
the money, which totaled $26.7 million in the first contract year,
EAI was to pay the costs of operating the nine schools, including
employees' salaries and benefits, utilities, leasing costs for
computers and other equipment, and repairs and maintenance.\10 EAI
paid the district $3.4 million to cover the nine schools' share of
central support services provided by the district, leaving EAI $23.3
million for the direct operation of the schools. 

Under this payment arrangement, EAI's payment for managing the nine
schools was the difference between the total payment received from
the school district--excluding interest earned and grant money--and
the cost of operating the schools.  For the first 2 years of the
contract, EAI reported gross profits of $1.9 million and $3.3
million, respectively.  According to an EAI official, these were
gross, and not net, profits because they did not reflect time that
EAI's corporate staff spent on contract matters. 


--------------------
\9 Baltimore and EAI subsequently entered into three additional
contracts, each for one school.  However, we included only the first
contract, which covered nine schools, in our study because it was the
largest and was signed much earlier than the additional three. 

\10 EAI was only responsible for the ordinary maintenance and repair
of the nine schools.  The district was responsible for major repairs
or capital expense items, as well as repairs for asbestos-related
conditions. 


   EAI HAD AUTHORITY TO IMPLEMENT
   THE TESSERACT INSTRUCTIONAL
   APPROACH IN ONE DADE COUNTY
   SCHOOL
---------------------------------------------------------- Chapter 2:2

The Dade County Public School District's 5-year contract with EAI,
which covered June 1990 through June 1995, gave the company the
authority to implement its Tesseract instructional approach at Dade
County's South Pointe Elementary School.  However, EAI was to raise
the money needed to implement the Tesseract program with the
cooperation of the district.  EAI was also to help the district hire
the school's principal and its teachers.  The district was
responsible for managing all other aspects of the school, such as its
budget, food service, and building maintenance and repair.  The
contract allowed the district or EAI to terminate the contract at the
end of any academic year upon 90 days' prior written notice. 

Under the contract, EAI agreed to raise over $2 million, of which
$1.4 million was to be raised in the first 3 years.  This amount was
to supplement the funds the school ordinarily received from the
district to provide school services.  The district was to deposit the
money that EAI raised under the contract into an interest-bearing
account from which only the district could withdraw money.  The money
was to be used to pay for special staff needs, instructional
materials and equipment, and EAI's out-of-pocket expenses and
consulting fee. 


   EAI GIVEN AUTHORITY TO MANAGE
   THE OPERATIONS OF HARTFORD
   DISTRICT'S 32 SCHOOLS
---------------------------------------------------------- Chapter 2:3

The Board of Education for the Hartford School District signed a
5-year contract with EAI in November 1994, which became retroactive
to July 1994.  Under the contract, EAI was to perform management and
operations tasks necessary to achieve the goals of the district's
strategic plan.  EAI was to also assume responsibility for managing
the operations of the 32 schools in the district while the Board of
Education retained policy-making and ultimate decision-making
authority.  Unlike the Baltimore and Dade County contracts, the
Hartford contract did not specify that EAI implement its Tesseract
instructional approach.  Rather, EAI was to recommend and implement
enhancements to the educational program.  In addition, the contract
stated that over the 5-year term EAI expected to spend about $20
million on technology and software initiatives and $1.6 million on
building improvements.  EAI was also to recommend and, with approval
of the Board of Education, implement enhancements to improve student
performance, including staff training and student evaluation. 

EAI was given authority to purchase materials and services from
commercial sources and to make recommendations concerning staffing
levels; organizational structures; and the hiring, assignment,
duties, compensation, discipline, and discharge of district
employees, including the superintendent.  However, the Board of
Education remained the final authority for all personnel and
organizational structure decisions.  The contract also allowed EAI to
provide advice during negotiations with labor organizations, but the
Board had the statutory responsibility for collective bargaining and
administering contracts with labor organizations. 

Under the contract, the district's entire annual school appropriation
and all grant money (including federal and state grants), which
totaled about $200 million in the first contract year, were to be
used by EAI to manage the schools.  EAI and Hartford, however,
ultimately disagreed on the interpretation of many key contract
provisions relating to payment and control of funds. 

The district had the right to terminate the contract upon 90 days'
written notice to EAI.  EAI could also terminate the contract if it
believed that the annual appropriation was insufficient to meet all
contractual financial requirements. 


   THE MINNEAPOLIS SCHOOL DISTRICT
   CONTRACTED WITH PSG FOR
   LEADERSHIP AND THE ACHIEVEMENT
   OF SPECIFIC GOALS
---------------------------------------------------------- Chapter 2:4

In December 1993, the Minneapolis School District signed a 3-year
contract with PSG.  Under the contract, the district employed PSG to
provide leadership to the district, including a superintendent. 
PSG's president was designated superintendent.  The company was to
implement the district's improvement agenda, which included, among
other things, to improve students' test scores and attendance rates,
instruction, and curriculum; reduce suspensions; enhance district
management and accountability; and maintain community trust and
involvement in the schools.  As superintendent, PSG was responsible
for the school district finances but within the district's budget and
payable systems.  The contract allowed the district or PSG to
terminate the contract with 30 days' advance written notice. 

Under the contract, PSG was to be paid a monthly fee of $5,000 for
serving as superintendent.  In addition, the company would receive
additional compensation for achieving the objectives in the
district's improvement agenda.  PSG characterizes the contract as a
"pay-for-performance" contract.  For each goal or objective, the
district established indicators, such as changes in test scores and
attendance rates, that were intended to help the district assess the
extent to which the goal had been achieved.  The agenda could be
amended from time to time to reflect additional goals and objectives. 


SCHOOL DISTRICTS' EXPERIENCES IN
IMPLEMENTING PRIVATE MANAGEMENT
CONTRACTS
============================================================ Chapter 3

The four school districts that awarded private management contracts
had different implementation experiences.  The Baltimore and Hartford
contracts proved difficult because of some opposition in the
communities to private management of the public schools, while Dade
County and Minneapolis implementation was easier because private
management was generally supported. 


   EAI, WITH SOME OPPOSITION, WAS
   GENERALLY ABLE TO IMPLEMENT ITS
   BALTIMORE CONTRACT
---------------------------------------------------------- Chapter 3:1

In Baltimore, the mayor, superintendent, and, initially, the teachers
union supported private management of the public schools.  The school
district selected for private management only those schools with
principals who were interested in private management.  In addition,
the district gave teachers who did not want to teach in privately
managed schools the option of transferring to other schools in the
district.  The teachers union, however, opposed EAI's requirement
that teaching assistants\11 have at least 90 college credits. 

The teaching assistants who did not meet this requirement were
transferred from the privately managed schools to other schools in
the district.  EAI replaced these assistants with a larger number of
teaching assistants, who usually had college degrees.  The Baltimore
teachers union protested the transfers as well as the replacements. 
The union contended that the original teaching assistants were
valuable because they were experienced and from schools' neighboring
communities and therefore provided inner-city children much needed
relationships with familiar, trusted adults. 

In addition, some people opposed EAI's special education inclusion
model and the manner in which it was implemented.\12

One group who opposed the model believed that the real motive of
EAI's special education inclusion program was to reduce the number of
special education teachers and thereby reduce costs.  In addition, a
national teachers union questioned whether the decision to implement
the inclusion model was in the children's best interest.  To further
complicate EAI's efforts to implement inclusion, the U.S.  Department
of Education found that EAI had not followed the procedures as
required by the Individuals With Disabilities Education Act before
changing special education students' placements by moving them to
regular classrooms.\13

Even with such opposition, EAI was able to generally implement its
contract.  However, with over 1 year remaining in the 5-year
contract, the school district decided to terminate the contract
effective March 1996.  School district officials had tried to
negotiate with EAI to reduce the amount of money EAI would receive to
operate the nine schools because of unexpected districtwide expenses. 
EAI would not agree to the reduced amount, and the district decided
in November 1995 to terminate the contract. 

Our analysis of reported operating expenses before and during private
management of the nine schools found that EAI spent about the same
each year on direct operations as the district had spent in the year
before the contract (see table 3.1).  However, EAI allocated funds
differently.  In its first year, EAI spent less on general
instruction and special education instruction but significantly more
on facilities.  In the second year, EAI's overall reported costs were
similar to those of its first year, but, again, EAI allocated funds
differently.  EAI increased its spending for general instruction
(primarily because of salary increases) but spent less than it had in
the previous year on special education.  EAI also spent less on
facilities than it had in the previous year (primarily, according to
an EAI official, because of the facility upgrading done during the
first year), but the company continued to spend more than the
district had previously spent on facilities. 



                               Table 3.1
                
                  Reported Operating Expenses in Nine
                  Baltimore Schools Before and During
                           Private Management

                         (Dollars in millions)


                                          First year of    Second year
                                                private     of private
                                             management     management
Program        Before private management   (school year   (school year
category         (school year 1991-92)         1992-93)       1993-94)
-------------  -------------------------  -------------  -------------
Teaching
General                  $14.0                    $13.2          $14.5
 instruction
Special                   2.3                       1.2            1.0
 education
 instruction
Food services             1.1                       1.1            1.1
Transportatio             0.2                       0.3            0.2
 n
Facilities                2.8                       4.4            3.8
======================================================================
Total                    $20.4                    $20.2          $20.6
----------------------------------------------------------------------
Note:  To make the figures comparable, we reclassified some
categories.  The table also excludes administrative overhead because
we could not get comparable data from Baltimore for the year before
private management. 


--------------------
\11 Baltimore refers to its teaching assistants as paraprofessionals
while EAI refers to them as instructional interns. 

\12 For a discussion of the advantages and disadvantages of such
programs, see Special Education Reform:  Districts Grapple With
Inclusion Programs (GAO/T-HEHS-94-160, Apr.  28, 1994). 

\13 These procedures include having a meeting about the change in
placement with parents, teachers, and other members of an educational
management team.  Baltimore and EAI, as a result, operated under a
plan whereby they are required to correct all violations. 


   EAI UNABLE TO FULLY IMPLEMENT
   ITS CONTRACT WITH THE HARTFORD
   SCHOOL DISTRICT
---------------------------------------------------------- Chapter 3:2

Although most school board members in Hartford supported private
management of public schools, many in the district did not agree with
them.  The superintendent, for example, had several concerns about
the district's decision to enter into a contract with EAI.  The
teachers union, as well as other unions, opposed private management
from the onset, possibly at least in part because they viewed EAI as
a vehicle for reducing teaching jobs. 

Opposition reached its peak when EAI submitted a budget proposal for
school year 1995-96 that would have eliminated a substantial number
of teaching positions.  EAI wanted to cut teacher costs and use the
savings to help fund technology initiatives specified in the
contract, as well as invest in clean and safe schools, implement
site-based management, and improve instruction.  However, most school
board members would not support the reduction in teachers.  In
commenting on a draft of this report, EAI said that it believed that
such a cut was warranted, saying school district teacher-pupil
ratios, determined by the district's contract with the teachers
union, had resulted in a system that was "overstaffed by millions of
dollars of personnel." According to Hartford, the cuts would have
resulted in massive violations of class size limitations contained in
the district's agreement with its teachers union. 

During the first year of the contract, the district and EAI agreed
that EAI would prioritize its efforts in 6 of the 32 schools. 
Hartford, in responding to a draft of this report, stated that EAI
suggested this six-school focus to achieve "showcase results quickly
as a strategy to build community support." EAI was to provide
specific education management services to the six schools.  These
services included (1) training teams of staff for site-based
management in five schools and one adult learning center and (2)
providing technology improvements, such as computers.  In January
1996, however, Hartford announced that it would terminate its
contract with EAI.  According to Hartford, the relationship broke
down and the contract will be terminated because EAI concluded that
it would not operate under the contract as written.  EAI, on the
other hand, said that it ceased services to Hartford due to the
district's failure to pay it for services rendered in accordance with
the contract.  In its 1995 annual report to its stockholders, EAI
stated that it had recorded costs totaling $5.5 million for the
Hartford contract but acknowledged uncertainties about whether the
district would reimburse it. 


   EAI MET THE TERMS OF ITS
   CONTRACT WITH DADE COUNTY
   PUBLIC SCHOOLS
---------------------------------------------------------- Chapter 3:3

The Dade County School Board and the teachers union supported EAI's
Tesseract program for one of the county's newly constructed public
schools.  EAI also helped the district hire the school's principal
and teachers, helping to ensure the support of those who would
implement the program. 

EAI raised $1.4 million of the over $2 million agreed to under the
contract and was able to implement its Tesseract instructional
approach.  The company also received in-kind donations estimated at
about $285,000.  Most of the money raised was used to pay for
teaching assistants' salaries and computer hardware and software. 

Dade County officials said that they were generally pleased with EAI. 
The contract expired in June 1995 but was not renewed.  According to
district officials, EAI wanted a larger role in school management,
but the district would not agree to this. 


   PSG ACHIEVED SOME OF THE GOALS
   SPECIFIED IN ITS MINNEAPOLIS
   CONTRACT
---------------------------------------------------------- Chapter 3:4

The Minneapolis school board unanimously agreed to sign the contract
with PSG amid general support for the company serving as school
superintendent.  District officials, however, did not view this
decision as private management of their schools.  The school board's
search for a superintendent was the same one used to select previous
superintendents.  This time, however, the board encouraged
nontraditional candidates--those without training or experience in
education--to apply. 

PSG's president was a well-known member of the Minneapolis community. 
He had previously served as a consultant to the district and had been
instrumental in improving financial management in the district's
schools. 

As of June 1995, after 18 months of the 3-year contract, the
Minneapolis School District determined that PSG had partially
achieved some of the goals and paid PSG 66 percent of the $716,500
that PSG was to receive under the contract if it had fully achieved
each goal.  For example, the company was paid when the suspension
rate dropped, attendance increased, and families' involvement in
their children's education increased.  PSG was also paid when it
developed baseline measures for assessing student performance,
identified the predictors of effective teaching, and developed a
strategic plan for the district.  It was not paid, however, when
student test scores did not improve or when it did not negotiate the
teachers' contract to the district's satisfaction.  The contract with
PSG remains in effect until December 1996. 


RESULTS OF PRIVATE MANAGEMENT ON
PUBLIC SCHOOL EDUCATION
============================================================ Chapter 4

Regardless of whether private companies fulfilled their contract
obligations, we found that contract services yielded some benefits
for students in all four districts.  For example, students received
individualized instruction and had greater access to computers and
cleaner school buildings.  In addition, private management companies
have served as catalysts for districts' rethinking of the status quo. 

Despite the positive impact of private management efforts, however,
one measure of student outcome--scores on standardized achievement
tests--did not substantially improve in Baltimore, where we analyzed
test scores.  Similar results were also reported in Dade County and
Minneapolis.  Hartford test score data, however, were not available
for analysis at the time of our review. 


   ATTENDANCE AND SUSPENSION RATES
   IMPROVED IN SOME DISTRICTS
---------------------------------------------------------- Chapter 4:1

In Dade County and Minneapolis, attendance rates improved; in
Minneapolis, suspension rates declined.  According to a Dade County
internal evaluation, after 2 years of private management,\14
attendance in the privately managed elementary school improved
significantly compared with another similar school in the district: 
Absences declined an average of 1.6 days per student, from 10.5 in
school year 1990-91 to 8.9 in school year 1992-93.\15

The Minneapolis School District reported slight districtwide
improvements in attendance.  Attendance rates increased from 90.5
percent in school year 1993-94 to 90.7 percent in school year
1994-95, according to district officials.  Suspension rates for all
students in school year 1994-95 decreased in 7 of 10 months compared
with school year 1993-94.  The decrease in suspensions, a PSG
contract goal, is a result of a change in how student discipline is
managed, district officials said. 

Our analysis of attendance rates for Baltimore showed that they did
not improve for students in the privately managed schools compared
with the rates for students in the nonprivately managed schools. 
(See app.  II for the details on our Baltimore attendance rate
results.)


--------------------
\14 Evaluation of the Saturn School Project at South Pointe
Elementary School, Dade County Public Schools, Office of Educational
Accountability (Miami, Fla.:  1994). 

\15 According to the Dade County evaluation, these results were found
to be statistically significant.  No follow-up studies have been
conducted to determine if the gains in attendance were sustained. 


   STUDENTS RECEIVED
   INDIVIDUALIZED INSTRUCTION, AND
   TEACHERS RECEIVED TRAINING
---------------------------------------------------------- Chapter 4:2

In Baltimore and Dade County, EAI implemented its Tesseract
instructional approach, resulting in more individualized instruction
and additional teacher training.  For example, in Baltimore, as part
of its Tesseract approach, EAI said it removed desks and replaced
them with tables to facilitate teacher-student and student-student
interaction.  EAI also provided weekly teacher training in its
Tesseract approach, which included training in the Tesseract
philosophy and implementation, instructional technology, and
leadership. 


   STUDENTS HAD GREATER ACCESS TO
   COMPUTERS
---------------------------------------------------------- Chapter 4:3

The Baltimore, Dade County, and Hartford contracts specified that EAI
provide computers.  In Baltimore, for example, EAI leased about 1,100
computers for the nine schools.  A typical classroom had four
networked computers, and each school had one or more computer labs. 
EAI also provided fax machines, copiers, and telephones for teachers
in their classrooms.  In Hartford, EAI's strategy was similar.  EAI
installed computer labs in 5 of Hartford's 32 schools.  EAI also said
that it provided copiers and fax machines for all 32 schools. 


   SCHOOL BUILDING MAINTENANCE AND
   REPAIRS IMPROVED
---------------------------------------------------------- Chapter 4:4

In Baltimore and Hartford, EAI provided school building repair and
maintenance.  In Baltimore, officials told us that the nine schools
were cleaner as well as better maintained.  EAI spent money to paint
the schools, improve heating, and install air conditioning.  In
addition, EAI spent money on bathroom repair and plumbing, fence
repair, landscaping, and preventive maintenance.  The company also
retrofitted lighting in the nine schools, which brightened classrooms
and is expected to reduce energy costs.  In Hartford, EAI completed
several types of repairs throughout the district, concentrating on
six of the district's schools. 


   COMPANIES HAVE BEEN CATALYSTS
   FOR DISTRICTS' RETHINKING THE
   STATUS QUO
---------------------------------------------------------- Chapter 4:5

Private companies have served as catalysts for school districts'
rethinking and challenging the status quo.  In Baltimore, the
superintendent said that the entire school district has become more
competitive because of EAI's presence.  For example, he said that
other schools--concerned that functions, such as maintenance, would
be contracted out--are doing a better job in those areas.  He also
said that the schools are operating in a more business-like manner. 

In Dade County, the district is seeking ways to continue to have
college-educated teaching assistants in classrooms, although the
contract with EAI has expired.  In this case, district officials
sought to continue an EAI innovation. 

In Hartford, some district officials were concerned that teachers'
salaries consumed too much of its budget and asked EAI to help it
negotiate the teachers' contract.  EAI succeeded in helping to secure
a zero increase in the teachers' salaries for 1 year. 

In Minneapolis, the school board hired what it considered to be a
nontraditional superintendent--a private company.  In addition, PSG's
president, who serves as the school district superintendent, does not
have the traditional background in education.  This may change the
way district officials view future candidates for superintendent,
officials stated. 


   IMPACT OF PRIVATE MANAGEMENT ON
   STUDENTS' STANDARDIZED TEST
   SCORES
---------------------------------------------------------- Chapter 4:6

In the Baltimore and Dade County privately managed schools--scores on
standardized achievement tests, a traditional measure of student
learning, showed no improvement when the scores were compared with
those in other comparable schools.  Similarly, in the Minneapolis
district, scores on standardized achievement tests did not improve. 
For Hartford, test score data were not yet available because the
district was still in the early stages of implementation. 

In our own test score analyses of Baltimore schools, we found little
or no difference between scores of students in privately managed
schools compared with students in other similar schools.  In general,
scores tended to be significantly lower in privately managed schools
in the 1992-93 school year, during the year in which private
management was beginning in Baltimore, compared both with school year
1991-92 scores and with scores of other similar schools.  During the
second and third years of private management, however, scores in
privately managed schools increased, so that by the end of the
1994-95 school year, little or no difference remained between scores
of students in privately managed and other similar schools.  (See
app.  II for the details on our Baltimore test score results.)

Dade County's evaluation of EAI's Tesseract instructional approach,
after 2 years, showed no improvement in test scores compared with
another similar Dade County school.\16

Although test scores of students in the privately managed schools
improved, they improved at a rate similar to that of students in the
comparison school.  In Minneapolis, children's test scores showed no
improvement after 2 years under private management.  In fact, the
learning gap between minority and nonminority children widened,
according to the district.  Thus, in Minneapolis, PSG did not achieve
two of its goals--improving test scores of all children and narrowing
test scores between minority and nonminority students. 

Although standardized achievement tests provide a widely accepted
measure of student learning, they do not measure attainment of the
entire range of educational objectives.  Standardized achievement
tests tend to measure core skills and knowledge common to the
curricula of most states, not necessarily the curricula of particular
classrooms or districts and not more abstract forms of learning. 
Thus, such tests may not measure a child's creative thinking, values,
and sense of personal and community responsibility--all qualities
parents and teachers we interviewed mentioned as important
educational goals.  In some districts, some parents and teachers said
they saw positive effects on children in privately managed schools,
even though standardized test score results did not improve. 

In addition, standardized achievement tests tend to be better suited
to measuring the cumulative effects of instruction; thus, it may take
several years before improved learning is reflected in standardized
test scores.  This may partly account for the lack of overall
improvement in test scores in Baltimore, Dade County, and
Minneapolis. 


--------------------
\16 Evaluation of the Saturn School Project at South Pointe
Elementary School. 


GAO'S OBSERVATIONS AND COMMENTS
FROM SCHOOL DISTRICTS AND
COMPANIES
============================================================ Chapter 5

To improve educational outcomes for their students, school districts
in Baltimore, Dade County, Hartford, and Minneapolis contracted with
private, for-profit companies to manage individual schools or entire
school districts.  These contracts reflected considerable variation
in the authority and responsibilities school districts were willing
to give these private companies.  For example, EAI was to focus its
instructional approach in only one school in Dade County but was to
manage the entire Hartford School District. 

Our work suggests that implementing the contracts was more difficult
in some school districts than in others because the companies faced
more community opposition.  The level of opposition appeared to
depend on the extent to which groups in the community perceived
themselves as losing something.  The strongest opposition occurred in
Hartford, where the company was viewed as a vehicle for reducing the
number of teachers.  On the other hand, companies faced almost no
opposition in Dade County, where private management was used in a new
school for which teachers and staff had not been hired, or in
Minneapolis, where the private intervention involved primarily the
superintendent position. 

Although scores on standardized achievement tests did not
substantially improve in the three districts where test scores were
available for analysis, in all four districts, the private management
companies made changes that benefited students.  For example,
students enjoyed individualized instruction, greater access to
computers, and improved building maintenance and repair. 

In the end, three of the four contracts had either expired, been
terminated, or were in the process of being terminated.  In
Baltimore, the cited reason for contract termination was budget
constraints.  In Hartford, the company cited the district's failure
to pay for services rendered in accordance with the contract; the
district cited the unwillingness of the company to operate under the
contract as written as the reason for the impending contract
termination.  The Dade County contract expired at the end of its
5-year term.  The Minneapolis contract remains in force until
December 1996. 


   SCHOOL DISTRICT AND COMPANY
   COMMENTS
---------------------------------------------------------- Chapter 5:1

A draft of this report was provided for review and comment to the
Department of Education as well as the school districts (Baltimore
City, Dade County, Hartford, and Minneapolis) and companies (EAI and
PSG) that were the primary focus of our work on private management of
public schools.  Only the Department of Education declined to provide
comments.  In general, the comments clarified or provided additional
information pertinent to issues discussed in the report. 


   SCHOOL DISTRICTS' COMMENTS
---------------------------------------------------------- Chapter 5:2

Baltimore's Superintendent of Public Instruction stated that he and
his staff had no recommended changes or concerns about the report's
contents.  The superintendent added that much was learned from the
district's experience with private management, although Baltimore's
relationship with EAI had ended.  According to the superintendent,
the district plans to continue to implement EAI strategies and
programs, which he said have proven successful and beneficial to
students. 

The Dade County School Board provided several technical comments that
we incorporated in the report where appropriate. 

The Hartford Board of Education provided its interpretation of the
fundamental structure and essential terms of its private management
contract with EAI.  As noted in our report, the Hartford district and
EAI disagree on the interpretation of many key contract provisions. 

The Hartford School District also commented that neither opposition
nor the district's budgetary problems prevented EAI from implementing
the contract.  We modified the report to reflect the district's view. 
The district also made several technical comments that we
incorporated throughout the report where appropriate. 

Comments from the Minneapolis School District were incorporated in
the comments provided by PSG.  PSG's comments are discussed in the
following paragraphs. 


   PRIVATE MANAGEMENT COMPANIES'
   COMMENTS
---------------------------------------------------------- Chapter 5:3

EAI was particularly concerned that we used spring 1992, rather than
spring 1993, test score results to determine the base year for
measuring the academic performance of Baltimore students based on the
Comprehensive Test of Basic Skills (CTBS).  In explaining its
preference for using spring 1993 test score results, EAI stated that
its first priority after it was awarded the Baltimore contract in
July 1992 was to focus on failures of the physical plant and safety
issues.  According to EAI, initial resistance to staff development
programs also hindered progress in implementing improved
instructional techniques.  In addition, it took until February 1993
to install approximately 1,100 computers in the nine Baltimore
contract schools.  Because the CTBS test was again given in March
1993, EAI believes that the spring 1993 testing date is a far more
accurate baseline date than is March 1992. 

We disagree with EAI's position on this issue and believe that 1992
is the appropriate baseline year.  Using spring 1992 scores as a
baseline provides a benchmark of student performance immediately
before EAI's assuming management in the fall of 1992.  Our goal was
to assess changes in student performance that occurred while these
schools were under EAI management.  Using 1993 scores as the baseline
ignores the impact of the first year of EAI management and thus fails
to assess the entire experience. 

EAI also maintained that we failed to note the results of the
Maryland State Performance Assessment Program (MSPAP) in our report. 
MSPAP is a test that measures student performance on tasks that
require critical thinking, high-level problem solving, and the
integration of knowledge.  Our analyses do include MSPAP test scores
to the extent they were available, and these results are discussed in
appendix II of the report.  EAI also made several technical comments
that we incorporated in the report where appropriate. 

In PSG's comments, which also included comments from the Minneapolis
School District, the company discussed the issue of accountability. 
In PSG's view, its willingness to be accountable for the results it
produces for the entire school district is the key to building
accountability throughout the rest of the local system.  In this
regard, the company believes accountability must be adopted by
stakeholders throughout the educational system, including teachers in
their classrooms, principals in their schools, and parents and
community members.  Accountability throughout the system is
imperative, PSG believes, for students to succeed. 

PSG also provided technical comments on the information contained in
our draft report that we incorporated as appropriate.  The most
significant point on which we disagree with PSG concerns payment for
superintendent services.  Our report states that PSG was to receive a
monthly fee of $5,000 for serving as superintendent.  PSG said that,
instead, the fee for superintendent services is paid on a
pay-for-performance basis; that is, it would only be paid if the
district were satisfied with PSG's progress as superintendent. 
Because we found no such qualifications in our analysis of the PSG
contract, we did not change that statement in our report. 


METHODOLOGY USED IN BALTIMORE TEST
SCORE AND ATTENDANCE ANALYSES
=========================================================== Appendix I

This appendix discusses the methodology used in analyzing
standardized test score and attendance data of seven privately
managed schools in the Baltimore City Public School District.\17 It
also presents general limitations of the study and defines terms used
in the report. 

A number of individuals with educational, statistical, or
methodological expertise assisted us in various portions of our test
score and attendance analyses.  (See app.  IX for the names and
affiliations of these people.)


--------------------
\17 We limited our analysis to the seven elementary schools that
included kindergarten to fifth grades because more test score data
were available for this group and the pool from which to select a
comparison school was larger.  (See table III.1 in app.  III for a
list of these schools.)


   METHODOLOGY
--------------------------------------------------------- Appendix I:1

Using student data provided by the district's Accountability Office
for the school year before private management (1991-92), we
statistically matched each privately managed school with a similar
district-managed school.  Students in the latter schools served as a
comparison group for two types of analyses:  (1) a longitudinal
analysis, which analyzed test scores and attendance of students who
remained in the same school for the entire 3-year period (1992-93
through 1994-95), and (2) a cross-sectional analysis, which analyzed
test scores and attendance of students on a year-by-year basis
without regard to changes in student population from one year to the
next.  For all statistical tests, we rejected the null hypotheses if
the observed significance levels were less than .05. 

We limited our test score analyses to the areas of reading and math. 
We extracted from the district's data files reading and math scores
as dependent measures from routinely administered achievement and
performance tests.  For the 1991-92 through 1994-95 school years,\18
we used normal curve equivalent (NCE) scores for reading and math
from the Comprehensive Test of Basic Skills (CTBS).  In addition, for
students in third and fifth grades, we used reading and math outcome
scale scores from the Maryland State Performance Assessment Program
(MSPAP) for school years 1992-93 and 1993-94.\19

We used the number of days the student was absent for the year as the
dependent variable in the attendance analyses.  We standardized this
variable by taking the ratio of days absent to days on roll and
projecting number of days absent to a 180-day school year. 

We included in our analyses students enrolled in the selected
privately managed or comparison schools who met the following
criteria:  (1) they had been on the rolls of the school at least 170
days of the school year and (2) they were not identified as receiving
special education services greater than intensity level 3.\20
Students were excluded from test score analyses if they had no score
on the particular outcome (for example, CTBS reading score) used in
the analysis. 


--------------------
\18 We used 1991-92 data in selecting the comparison schools and as
covariates in the longitudinal analyses. 

\19 Data were available for MSPAP only for school years 1992-93 and
1993-94.  Only students in third, fifth, and eighth grades
participate in the assessment. 

\20 See "Definitions of Terms" in this app.  for an explanation of
intensity level 3. 


      SELECTING THE COMPARISON
      SCHOOLS
------------------------------------------------------- Appendix I:1.1

We used the Statistical Package for the Social Sciences k-means
clustering procedure to match each privately managed school with a
similar district-managed school.  Matching was done on the basis of
the following school characteristics:  enrollment, attendance rate,
promotion rate, racial composition of student body, proportion of
students receiving free/reduced lunch, proportion of special
education students, mobility rate, and CTBS scores.  The clustering
procedure compared the characteristics of the schools and calculated
a "distance" measure to represent the similarity between the cluster
center (that is, the privately managed school) and each of the other
schools.  We selected as the comparison school the school identified
as being closest (that is, with the smallest distance measure) to the
privately managed school. 


      THE LONGITUDINAL ANALYSIS
------------------------------------------------------- Appendix I:1.2

The longitudinal analysis assessed test scores and attendance of
students enrolled in a privately managed or comparison school in the
1992-93 school year and who remained in that school through the
1994-95 school year.  For these analyses, we combined school
populations to form a single group of students from privately managed
schools and a single group of students from comparison schools. 

We performed separate analyses for reading, math, and attendance. 
The groups for the reading analysis consisted of 689 students in the
privately managed group and 733 in the comparison group.  For the
math analysis, the privately managed group had 470 students, and the
comparison group had 488 students.  The attendance analysis consisted
of 870 privately managed school students and 855 comparison students. 

We used a repeated measure analysis of variance (ANOVA) technique to
provide the overall test of group effect for the longitudinal
analyses.  The repeated measure ANOVA procedure produces an omnibus
test of significance that adjusts for the increased probability of
Type I error (that is, concluding that the groups in question are
different when in reality they are not) inherent in multivariate
designs. 

Although schools were matched on the basis of data from the school
year immediately preceding private management (1991-92), attrition
that occurred in school years 1992-93 through 1994-95 threatened the
equivalency of the surviving student groups.  To mitigate this
threat, we used 1991-92 data as covariates in each of the
longitudinal analyses to provide statistical adjustment for
nonequivalence introduced by attrition.  Since the covariate adjusted
for main effects only, means used in discussion of year-to-year
changes do not reflect this adjustment. 

Because longitudinal analysis focuses on changes in an individual
student's performance over time, it is well suited for assessing
educational interventions that may impact outcome measures slowly or
over a longer period of time.  However, because the analysis included
only that subset of students who remained in the same school for the
3-year period, results are not necessarily generalizable to the
relatively large proportion of the district's students who are more
mobile and may change schools one or more times during a school year. 


      THE CROSS-SECTIONAL ANALYSIS
------------------------------------------------------- Appendix I:1.3

The cross-sectional analysis compared student test scores and
attendance for each year of the private management contract (1992-93
through 1994-95), without regard to changes in the student population
from year to year. 

We performed cross-sectional test score analyses at the school level. 
One-way ANOVA was used to compare reading and math scores of students
in each pair of schools.  We used means and variances from these
analyses to compute effect sizes, which were then cumulated to
provide summary findings.  (See tables III.3, III.5, III.7, and III.9
in app.  III for group sizes used in these analyses.)

To perform the cross-sectional attendance analyses, we combined
school populations to form a single group of students from privately
managed schools and a single group of students from comparison
schools.  (See table III.11 in app.  III for group sizes for these
analyses.)

Cross-sectional analysis does not take into account changes in
individual students' scores.  Instead, it compares the averages of
students in the two matched schools for each year under study. 
Cross-sectional analysis is more sensitive to detecting short-term
effects of educational changes and is commonly employed to provide
descriptive information about school performance. 


      INTERPRETATION AND
      COMPUTATION OF EFFECT SIZES
------------------------------------------------------- Appendix I:1.4


         INTERPRETATION OF EFFECT
         SIZES
----------------------------------------------------- Appendix I:1.4.1

Effect sizes (referred to as "d" in the computation formula) are
measures of the overlap between the distributions of the experimental
and control groups when the underlying scale has been standardized so
that  equals 1.  Thus, in this study, d can be translated
into the proportion of comparison school scores that are less than
(or more than) the average score in the privately managed school.\21

An effect size of zero means that the average score in the comparison
school was equal to the average score in the privately managed
school, or, alternatively, the scores of 50 percent of students in
the comparison school were higher than the average score in the
privately managed school.  A negative effect size means that the
scores of more than 50 percent of students in the comparison school
were higher than the average student in the privately managed school;
a positive effect means that fewer than 50 percent of students in the
comparison group scored better than the average student in the
privately managed group. 

As with other estimates, an effect size must be interpreted on the
basis of two factors:  (1) the confidence with which the estimate is
made and (2) the practical importance of an effect of this size. 

Tables III.2 through III.9 in appendix III show the computed effect
sizes and variances.  Indications of statistical significance were
based on a 95-percent confidence interval. 

Once statistical significance of the estimate has been established,
the issue of practical importance must be considered.  Judgments
about practical importance are typically made relative to the
experimental context and the standards that have been established in
that particular field.  In educational research, effect sizes of .5
have been offered as a conventional measure of practical
significance.\22 The National Institute of Education's Joint
Dissemination Review Panel suggested that usually an effect size of
.33, but at times as small as .25, may be considered educationally
significant.\23 To recognize relatively small effects, we chose to
target our discussions to effects of .25 or greater. 


--------------------
\21 L.V.  Hedges and I.  Olkin, Statistical Methods for Meta-Analysis
(San Diego:  Academic Press, Inc., 1985), p.  76. 

\22 P.  Rossi and S.  Wright, "Evaluation Research:  An Assessment of
Theory, Practice and Politics," Evaluation Quarterly, Vol.  I (1977),
pp.  5-52. 

\23 G.K.  Tallmadge, The Joint Dissemination Review Panel Ideabook
(Washington, D.C.:  National Institute of Education and U.S.  Office
of Education, 1977). 


         COMPUTATION OF EFFECT
         SIZES
----------------------------------------------------- Appendix I:1.4.2

To aid in interpreting findings and aggregating results, effect sizes
were computed for each finding\24 using the formula



where Y\E is the mean of the experimental group (privately managed
schools), Y\C is the mean of the control group (comparison schools),
and s is the pooled standard deviation.  Variances were calculated
for the individual effects by the following formula: 



Effect sizes were cumulated using the following method that gives
greater weight to those individual effects having lower variances: 



Confidence intervals for effect sizes were computed using the
following standard formulas: 






--------------------
\24 For more information, see Hedges and Olkin, pp.  75-106. 


   GENERAL LIMITATIONS OF THE
   STUDY
--------------------------------------------------------- Appendix I:2

Assessing the impact of program change in a field setting is
extremely difficult.  Any study that attempts to do so encounters
problems such as the following: 

  -- No random assignment to groups.  We could not randomly assign
     students to a treatment or control group to ensure initial
     equivalency of the groups.  We attempted to adjust for this
     shortcoming by selecting comparison schools matched as closely
     as possible on relevant characteristics and by using statistical
     techniques (for example, covariate analysis) to control for
     initial differences on key characteristics. 

  -- Lack of experimental control.  The independent variable could
     not be well defined and may have differed from school to school. 
     The independent variable in this study was the type of school
     management, privately managed or district managed.  However,
     this variable was operationalized in somewhat different fashion
     from school to school as teachers and principals implemented the
     Tesseract approach under somewhat different circumstances and
     with varying levels of enthusiasm. 

For example, some interviewees told us that veteran teachers tended
to resist implementing the Tesseract approach.  On the other hand, a
number of the newer teachers we interviewed were enthusiastic about
it.  We were told in one school that teachers were given a great deal
of flexibility in deciding which parts of the approach to use in
their classrooms.  Thus, a great deal of variability may have existed
in the operationalization of private management from school to
school. 

  -- Attrition.  Students in these schools were highly mobile, so
     loss of subjects affected these analyses.  The long-term effects
     of program change may take several years for student scores to
     fully reflect them and so may be difficult to measure with
     highly mobile students, as are students in Baltimore City Public
     Schools. 

  -- Measurement difficulties.  Standardized achievement tests, such
     as the ones used to provide the dependent variables in these
     analyses, are widely accepted as measures for use in educational
     evaluations; however, they have shortcomings when used for this
     purpose.  One of the major criticisms of using standardized
     tests for evaluating educational programs is that they may not
     be capable of measuring mastery of the specific objectives of
     the instructional program being evaluated; rather, they measure
     more general skills and knowledge.  Thus, standardized tests may
     be relatively insensitive to detecting specific gains,
     especially in the short term. 

Furthermore, school personnel and parents associated with schools in
this study mentioned broad goals as some of the desired outcomes of
educational changes.  For example, they wanted students to become
better citizens, become self-directed learners, and show improved
general problem-solving abilities.  In general, standardized test
scores are not intended for, nor well suited to, measuring these
broader goals. 


   DEFINITIONS OF TERMS
--------------------------------------------------------- Appendix I:3


      COMPREHENSIVE TEST OF BASIC
      SKILLS (CTBS)
------------------------------------------------------- Appendix I:3.1

A standardized achievement test given annually to all students in
kindergarten through fifth grade in the Baltimore City Public School
District that measures basic skills in reading, language, spelling,
math, study skills, science, and social studies. 


      DAYS ABSENT
------------------------------------------------------- Appendix I:3.2

A variable constructed from student attendance records to standardize
the attendance base to a full school year.  We calculated an absentee
rate by dividing days absent by days on roll, then projected this
rate to a 180-day school year basis. 


      MARYLAND STATE PERFORMANCE
      ASSESSMENT PROGRAM (MSPAP)
------------------------------------------------------- Appendix I:3.3

Assessments given in third, fifth, and eighth grades in reading,
writing, language usage, math, social studies, and science.  These
tests measure student performance on tasks that require critical
thinking, high-level problem solving, and the integration of
knowledge. 


      MSPAP OUTCOME SCALE SCORE
------------------------------------------------------- Appendix I:3.4

A version of the MSPAP outcome score described in the 1994 Score
Interpretation Guide for MSPAP as "directly comparable across
outcomes in the same content area, across years, and to the MSPAP
proficiency levels.  These scores are expressed on the MSPAP scale
score scale and range, as do the content area scale scores, from 350
to 700."


      NORMAL CURVE EQUIVALENT
      (NCE)
------------------------------------------------------- Appendix I:3.5

A normalized version of a test score often used for federal reporting
purposes.  These scores have a mean of 50 and a standard deviation of
about 21.  The values of 1, 50, and 99 on this scale coincide with
percentile ranks of 1, 50, and 99; but other values do not coincide. 


      SPECIAL EDUCATION, INTENSITY
      LEVEL 3
------------------------------------------------------- Appendix I:3.6

One of six levels describing intensity of special education services
available.  Students classified as Level 3 may receive special
educational services not to exceed an average of 3 hours per school
day. 


RESULTS OF BALTIMORE TEST SCORE
AND ATTENDANCE ANALYSES
========================================================== Appendix II

This appendix discusses the results of our analyses of achievement
and attendance data conducted on selected privately managed schools
in the Baltimore City Public School District.\25 (For a discussion of
the methodology used in these analyses, see app.  I.)


--------------------
\25 The University of Maryland, Baltimore County, conducted an
evaluation of the privately managed schools in Baltimore that
included test score and attendance analyses.  Those findings
generally agreed with the results presented here. 


   TEST SCORE ANALYSES
-------------------------------------------------------- Appendix II:1

Our test score analyses of Baltimore's schools indicated that,
overall, test scores of students attending privately managed schools
were similar to those of students attending the matched,
district-managed, comparison schools.  Although schools varied in
some of the analyses, overall, we found little difference between
test scores in privately managed and comparison schools. 


      LONGITUDINAL ANALYSIS
------------------------------------------------------ Appendix II:1.1

The longitudinal analysis, which tracked the group of students who
remained in the same school during the 3-year period (1992-93 through
1994-95), showed overall test score results in reading and math to be
similar between students in privately managed and comparison schools
after adjusting for the presence of initial difference.\26 Reading
and math scores for both groups declined over the 3-year period
compared with 1991-92 scores.  For students in privately managed
schools, these scores sharply declined in 1992-93 (the implementation
year), followed by increases the next 2 years.  Scores of students in
comparison schools, on the other hand, declined less sharply in
1992-93 but continued to decline in each of the next 2 years.  Figure
II.1 shows that reading scores showed this pattern.\27

   Figure II.1:  Longitudinal
   Analysis:  CTBS Reading Scores

   (See figure in printed
   edition.)

Note:  Means reported in this graph illustrate changes in average
scores over time.  They have not been adjusted for variation
accounted for by 1992 test scores.  After adjusting for this initial
difference, no significant difference was found between scores of
students in the privately managed group and scores of students in the
comparison group. 

Although year-to-year changes were sometimes significantly different
between the privately managed and comparison groups, offsetting
patterns of increase and decline resulted in an overall finding of no
statistically significant difference in reading after adjusting for
the presence of initial difference.  As figure II.2 shows, the
pattern of change in math scores from year to year was similar to
that found in the reading scores.  Although both groups declined in
the first year, the decline in the privately managed schools was
greater.  However, in the following years, scores in the privately
managed schools increased, while the scores in the comparison group
continued to decline. 

   Figure II.2:  Longitudinal
   Analysis:  CTBS Math Scores

   (See figure in printed
   edition.)

Note:  Means reported in this graph illustrate changes in average
scores over time.  They have not been adjusted for variation
accounted for by 1992 test scores.  After adjusting for this initial
difference, the overall decline in scores was somewhat greater for
the privately managed group. 

Unlike reading scores, however, when the 3 years are considered
together, the decline in math scores in the privately managed group
was statistically significant, even though the 1994-95 scores of the
privately managed group were slightly higher than the comparison
group's.  This finding of overall difference between the two groups
may be accounted for largely by the substantial decline in scores in
the privately managed group in 1992-93, the implementation year. 

Figures II.3 and II.4 depict the pattern of change over the years
expressed in change from the groups' 1991-92 averages.  The initial
decline both in reading and math scores is apparent in these figures. 
For both reading and math, after this initial large decline, scores
in the privately managed group show a pattern of steady improvement. 

   Figure II.3:  Changes in CTBS
   Reading Scores (Longitudinal
   Analysis)

   (See figure in printed
   edition.)

   Figure II.4:  Changes in CTBS
   Math Scores (Longitudinal
   Analysis)

   (See figure in printed
   edition.)


--------------------
\26 Although schools were matched for this analysis on the basis of
1991-92 characteristics, the equivalency of the groups was affected
by attrition over the 3-year period.  To compensate for this fact, we
used 1991-92 test score and attendance data as a statistical control
for this inequality. 

\27 Means for individual years shown in figs.  II.1 and II.2 have not
been adjusted for variance accounted for by 1991-92 scores. 


      CROSS-SECTIONAL ANALYSES
------------------------------------------------------ Appendix II:1.2

Results of the cross-sectional analyses were similar to those of the
longitudinal analysis.  Overall, we found little difference in
student achievement in privately managed and comparison schools. 
When MSPAP scores were used as the outcome measure, no significant
difference was found between the privately managed group of students
and the comparison group in either reading or math scores.  (Tables
III.2 and III.4 in app.  III show school-level and overall effects
sizes\28 for the MSPAP analyses.)

Tables III.6 and III.9 (in app.  III) show similar findings using
CTBS scores as the outcome measure.  Although small, the cumulated
effects for CTBS scores in both reading and math were statistically
significant in favor of the comparison group.  However, effects of
this small size are not conventionally considered to have practical
educational significance.  (See app.  I for additional discussion of
interpreting effect sizes.)

Because the cross-sectional analyses were done at the school level,
these analyses revealed the variation in effect among pairs of
schools.  In some pairs, students in the privately managed school
performed better; in others, students in the comparison school
performed better.  In most cases, the effects were relatively small. 

Three of the seven privately managed schools (Dr.  Rayner Browne,
Harlem Park Elementary, and Sarah M.  Roach), however, show a more
definite pattern of underperformance compared with their matched
schools.  In these cases, effect sizes were great enough to warrant
further attention.  Determining whether this underperformance is an
artifact of the comparison or a result of some change in the
privately managed school or in the comparison school is beyond the
scope of this study. 


--------------------
\28 Effect sizes are measures of the overlap between the
distributions of the experimental and control groups when the
underlying scale has been standardized so that  equals 1. 
See app.  I for additional information on interpreting effect size. 


   ATTENDANCE ANALYSES
-------------------------------------------------------- Appendix II:2

Overall, our analyses of student attendance data showed little
difference between attendance patterns of students in privately
managed schools and students in comparison schools.  We found no
difference in the number of days absent for nonmobile students (those
remaining in the same school for the 3 years) in the longitudinal
analysis and little difference when the entire student population was
considered in the cross-sectional analysis. 


      LONGITUDINAL ANALYSES
------------------------------------------------------ Appendix II:2.1

Our longitudinal analysis of attendance data revealed that attendance
patterns of nonmobile students attending privately managed schools
were similar to attendance patterns of nonmobile students attending
comparison schools.  When we analyzed attendance of nonmobile
students in privately managed and comparison schools, we found no
significant difference between the groups.  Both groups showed a
general pattern of improved attendance for the period.  (Table III.10
in app.  III presents these results.)


      CROSS-SECTIONAL ANALYSIS
------------------------------------------------------ Appendix II:2.2

Our year-by-year analysis of the period also showed a general trend
toward improved attendance for both privately managed and comparison
groups.  As table III.11 (in app.  III) shows, no significant
difference existed in attendance rates of students attending
privately managed schools and students attending comparison schools
in 1992-93, the first year of private management.  However, in
1993-94 students in the privately managed schools were absent, on
average, about 3 days per year more than students in the comparison
schools.  In 1994-95, this difference declined to 1 day a year. 
Findings for both 1993-94 and 1994-95 were statistically significant. 


DATA SUPPORTING THE BALTIMORE TEST
SCORE AND ATTENDANCE ANALYSES
========================================================= Appendix III



                        Table III.1
          
              Pairs of Schools in the Analyses

                Privately managed
                school                Comparison school
--------------  --------------------  --------------------
Pair 1          Dr. Rayner Browne     Furman L. Templeton

Pair 2          Mildred Monroe        Park Heights

Pair 3          Harlem Park           Pimlico
                Elementary

Pair 4          Edgewood              Hilton

Pair 5          Sarah M. Roach        James Mosher

Pair 6          Mary E. Rodman        Liberty

Pair 7          Graceland Park-       Charles Carroll,
                O'Donnell Heights     Barrister
----------------------------------------------------------


                                            Table III.2
                              
                               Means and Effect Sizes, MSPAP Reading
                                               Scores


                                                School                             School
                                                effect                             effect  Cumulate
School                 Grade 3     Grade 5        size     Grade 3     Grade 5       size  d effect
------------------  ----------  ----------  ----------  ----------  ----------  ---------  --------
Dr. Rayner Browne          444         469        -.03         451         458       +.23      +.10
Furman L.                  447         469                     430         457
 Templeton
Mildred Monroe             456         480        +.19         431         487       -.24         0
Park Heights               450         470                     465         481
Harlem Park                434         469        -.21         443         446     -.43\a      -.33
 Elementary
Pimlico                    461         465                     467         465
Edgewood                   464         487        +.04         473         474       -.11      -.04
Hilton                     468         479                     483         474
Sarah M. Roach             474         488        -.09         456         484       -.24      -.17
James Mosher               483         485                     491         475
Mary E. Rodman             450         465        +.11         456         456       -.26      -.07
Liberty                    451         453                     462         476
Graceland Park-            455         469        -.37         453         459       -.22      -.29
 O'Donnell Heights
Charles Carroll,           487         473                     459         474
 Barrister
Overall effect                                                                                 -.12
---------------------------------------------------------------------------------------------------
Note:  Effect sizes are measures of the overlap between the
distributions of the experimental and control groups when the
underlying scale has been standardized so that  =1.  See
app.  I for additional information on interpretation of effect sizes. 

\a Significantly different from 0 to .05 level. 



                                       Table III.3
                         
                         Ns and Variances for Effect Sizes, MSPAP
                                      Reading Scores


School                \2             N   \2             N   \2
-------------------  ------------  ------------  ------------  ------------  ------------
Dr. Rayner Browne            .027            63          .025            69          .052
Furman L. Templeton                          94                          98
Mildred Monroe               .031            56          .037            54          .068
Park Heights                                 75                          58
Harlem Park                  .019            88          .015           128          .034
 Elementary
Pimlico                                     139                         156
Edgewood                     .015           110          .014           128          .030
Hilton                                      165                         154
Sarah M. Roach               .023           107          .021           101          .044
James Mosher                                 75                          97
Mary E. Rodman               .012           188          .013           184          .025
Liberty                                     146                         133
Graceland Park-              .029            80          .028            84          .057
 O'Donnell Heights
Charles Carroll,                             66                          63
 Barrister
Overall effect                                                                       .006
-----------------------------------------------------------------------------------------


                                       Table III.4
                         
                            Means and Effect Sizes, MSPAP Math
                                          Scores


                                           Effect                        Effect  Cumulate
School                Grade 3   Grade 5      size   Grade 3   Grade 5      size  d effect
-------------------  --------  --------  --------  --------  --------  --------  --------
Dr. Rayner Browne         450       470    +.49\a       473       472    +.74\a    +.61\a
Furman L. Templeton       422       456                 424       447
Mildred Monroe            450       462      +.33       432       487      -.01      +.17
Park Heights              418       464                 474       461
Harlem Park               434       467   -0.62\a       446       448    -.47\a     -54\a
 Elementary
Pimlico                   469       490                 475       462
Edgewood                  483       487    +.66\a       493       458      +.05      +.35
Hilton                    448       464                 485       461
Sarah M. Roach            489       469    -.39\a       453       457    -.87\a    -.63\a
James Mosher              484       513                 512       474
Mary E. Rodman            437       449      +.22       454       461      -.16      +.03
Liberty                   437       436                 487       448
Graceland Park-
 O'Donnell Heights        452       460     -0.32       456       444     -0.30      -.31
Charles Carroll,          480       462                 477       468
 Barrister
Overall effect                                                                       -.05
-----------------------------------------------------------------------------------------
Note:  Effect sizes are measures of the overlap between the
distributions of the experimental and control groups when the
underlying scale has been standardized so that  = 1.  See
app.  I for additional information on interpretation of effect sizes. 

\a Significantly different from 0 at .05 level. 



                                       Table III.5
                         
                         Ns and Variances for Effect Sizes, MSPAP
                                       Math Scores


               \
School                  2               N     \2               N     \2
-------------  ----------  --------------  --------------  --------------  --------------
Dr. Rayner           .029              57            .031              58            .060
 Browne
Furman L.                              98                              86
 Templeton
Mildred              .034              53            .040              50            .074
 Monroe
Park Heights                           73                              57
Harlem Park          .018             103            .017             109            .035
 Elementary
Pimlico                               137                             136
Edgewood             .016             105            .016             113            .032
Hilton                                168                             144
Sarah M.             .026             106            .026              85            .051
 Roach
James Mosher                           68                              89
Mary E.              .013             181            .013             179            .026
 Rodman
Liberty                               145                             131
Graceland            .030              75            .027              76            .057
 Park-
 O'Donnell
 Heights
Charles                                63                              53
 Carroll,
 Barrister
Overall                                                                              .006
 effect
-----------------------------------------------------------------------------------------


                                         Table III.6
                           
                             Means and Effect Sizes, CTBS Reading
                                            Scores


                                     Effect              Effect              Effect  Cumulate
School                       Mean      size      Mean      size      Mean      size  d effect
-------------------  ------------  --------  --------  --------  --------  --------  --------
Dr. Rayner Browne            28.3    -.45\a      32.3    -.21\a      30.4    -.43\a    -.36\a
Furman L. Templeton          37.0                36.7                38.9
Mildred Monroe               36.0      -.05      43.2      +.07      42.6      -.05      -.01
Park Heights                 36.8                41.9                43.6
Harlem Park                  32.2    -.54\a      30.3    -.60\a      37.2      -.06    -.41\a
 Elementary
Pimlico                      42.4                42.0                38.6
Edgewood                     49.0      +.03      47.5      -.09      47.5      +.14      +.02
Hilton                       48.5                49.4                44.7
Sarah M. Roach               41.6    -.30\a      38.2    -.52\a      42.4    -.48\a    -.43\a
James Mosher                 47.6                48.6                53.3
Mary E. Rodman               41.0      -.02      38.8       .00      45.3      +.10      +.02
Liberty                      41.3                38.9                43.1
Graceland Park-              35.7       .00      33.6    -.34\a      35.8    -.22\a   -0.19\a
 O'Donnell Heights
Charles Carroll,             35.6                39.8                40.2
 Barrister
Overall effect                                                                         -.17\a
---------------------------------------------------------------------------------------------
Note:  Effect sizes are measures of the overlap between the
distributions of the experimental and control groups when the
underlying scale has been standardized so that  =1.  See
app.  I for additional information on interpreting effect sizes. 

\a Significantly different from 0 at .05 level. 



                                       Table III.7
                         
                         Ns and Variances for Effect Sizes, CTBS
                                      Reading Scores


                     \2         N       >\2         N       >\2         N       >\2
-------------------  --------  --------  --------  --------  --------  --------  --------
Dr. Rayner Browne        .008       216      .007       252      .009       219      .003
Furman L. Templeton                 315                 330                 269
Mildred Monroe           .009       193      .010       180      .010       186      .003
Park Heights                        302                 235                 212
Harlem Park              .005       406      .005       435      .005       366      .002
 Elementary
Pimlico                             470                 475                 452
Edgewood                 .005       403      .005       386      .005       347      .002
Hilton                              498                 474                 449
Sarah M. Roach           .006       353      .006       346      .006       313      .002
James Mosher                        311                 311                 330
Mary E. Rodman           .004       587      .004       560      .004       536      .001
Liberty                             478                 436                 412
Graceland Park-          .008       280      .007       321      .008       321      .003
 O'Donnell Heights
Charles Carroll,                    235                 235                 219
 Barrister
Overall effect                                                                       .000
-----------------------------------------------------------------------------------------


                                       Table III.8
                         
                         Means and Effect Sizes, CTBS Math Scores


                                 Effect              Effect              Effect  Cumulate
School                   Mean      size      Mean      size      Mean      size  d effect
-------------------  --------  --------  --------  --------  --------  --------  --------
Dr. Rayner Browne        28.2    -.42\a      32.0      -.16     32.02    -.41\a    -.32\a
Furman L. Templeton      36.9                35.1                41.8
Mildred Monroe           39.2      +.12      46.7      +.17      47.5    +.23\a    +.17\a
Park Heights             36.7                43.2                42.7
Harlem Park              32.4    -.50\a      37.2    -.30\a      37.9      -.04    -.28\a
 Elementary
Pimlico                  43.0                44.0                38.7
Edgewood                 49.5      +.04      51.4      +.12      51.3    +.27\a    +.14\a
Hilton                   48.6                48.6                45.0
Sarah M. Roach           37.7    -.51\a      39.0    -.38\a      48.8    -.12\a    -.34\a
James Mosher             48.3                47.2                51.5
Mary E. Rodman           38.3      -.01      39.5      +.12      47.4    +.27\a     +12\a
Liberty                  38.6                36.9                41.0
Graceland Park-          36.8      +.11      35.4    -.17\a      40.6    -.09\a      -.05
 O'Donnell Heights
Charles Carroll,         34.5                38.9                42.4
 Barrister
Overall Effect                                                                     -.08\a
-----------------------------------------------------------------------------------------
Note:  Effect sizes are measures of the overlap between the
distributions of the experimental and control groups when the
underlying scale has been standardized so that  =1.  See
app.  I for additional information on interpreting effect sizes. 

\a Significantly different from 0 at .05 level. 



                                       Table III.9
                         
                         Ns and Variances for Effect Sizes, CTBS
                                       Math Scores


                     \2         N       >\2         N       >\2         N       >\2
-------------------  --------  --------  --------  --------  --------  --------  --------
Dr. Rayner Browne        .010       175      .009       200      .010       173      .003
Furman L. Templeton                 260                 280                 233
Mildred Monroe           .010       166      .012       152      .012       154      .004
Park Heights                        249                 196                 171
Harlem Park              .006       338      .005       351      .006       300      .002
 Elementary
Pimlico                             399                 399                 377
Edgewood                 .005       341      .006       321      .006       291      .002
Hilton                              431                 410                 379
Sarah M. Roach           .008       302      .007       291      .007       275      .003
James Mosher                        251                 261                 264
Mary E. Rodman           .005       484      .005       471      .005       437      .002
Liberty                             388                 367                 340
Graceland Park-          .009       232      .009       249      .009       271      .003
 O'Donnell Heights
Charles Carroll,                    195                 191                 176
 Barrister
Overall effect                                                                       .000
-----------------------------------------------------------------------------------------


                                       Table III.10
                         
                           Days Absent (Longitudinal Analysis)

                                    1991-92        1992-93        1993-94        1994-95
----------------------------  -------------  -------------  -------------  -------------
Students in privately                  13.8           12.1            9.5            9.7
 managed schools (N=870)
Students in comparison                 13.4           13.0            8.6            9.4
 schools (N=855)
----------------------------------------------------------------------------------------
Note:  Significance test of overall difference not significant at .05
level. 



                                       Table III.11
                         
                          Days Absent (Cross-Sectional Analysis)

                                    1991-92        1992-93        1993-94        1994-95
----------------------------  -------------  -------------  -------------  -------------
Students in privately                  14.0           14.4         13.0\a         12.9\a
 managed schools                  (N=2,425)      (N=3,118)      (N=3,127)      (N=2,961)
Students in comparison                 14.0           14.9           10.2           11.8
 schools                          (N=2,490)      (N=3,242)      (N=3,103)      (N=3,049)
----------------------------------------------------------------------------------------
\a Differences found in 1993-94 and in 1994-95 were statistically
significant at .05 level. 




(See figure in printed edition.)Appendix IV
COMMENTS FROM THE BALTIMORE CITY
PUBLIC SCHOOL DISTRICT
========================================================= Appendix III




(See figure in printed edition.)Appendix V
COMMENTS FROM DADE COUNTY PUBLIC
SCHOOLS
========================================================= Appendix III




(See figure in printed edition.)Appendix VI
COMMENTS FROM THE HARTFORD SCHOOL
DISTRICT
========================================================= Appendix III



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)




(See figure in printed edition.)Appendix VII
COMMENTS FROM EAI
========================================================= Appendix III



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)




(See figure in printed edition.)Appendix VIII
COMMENTS FROM THE MINNEAPOLIS
SCHOOL DISTRICT AND PSG
========================================================= Appendix III



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)


ACKNOWLEDGMENTS
========================================================== Appendix IX

We acknowledge the assistance of the following experts:  Dr.  Jack
Edwards, Defense Manpower Data Center; Dr.  John A.  Nunnery,
Director, Bureau of Educational Research, College of Education at the
University of Memphis; Dr.  Nambury Raju, Professor of Psychology at
the Georgia Institute of Technology; and Dr.  Randall R.  Robey,
Assistant Professor, Curry School of Education at the University of
Virginia.  These individuals provided valuable assistance for our
test score and attendance analyses; however, they do not necessarily
endorse the positions taken in the report. 


GAO CONTACTS AND STAFF
ACKNOWLEDGMENTS
=========================================================== Appendix X

GAO CONTACTS

Fred Yohey, Assistant Director, (202) 512-7218
Sherri Doughty, Project Manager, (202) 512-7273

ACKNOWLEDGMENTS

In addition to those named above, the following individuals made
important contributions to this report:  Lemuel Jackson coauthored
the report and led the Minneapolis case study, Patricia Cazares
coauthored the report and led the Hartford and Dade County case
studies, Benjamin Jordan coauthored the report and assisted with the
case studies, Deborah McCormick led the Baltimore test score and
attendance analysis and assisted with the case studies, Susan Poling
led the contract analysis and assisted with the case studies, Louise
DiBenedetto and Suzanne Murphy led the Baltimore expense analysis,
and Laurel Rabin and Elizabeth Williams assisted in writing the
report. 


*** End of document. ***