No Child Left Behind Act: Assistance from Education Could Help	 
States Better Measure Progress of Students with Limited English  
Proficiency (26-JUL-06, GAO-06-815).				 
                                                                 
For the Spanish translation of the highlights page for this	 
document, see GAO-06-1111. Ley para que ningun nino se quede	 
atras: La ayuda del Departamento de Educacion puede contribuir a 
que los Estados midan mejor el progreso de los alumnos que no	 
dominan bien el ingles. GAO-06-1111, Julio de 2006. The No Child 
Left Behind Act of 2001 (NCLBA) focused attention on the academic
achievement of more than 5 million students with limited English 
proficiency. Obtaining valid test results for these students is  
challenging, given their language barriers. This report describes
(1) the extent to which these students are meeting annual	 
academic progress goals, (2) what states have done to ensure the 
validity of their academic assessments, (3) what states are doing
to ensure the validity of their English language proficiency	 
assessments, and (4) how the U.S. Department of Education	 
(Education) is supporting states' efforts to meet NCLBA's	 
assessment requirements for these students. To collect this	 
information, we convened a group of experts and studied five	 
states (California, Nebraska, New York, North Carolina, and	 
Texas). We also conducted a state survey and reviewed state and  
Education documents.						 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-06-815 					        
    ACCNO:   A57487						        
  TITLE:     No Child Left Behind Act: Assistance from Education Could
Help States Better Measure Progress of Students with Limited	 
English Proficiency						 
     DATE:   07/26/2006 
  SUBJECT:   Academic achievement				 
	     Education						 
	     Educational testing				 
	     Federal/state relations				 
	     Foreign languages					 
	     Performance measures				 
	     Students						 
	     Educational standards				 
	     State-administered programs			 
	     Data collection					 
	     English as a Second Language Program		 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO 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.                                         **
**                                                              **
******************************************************************
GAO-06-815

     

     * Results in Brief
     * Background
          * NCLBA Requirements
          * Test Development
          * Education's Responsibilities
     * Students with Limited English Proficiency Performed below Pr
          * In Almost Two-Thirds of States, the Percentage of Students w
          * Factors beyond Student Performance Influence Progress Measur
          * States and Districts We Visited Have Taken Steps to Improve
     * Selected States Considered Language Issues When Developing A
          * States Reported Efforts to Improve Validity of Assessment Re
          * Both Education's Peer Reviews and Our Group of Experts Raise
          * Accommodations Can Increase Validity of Assessment Results,
          * Native Language and Alternate Assessments May Improve the Va
     * Most States Implemented New English Language Proficiency Ass
          * States Are Working with Consortia and Test Developers and In
          * Many States Are Still in the Process of Establishing the Val
     * Education Has Provided Assistance, but States Reported Need
          * Education Has Provided a Variety of Support on Assessment Is
          * Education Has Offered Different Accountability Options for S
     * Conclusions
     * Recommendations for Executive Action
     * Agency Comments
     * GAO Contacts
     * Staff Acknowledgments
     * GAO's Mission
     * Obtaining Copies of GAO Reports and Testimony
          * Order by Mail or Phone
     * To Report Fraud, Waste, and Abuse in Federal Programs
     * Congressional Relations
     * Public Affairs

Report to Congressional Requesters

United States Government Accountability Office

GAO

July 2006

NO CHILD LEFT BEHIND ACT

Assistance from Education Could Help States Better Measure Progress of
Students with Limited English Proficiency

This report was originally issued on July 26, 2006. It was reissued on
September 15, 2006, with the following change: GAO reposted the web
version of this report on September 15, 2006, to include a link to a
Spanish translation of the Highlights page,
http://www.gao.gov/highlights/d061111high.pdf . Para ver la version de
"Highlights" en espanol, http://www.gao.gov/highlights/d061111high.pdf ,
con la fecha del 15 de Septiembre de 2006, oprima aqui.

GAO-06-815

Contents

Letter 1

Results in Brief 3
Background 6
Students with Limited English Proficiency Performed below Progress Goals
in 2004 in Two-Thirds of States, but States We Studied Are Working to
Improve Student Academic Performance 13
Selected States Considered Language Issues When Developing Academic
Assessments, but Validity and Reliability Concerns Remain 25
Most States Implemented New English Language Proficiency Assessments but
Faced Challenges Establishing Their Validity and Reliability 35
Education Has Provided Assistance, but States Reported Need for Additional
Guidance and Flexibility 41
Conclusions 49
Recommendations for Executive Action 50
Agency Comments 51
Appendix I GAO's Group of Experts on Assessing the Academic Knowledge of
Students with Limited English Proficiency 53
Appendix II Determining Adequate Yearly Progress for Student Groups 55
Appendix III Percentage of Districts Making AYP Goals for Mathematics for
Students with Limited English Proficiency 57
Appendix IV Proficiency Scores on Mathematics Tests for All Students and
Students with Limited English Proficiency 58
Appendix V Enhanced Assessment Consortia Participation 60
Appendix VI English Language Proficiency Assessments Used in the 2005-2006
School Year, by State 61
Appendix VII Comments from the Department of Education 63
Appendix VIII GAO Contacts and Acknowledgments 68
Related GAO Products 69

Tables

Table 1: Selected Provisions from Title I of NCLBA 8
Table 2: Selected Provisions from Title III of NCLBA 10
Table 3: Examples of English Language Proficiency and Language Arts
Standards for a Fifth-Grade Student 10
Table 4: Percentage of Elementary Students Scoring at the Proficient Level
or Above on State Mathematics Assessment for Selected Student Groups,
School Year 2003-2004 19
Table 5: Examples of Issues Relating to Assessing Students with Limited
English Proficiency Raised in Education's Peer Review Reports 28
Table 6: Most Frequently Cited Accommodations in 42 States 31

Figures

Figure 1: NCLBA's Requirements for Students with Limited English
Proficiency under Title I and Title III 11
Figure 2: School Year 2003-2004 Comparison of Percentage of Students with
Limited English Proficiency Who Achieved Proficient Scores in Language
Arts and Mathematics with State-Established Progress Goals 15
Figure 3: Percentage of Districts in 18 Selected States Reporting Adequate
Yearly Progress Results in School Year 2003-2004 for Students with Limited
English Proficiency 17
Figure 4: Use of Native Language and Alternate Assessments for Students
with Limited English Proficiency 34
Figure 5: Type of English Language Proficiency Assessment Administered in
2005-2006 School Year 38
Figure 6: Movement of Students In and Out of Limited English Proficient
Student Group and Other Student Groups 47
Figure 7: Process for Determining Adequate Yearly Progress for a Student
Group 56

Abbreviations

AYP adequate yearly progress

CELLA Comprehensive English Language Learning Assessment

ELDA English Language Development Assessment

ESL English as a second language

GAO Government Accountability Office

MWAC Mountain West Assessment Consortium

NCLBA No Child Left Behind Act

PA EAG Pennsylvania Enhanced Assessment Grant

SCASS State Collaborative on Assessment and Student Standards

WIDA World-Class Instructional Design and Assessment

This is a work of the U.S. government and is not subject to copyright
protection in the United States. It may be reproduced and distributed in
its entirety without further permission from GAO. However, because this
work may contain copyrighted images or other material, permission from the
copyright holder may be necessary if you wish to reproduce this material
separately.

United States Government Accountability Office

Washington, DC 20548

July 26, 2006

The Honorable George Miller Ranking Minority Member Committee on Education
and the Workforce House of Representatives

The Honorable Ruben Hinojosa Ranking Minority Member Subcommittee on
Select Education Committee on Education and the Workforce House of
Representatives

The Honorable Lynn Woolsey Ranking Minority Member Subcommittee on
Education Reform Committee on Education and the Workforce House of
Representatives

The Honorable Raul Grijalva House of Representatives

An estimated 5 million children with limited English proficiency were
enrolled in U.S. public schools during the 2003-2004 school year,
representing about 10 percent of the total school population. They speak
over 400 languages, with almost 80 percent of students with limited
English proficiency speaking Spanish. These students have difficulties in
speaking, reading, writing, or understanding English that interfere with
their ability to successfully participate in school. Because of these
language barriers, obtaining information on the academic knowledge of
these students from an assessment that is valid and reliable (i.e., it
measures what it is designed to measure in a consistent manner) presents
challenges. As a result, students with limited English proficiency have
historically been excluded from statewide assessments, leaving states and
districts with little information about how these students are performing
academically.

In 1994, the enactment of the Improving America's Schools Act required
states to assess these students, to the extent practicable, in the manner
most likely to yield accurate information about their academic knowledge.
Subsequently, Congress passed the No Child Left Behind Act of 2001 (NCLBA)
with the goal of increasing academic achievement and closing achievement
gaps among different student groups. Specifically, NCLBA required states
to demonstrate that all students have reached the "proficient" level on a
state's language arts and mathematics assessments by 2014. States are
obligated to demonstrate "adequate yearly progress" toward this goal each
year-that is, they must show that increasing percentages of students are
reaching proficient achievement levels over time. Students with limited
English proficiency, along with other targeted student groups, must
separately meet the same academic progress goals as other students.
Further, NCLBA required states to annually assess the English proficiency
of these students and to demonstrate that they are making progress in
becoming proficient in English. Because these students are defined by a
temporary characteristic-unlike other student groups targeted in
NCLBA-once a state determines that students with limited English
proficiency have attained English proficiency, they are no longer included
in the group of students with limited English proficiency, although
Education has given states some flexibility in this area.

Given your interest in the academic achievement of these students and the
validity and reliability of assessments used to measure their performance,
we are providing information on (1) the extent to which students with
limited English proficiency are meeting adequate yearly progress goals and
what selected states and districts are doing to support the improved
academic performance of these students, (2) what states have done to
ensure that results from language arts and mathematics assessments are
valid and reliable for students with limited English proficiency, (3) how
states are assessing English proficiency and what they are doing to
address the validity and reliability of these assessment results, and (4)
how the Department of Education (Education) is supporting states' efforts
to meet NCLBA's assessment requirements for these students.

To determine the extent to which students with limited English proficiency
are meeting adequate yearly progress goals, we collected school year
2003-2004 state-level data for 48 states, including the District of
Columbia. We obtained the majority of our data from state Web sites and,
when necessary, contacted state officials for these data. Three states did
not publish data in a format that allowed us to determine if students with
limited English proficiency had met the state's adequate yearly progress
goals. We also collected additional achievement data for 2003-2004 at the
school district level from 18 states. We chose a nonrandom sample of
states with the largest percentage of the national population of students
with limited English proficiency, states with the largest percentage
increases in these students between 1990 and 2000, and included at least 2
states from each region represented by Education's regional education
laboratories (with the exception of one region that included only one
state). When district-level achievement data for school year 2003-2004
were not available on a state's Web site or a state had more than 500
districts, we requested the data directly from state officials; 2 states
did not respond to our request for these data. We determined that the
state and district data were sufficiently reliable for our purposes. We
studied 5 states in depth (California, Nebraska, New York, North Carolina,
and Texas) to collect detailed information from state and district
officials on their assessment practices, efforts to ensure the validity
and reliability of their assessments for students with limited English
proficiency, and their approaches to improve the performance of these
students. These 5 states had relatively large percentages of students with
limited English proficiency or had experienced large increases in their
populations of these students. In addition, we selected these particular
states to ensure variation in geography, types of English language
proficiency tests used, and use of different approaches to assessing the
content knowledge of this student group. To obtain information on the
assessments used by other states, we reviewed accountability workbooks and
other documents that states submit to Education, available reports from
state monitoring visits conducted by Education, and available peer review
reports from 38 states on their assessment and accountability systems. In
addition to studying 5 states, we directly contacted officials in 28
states to confirm what English language proficiency assessment they
planned to administer in 2005-2006 and to discuss what guidance Education
had provided regarding these assessments. We also interviewed officials
from major test development companies, from state consortia that are
developing English language proficiency assessments, and from Education.
To assess state efforts to ensure the validity and reliability of their
assessments, we reviewed national assessment standards developed by
professional organizations and convened a group of experts to discuss
states' efforts to develop and implement valid and reliable academic
assessments for students with limited English proficiency (see app. I for
more information about these experts). Finally, we obtained information
from the 50 states and the District of Columbia on their use of native
language assessments using a short e-mail survey. We conducted our review
between June 2005 and June 2006 in accordance with generally accepted
government auditing standards.

                                Results in Brief

In school year 2003-2004, state data showed that the percentage of
students with limited English proficiency scoring proficient on a state's
language arts and mathematics tests was lower than the state's annual
progress goals in nearly two-thirds of the 48 states for which we obtained
data. To help these students progress academically, state and district
officials in the 5 states we visited reported using a variety of
strategies, including training teachers to incorporate language
development into academic classes. Further, our review of data 49 states
submitted to Education showed that the performance of students with
limited English proficiency on states' mathematics assessments for
elementary school students was lower than that of the total student
population in all of these states but 1. Although the student groups are
not mutually exclusive, in most of the 49 states, the performance of
students with limited English proficiency was generally lower than that of
other groups, such as economically disadvantaged students. Factors other
than student academic knowledge, however, can influence whether states and
districts meet their academic progress goals for students with limited
English proficiency, such as how a state establishes its annual progress
goals. To support improved academic progress for these students, district
and state officials we spoke with in our 5 study states reported using
strategies similar to those considered good practices for all students. In
particular, they cited providing teacher training focused on these
students, having school leadership focused on their needs and using data
to target interventions as key to the success of these students.

For assessments of academic knowledge in language arts and mathematics, we
found that our 5 study states have taken some steps to address the
specific challenges associated with assessing students with limited
English proficiency. Although officials in these states reported taking
steps to follow generally accepted test development procedures to ensure
the validity and reliability of results from these assessments for the
general student population, these assessments may not provide valid
results for students with limited English proficiency. Our group of
experts expressed concerns about whether all states are assessing these
students in a valid and reliable manner, noting that states are not taking
all the critical steps needed to do so. Although states have been required
to include these students in their assessments since 1994, Education's
recent peer reviews of 38 states cited 25 for not providing sufficient
evidence on the validity or reliability of results for students with
limited English proficiency. In 1 state, for example, procedures to
develop test questions did not include an adequate check for language
biases. To increase the validity and reliability of assessment results for
this population, most states offered accommodations, such as offering
extra time to complete the test and using a bilingual dictionary. While
most states offered some accommodations, our experts indicated that
research is lacking on what specific accommodations are appropriate for
students with limited English proficiency. Our survey of states and review
of state documents found that 16 states used statewide native language
assessments for some grades and 13 states used statewide alternate
assessments (such as portfolios of classroom work) in 2005 to better
accommodate certain students with limited English proficiency. While such
assessments may improve the validity of test scores, our group of experts
noted that developing native language and alternate assessments requires
resources and expertise that not all states have. Further, our experts
told us that native language assessments may not provide valid results for
students who are not receiving instruction in their native language. In
addition, developing assessments in all languages spoken by students with
limited English proficiency would likely not be practicable for most
states.

With respect to English language proficiency assessments, many states are
still in the preliminary phases of developing and administering new
assessments to measure students' progress in learning English, as required
by NCLBA; as a result, complete information on the validity and
reliability of these assessments is not yet available. To assess these
students in the 2005-2006 school year, 22 states used new assessments
developed by one of four state consortia; 8 states used customized,
off-the-shelf assessments offered by testing companies; 14 states used
off-the-shelf assessments; and 7 states used state-developed assessments.
While a few states already had the required English language proficiency
assessments in place, others will be administering these assessments for
the first time in 2006; as a result, states and test developers are still
collecting evidence to document the validity and reliability of the
results for most of these tests. An Education-funded study by a national
education research organization reviewed the available documentation for
the English language proficiency assessments used by 33 states in the
2005-2006 school year and found insufficient documentation on the validity
and reliability of results from these assessments.

Education has offered states support and technical assistance in a variety
of ways to help them appropriately assess students with limited English
proficiency, such as providing training, conducting peer reviews of
states' academic assessments, and providing flexibility in assessing these
students. However, Education has issued little written guidance to states
on developing English language proficiency assessments that meet NCLBA's
requirements and on tracking the progress of students in acquiring
English. Officials in about one-third of the 33 states we visited or
contacted told us that they were uncertain about Education's requirements
for the new English language proficiency assessments and that they wanted
more guidance. In addition, our group of experts reported that some states
need more assistance to develop language arts and mathematics assessments
that provide valid measures of the academic knowledge of this group of
students. To support states' efforts to incorporate these students into
their accountability systems for academic performance, Education has
offered states some flexibilities in how they track progress goals for
these students. For example, students who have been in the United States
for less than 1 year do not have to be assessed for language arts.
However, several state and district officials in the 5 states we studied
told us that additional flexibility, such as excluding students from
testing for a longer period, is needed to ensure that adequate yearly
progress measures accurately track the academic progress of these
students.

To help states improve their assessment of students with limited English
proficiency, we are recommending that the Secretary of Education (1)
support additional research on accommodations, (2) identify and provide
additional technical support states need to ensure the validity and
reliability of academic assessments for these students, (3) publish more
detailed guidance on assessing the English language proficiency of these
students, and (4) explore ways to provide additional flexibility with
respect to measuring annual progress for these students. In its comments,
Education generally agreed with our recommendations.

                                   Background

Students with limited English proficiency are a diverse and complex group.
They speak many languages and have a tremendous range of educational needs
and include refugees with little formal schooling and students who are
literate in their native languages. Accurately assessing the academic
knowledge of these students in English is challenging. If a student
responds incorrectly to a test item, it may not be clear if the student
did not know the answer or misunderstood the question because of language
barriers.

Several approaches are available to allow students to demonstrate their
academic knowledge while they are becoming proficient in English, although
each poses challenges. First, a state can offer assessments in a student's
native language. However, vocabulary in English is not necessarily
equivalent in difficulty to the vocabulary in another language. As a
result, a test translated from English may not have the same level of
difficulty as the English version. If a state chooses to develop a
completely different test in another language instead of translating the
English version, the assessment should measure the same standards and
reflect the same level of difficulty as the English version of the test to
ensure its validity. Second, states can also offer accommodations, such as
providing extra time to take a test, allowing the use of a bilingual
dictionary, reading test directions aloud in a student's native language,
or administering the test in a less distracting environment.
Accommodations alter the way a regular assessment is administered, with
the goal of minimizing the language impediments faced by students with
limited English proficiency; they are intended to level the playing field
without providing an unfair advantage to these students. Finally, states
can use alternate assessments that measure the same things as the regular
assessment while minimizing the language burden placed on the student. For
example, an alternate assessment can be a traditional standardized test
that uses simplified English or relies more on pictures and diagrams. It
can also be a portfolio of a student's class work that demonstrates
academic knowledge. In either case, studies would be needed to demonstrate
that the alternate assessment is equivalent to the regular assessment.

NCLBA Requirements

Title I of NCLBA seeks to ensure that all children have a fair and equal
opportunity to obtain a high-quality education and become proficient in
academic subjects. It requires states to administer tests in language arts
and mathematics to all students in certain grades and to use these tests
as the primary means of determining the annual performance of states,
districts, and schools. These assessments must be aligned with the state's
academic standards-that is, they must measure how well a student has
demonstrated his or her knowledge of the academic content represented in
these standards. States are to show that increasing percentages of
students are reaching the proficient level on these state tests over time.
NCLBA also requires that students with limited English proficiency receive
reasonable accommodations and be assessed, to the extent practicable, in
the language and form most likely to yield accurate data on their academic
knowledge. Somewhat similar versions of these provisions, such as
reporting testing results for different student groups, had been included
in legislation enacted in 1994. One new NCLBA requirement was for states
to annually assess the English language proficiency of students identified
as having limited English proficiency. Table 1 summarizes some key Title I
provisions from NCLBA.

Table 1: Selected Provisions from Title I of NCLBA

State academic         Beginning in the 2005-2006 school year, states must 
assessments            implement annual, high-quality state assessments in 
                          reading and mathematics in grades 3-8 and at least  
                          once in high school.a These assessments must be     
                          aligned with challenging state academic standards   
                          and must be "consistent with relevant, nationally   
                          recognized professional and technical standards for 
                          such assessments" and used in ways that are valid   
                          and reliable. States must provide for the           
                          participation of all students, including those with 
                          limited English proficiency.                        
Academic assessment    Students with limited English proficiency are to be 
provisions related to  assessed in a valid and reliable manner. In         
students with limited  addition, they must be provided with reasonable     
English proficiency    accommodations and be assessed, to the extent       
                          practicable, "in the language and form most likely  
                          to yield accurate data" on their academic           
                          knowledge. In addition, for language arts, students 
                          who have been in U.S. schools for 3 years or more   
                          generally must be assessed in English.              
Adequate yearly        States must set annual goals that lead to all       
progress               students achieving proficiency in language arts and 
                          mathematics by 2014. To be deemed as having made    
                          adequate yearly progress for a given year, each     
                          district and school must show that the requisite    
                          percentage of each designated student group, as     
                          well as the student population as a whole, met the  
                          state proficiency goal (that is, the percentage of  
                          students who have achieved the proficient level on  
                          the state's assessments). Schools must also show    
                          that at least 95 percent of students in each        
                          designated student group participated in these      
                          assessments. Further, schools must also demonstrate 
                          that they have met state targets on other academic  
                          indicators-graduation rates in high school or       
                          attendance or other measures in elementary and      
                          middle schools. Alternatively, a district or school 
                          can make adequate yearly progress through the "safe 
                          harbor" provision, if the percentage of students in 
                          a group considered not proficient decreased by at   
                          least 10 percent from the preceding year and the    
                          group made progress on one of the state's other     
                          academic indicators.                                
Actions when adequate  Schools that receive funding under Title I of NCLBA 
yearly progress not    must take specified actions if they do not meet     
achieved               state progress goals. Specifically, schools that do 
                          not make adequate yearly progress for 2 consecutive 
                          years or more are identified for improvement and    
                          must, among other things, offer parents an          
                          opportunity to transfer students to another school  
                          and provide supplemental services (e.g., tutoring). 
                          Those that miss the annual goals for additional     
                          years are identified for successive stages of       
                          intervention, including corrective action and       
                          restructuring.                                      
State English language States must annually assess the English language    
proficiency            proficiency of all students with limited English    
assessments            proficiency, measuring students' oral language,     
                          reading, and writing skills in English.             

Source: Pub. L. No. 107-110.

aBeginning in school year 2007-2008, states must implement similar
assessments in science.

Accurately assessing the academic knowledge of students with limited
English proficiency has become more critical because NCLBA designated
specific groups of students for particular focus. These four groups are
students who (1) are economically disadvantaged, (2) represent major
racial and ethnic groups, (3) have disabilities, and (4) are limited in
English proficiency. These groups are not mutually exclusive, so that the
results for a student who is economically disadvantaged, Hispanic, and has
limited English proficiency could be counted in all three groups. States
and school districts are required to measure the progress of all students
in meeting academic proficiency goals, as well as to measure separately
the progress of these designated groups. To be deemed as having made
adequate yearly progress, generally each district and school must show
that each of these groups met the state proficiency goal (that is, the
percentage of students who have achieved the proficient level on the
state's assessments) and that at least 95 percent of students in each
designated group participated in these assessments.

Although NCLBA placed many new requirements on states, states have broad
discretion in many key areas. States establish their academic content
standards and then develop their own tests to measure the academic content
students are taught in school. States also set their own standards for
what constitutes proficiency on these assessments. In addition, states set
their own annual progress goals for the percentage of students achieving
proficiency, using guidelines outlined in NCLBA.1

Title III of NCLBA focuses specifically on students with limited English
proficiency, with the purpose of ensuring that these students attain
English proficiency and meet the same academic content standards all
students are expected to meet. This title established new requirements
intended to hold states and districts accountable for student progress in
attaining English proficiency. It requires states to establish goals to
demonstrate, among other things, annual increases in (1) students making
progress in learning English and (2) students attaining English
proficiency. Specifically, states must establish English language
proficiency standards that are aligned with a state's academic content
standards. The purpose of these alignment requirements is to ensure that
students are acquiring the academic language they will need to
successfully participate in the classroom. Education also requires that a
state's English language proficiency assessment be aligned to its English
language proficiency standards. While NCLBA requires states to administer
academic assessments to students in specific grades, it requires states to
administer an annual English language proficiency assessment to all
students with limited English proficiency, from kindergarten to grade 12.
See table 2 for summary of key Title III provisions.

1For more information on differences in state approaches to meeting NCLBA
requirements, see GAO, No Child Left Behind Act: Improvements Needed in
Education's Process for Tracking States' Implementation of Key Provisions,
GAO-04-734 (Washington, D.C.: Sept. 30, 2004).

Table 2: Selected Provisions from Title III of NCLBA

                             States must establish English language           
State English language    proficiency standards that are aligned with the  
proficiency standards     state's challenging academic content standards.  
Tracking student progress States must establish objectives for improving   
in learning English       students' English proficiency in four areas:     
                             speaking, listening, reading, and writing.a      
                             States receiving grants under Title III must     
                             establish annual goals for increasing and        
                             measuring the progress of students with limited  
                             English proficiency in (1) learning English, (2) 
                             attaining English proficiency, and (3) meeting   
                             adequate yearly progress goals in attaining      
                             academic proficiency outlined in Title I.        
Actions when annual goals Districts that receive funding under Title III   
for students with limited are subject to certain consequences if they do   
English proficiency not   not meet a state's annual Title III goals. If a  
met                       district does not meet the goals for 2           
                             consecutive years, it must develop an            
                             improvement plan that addresses the factors that 
                             prevented the district from meeting the goals.   
                             If a district does not meet the goals for 4      
                             consecutive years, it must modify its curriculum 
                             and method of instruction or the state must      
                             determine whether to continue to fund the        
                             district and require the district to replace all 
                             personnel related to the district's inability to 
                             meet the goals.                                  

Source: Pub. L. No. 107-110.

aTitle I refers to oral language skills, which encompass listening and
speaking.

Language arts standards define the academic skills a student is expected
to master, while English language proficiency standards define progressive
levels of competence in the acquisition of English necessary to
participate successfully in the classroom. Examples of standards for
English language proficiency and language arts are provided in table 3.

Table 3: Examples of English Language Proficiency and Language Arts
Standards for a Fifth-Grade Student

English language proficiency                                               
standards                             Language arts standards
The student can comprehend reading    The student can independently read   
passages written in familiar or short and comprehend a grade-level         
sentence patterns and verbalize some  appropriate text and write a short   
of the main points of the passage.    essay describing the main idea of    
                                         the text.                            
The student can use acquired          The student can apply knowledge of   
knowledge of the English language to  reading strategies to comprehend the 
learn and understand new vocabulary   text of the next higher level of     
in context.                           difficulty.                          

Source: U.S. Department of Education, "Final Non-Regulatory Guidance on
the Title III State Formula Grant Program-Standards, Assessments and
Accountability," February 2003.

Under NCLBA, states, districts, and schools have two sets of
responsibilities for students with limited English proficiency. As shown
in figure 1, they are responsible for ensuring that these students make
progress in learning English under Title III and that they become
proficient in language arts and mathematics under Title I. Beginning with
the 2004-2005 school year, Education is required to annually review
whether states have made adequate yearly progress (as defined by the
state) for each of the student groups and have met their objectives for
increasing the number or percentage of students who become proficient in
English.

Figure 1: NCLBA's Requirements for Students with Limited English
Proficiency under Title I and Title III

Test Development

NCLBA's emphasis on validity and reliability reflects the fact that these
concepts are among the most important in test development. Validity refers
to whether the test measures what it is intended to measure. Reliability
refers to whether or not a test yields consistent results across time and
location and among different sections of the test. A test cannot be
considered valid if it is unreliable. The Standards for Educational and
Psychological Testing provide universally accepted guidance for the
development and evaluation of high-quality, psychometrically sound
assessments.2 They outline specific standards to be considered when
assessing individuals with limited English proficiency, including (1)
determining when language differences produce threats to the validity and
reliability of test results, (2) providing information on how to use and
interpret results when tests are used with linguistically diverse
individuals, and (3) collecting the same evidence to support claims of
validity for each linguistic subgroup as was collected for the population
as a whole.

2These standards are sponsored and published jointly by the American
Educational Research Association, the American Psychological Association,
and the National Council on Measurement in Education.

Test development begins with determining the purpose of the test and the
content to be measured by the test. NCLBA outlines several purposes of
statewide assessments, including determining the yearly performance of
schools and districts, interpreting individual student academic needs, and
tracking the achievement of several groups of students. NCLBA requires
that the content of statewide assessments reflects state standards in
language arts and mathematics, but the specific skills measured can vary
from state to state. For example, a language arts assessment could measure
a student's knowledge of vocabulary or ability to write a persuasive
essay. Variations in purpose and content affect test design, as well as
the analyses necessary to determine validity and reliability.

After determining the purpose and content of the test, developers create
test specifications, which delineate the format of the questions and
responses, as well as the scoring procedures. Specifications may also
indicate additional information, such as the intended difficulty of
questions, the student population that will take the test, and the
procedures for administering the test. These specifications subsequently
guide the development of individual test questions. The quality of the
questions is usually ascertained through review by knowledgeable educators
and statistical analyses based on a field test of a sample of
students-ideally the sample is representative of the overall target
student population so the results will reflect how the questions will
function when the test is administered to the population. These reviews
typically evaluate a question's quality, clarity, lack of ambiguity, and
sometimes its sensitivity to gender or cultural issues; they are intended
to ensure that differences in student performance are related to
differences in student knowledge rather than other factors, such as
unnecessarily complex language. Once the quality has been established,
developers assemble questions into a test that meets the requirements of
the test specifications. Developers often review tests after development
to ensure that they continue to produce accurate results.

Education's Responsibilities

Education has responsibility for general oversight of Titles I and III of
NCLBA. The department's Office of Elementary and Secondary Education
oversees states' implementation of Title I requirements with respect to
academic assessments and making adequate progress toward achieving
academic proficiency for all students by 2014. Education's Office of
English Language Acquisition, Language Enhancement and Academic
Achievement for Limited English Proficient Students oversees states' Title
III responsibilities, which include administering annual English language
proficiency assessments to students with limited English proficiency and
demonstrating student progress in attaining English language proficiency.

Students with Limited English Proficiency Performed below Progress Goals in 2004
 in Two-Thirds of States, but States We Studied Are Working to Improve Student
                              Academic Performance

In school year 2003-2004, the percentage of students with limited English
proficiency reported by states as scoring proficient on a state's language
arts and mathematics tests was lower than the state's annual progress
goals (established for all students) in nearly two-thirds of the 48 states
for which we obtained data.3 Further, data from state mathematics tests
showed that these students were generally achieving lower rates of
academic proficiency than the total student population. However, factors
other than student academic performance can influence whether a state
meets its progress goals, such as which students a state includes in the
limited English proficient group and how a state establishes its annual
progress goals. Officials in our study states reported using several
common approaches, including providing teacher training specific to the
needs of limited English proficient students and using data to guide
instruction and identify areas for improvement.

3We included the District of Columbia in our 48-state total.

In Almost Two-Thirds of States, the Percentage of Students with Limited English
Proficiency Achieving Proficient Scores Was Below the State's Annual Progress
Goals

In nearly two-thirds of the 48 states for which we obtained data, state
data showed that the percentage of students with limited English
proficiency scoring proficient on language arts and mathematics tests was
below the annual progress goal set by the state for school year 2003-2004.
Students with limited English proficiency met academic progress goals in
language arts and mathematics in 17 states.4 In 31 states, state data
indicated that these students missed the goals either for language arts or
for both language arts and mathematics (see fig. 2). In 21 states, the
percentage of proficient students in this group was below both the
mathematics and the language arts proficiency goals. See appendix II for
information on how adequate yearly progress measures are calculated.

4In 7 of the 17 states, students with limited English proficiency met a
state's adequate yearly progress goals through NCLBA's safe harbor
provision-that is, by decreasing the percentage of students scoring
nonproficient by 10 percent or more and showing progress on another
academic indicator.

Figure 2: School Year 2003-2004 Comparison of Percentage of Students with
Limited English Proficiency Who Achieved Proficient Scores in Language
Arts and Mathematics with State-Established Progress Goals

Notes: We obtained data for 42 states from their state Web sites and
contacted state officials in 6 states to obtain these data. Three states
did not report data in a format that allowed us to determine whether the
percentage of students with limited English proficiency met or exceeded
the annual progress goals established by the state.

When states reported proficiency data for different grades or groups of
grades, we determined that students with limited English proficiency met a
state's progress goals if the student group met all proficiency and
participation goals for all grades reported. An Education official told us
that a state could not make adequate yearly progress if it missed one of
the progress goals at any grade level.

All of the states on the map where the proficiency percentage for students
with limited English proficiency met or exceeded the state's annual
progress goal also met NCLBA's participation goals.

We incorporated states' use of confidence intervals and NCLBA's safe
harbor provision in determining whether the percentage of students with
limited English proficiency achieving proficient scores met or exceeded a
state's progress goals. If a state's published data did not explicitly
include such information, we contacted state officials to ensure that the
state did not meet its progress goals through the use of confidence
intervals or through NCLBA's safe harbor provision. In the following seven
states, the percentage of students with limited English proficiency was
below the state's annual progress goal for language arts or for both
language arts and mathematics, but the student group met the state's
requirements for progress through the safe harbor provision: Delaware,
Idaho, Maryland, Massachusetts, Oklahoma, Rhode Island, and Utah.

We reported 2004-2005 school year data for Oklahoma, New Mexico and Utah
because we could not obtain data for the 2003-2004 school year. Data from
Iowa, Massachusetts, and Rhode Island are for the 2002-2004 school years.

Rhode Island did not separately report participation rates for students
with limited English proficiency. Instead, it reported that all students
met the 95 percent participation goal.

We also obtained additional data from 18 states to determine whether
districts were meeting annual progress goals for students with limited
English proficiency in school year 2003-2004.5 In 14 of the 18 states,
however, we found that less than 40 percent of the districts in each state
reported separate results for this group of students (see fig. 3).6
Districts only have to report progress results for a student group if a
minimum number of students are included in the group.7 In Nebraska, for
example, only 4 percent of districts reported progress goals for students
with limited English proficiency. Except for Florida, Hawaii, and Nevada,
less than half of the districts in each state reported separate results
for this group of students. Even when districts do not have to report on
students with limited English proficiency, however, the test scores for
these students are included in the state's overall progress measures.

5We requested district data from 20 states, and 18 states responded to our
request.

6The number of districts reporting results for these students may increase
in the future, since states were required to assess students in more
grades beginning with the 2005-2006 school year. Testing in more grades
will likely increase the number of districts meeting the minimum number of
limited English proficient students that will be required to separately
report proficiency results.

7States determine the minimum number of students in a student group,
usually between 25 and 45 students.

Figure 3: Percentage of Districts in 18 Selected States Reporting Adequate
Yearly Progress Results in School Year 2003-2004 for Students with Limited
English Proficiency

Notes: If a district reported annual progress results for students with
limited English proficiency in either language arts proficiency or
mathematics proficiency, or both, we considered that the district reported
adequate yearly progress results for the student group.

Hawaii has only one school district. Since the state reported separate
results for students with limited English proficiency, it has been
included as 100 percent of districts reporting separate results for these
students.

For those districts that reported results for students with limited
English proficiency, district-level data showed that most districts in 13
of the 18 states met their mathematics progress goals for these students.
For example, 67 percent of reporting districts in Nebraska and 99 percent
of reporting districts in Texas met the state's goals. In 4 states, less
than half of the districts reporting results for these students met the
state mathematics progress goals. Specifically, 26 percent of Alaska
districts, 33 percent of Nevada districts, 48 percent of Oregon districts,
and 48 percent of Florida districts met these goals. (See app. III for
results from each of the 18 states.)

In addition to looking at whether students with limited English
proficiency met annual progress goals at the state and district level, we
also examined achievement levels on state assessments for this group of
students compared with the total student population (which also includes
students with limited English proficiency). Looking at mathematics results
reported by 49 states to Education, for example, in all but one state, we
found that a lower percentage of students with limited English proficiency
at the elementary school level achieved proficient scores, compared to the
total student population in school year 2003-2004 (see app. IV for the
results reported by the 49 states). Twenty-seven states reported that the
total student population outperformed students with limited English
proficiency by 20 percentage points or more. The differences among groups
in the percentage of students achieving proficient scores varied across
states. South Dakota, for example, reported a large achievement gap, with
37 percent of limited English proficient students scoring at the
proficient level, compared to 78 percent for the entire student
population. The gap was less pronounced in Texas, where 75 percent of
students with limited English proficiency achieved proficient scores on
the mathematics assessment, while 85 percent of the total student
population did. In Louisiana, these students performed about the same as
the total student population, with 58 percent of limited English
proficient students scoring at the proficient level on the elementary
mathematics assessment, compared to 57 percent of the total student
population.

We also found that, in general, a lower percentage of students with
limited English proficiency achieved proficient test scores than other
selected student groups (see table 4). All of the 49 states reported that
these students achieved lower rates of proficiency than white students.8
The performance of limited English proficient students relative to the
other student groups varied. In 37 states, for example, economically
disadvantaged students outperformed students with limited English
proficiency, while students with disabilities outperformed these students
in 14 states. In 12 states, all the selected student groups outperformed
students with limited English proficiency.

8Student groups are not mutually exclusive, with each of the ethnic and
racial categories probably including some number of students with limited
English proficiency. For example, the results for a student who is both
white and limited English proficient would be included in both groups.

Table 4: Percentage of Elementary Students Scoring at the Proficient Level
or Above on State Mathematics Assessment for Selected Student Groups,
School Year 2003-2004

            Students                                                             
                with                                                       
             Limited     Students                                          
             English         with                   Economically           
 States  Proficiency Disabilities African-American Disadvantaged  Hispanic White
 Ala.             53           31               58            62        61    81 
 Alaska           40           36               50            50        63    77 
 Ariz.a           32           31               46      data not        44    72 
                                                       available           
 Ark.b            49           24               38            53        56    74 
 Calif.           33           24               28            33        33    61 
 Colo.a           76           61               74            79        79    94 
 Conn.            47           49               69            61        61    88 
 Del.a            70           47               61            67        74    87 
 D.C.             34           14               49            48        57    89 
 Fla.             48           39               44            52        60    74 
 Ga.              53           46               65            66        67    84 
 Hawaiia           9            6               19            18        16    36 
 Idaho            62           55               69            76        68    87 
 Ill.a            53           57               54            61        64    89 
 Ind.a            47           40               54            60        60    75 
 Iowa             49           39               46            62        56    80 
 Kans.            58           67               61            70        65    84 
 Ky.a             32           29               28            36        38    51 
 La.              58           30               40            48        62    74 
 Maine            10           13               15            20        20    33 
 Md.              39           39               52            51        59    83 
 Mass.            22           20               20            22        19    49 
 Mich.            59           42               51            57        58    77 
 Minn.a           38           45               39            52        45    77 
 Miss.            79           61               69            72        87    91 
 Mo.              30           24               24            28        29    45 
 Mont.            15           22               32            33        36    49 
 Nebr.            73           65               72            79        80    90 
 Nev.a            22           22               27            32        32    57 
 N.H.a            57           61               72            71        67    85 
 N.J.             47           46               50            54        59    81 
 N.Mex.           36           31               50            47        49    72 
 N.C.             86           75               88            89        90   >95 
 N.Dak.           30           38               45            52        47    68 
 Ohio             48           38               39            48        51    72 
 Okla.a           41           23               38            46        47    61 
 Oreg.a           61           57               71            73        63    86 
 Pa.a             34           27               30            42        38    70 
 R.I.             23           34               32            36        31    60 
 S.C.             16           16               19            22        24    49 
 S.Dak.           37           48               56            65        62    83 
 Tex.             75           76               75            80        81    93 
 Utah             53           43               56            71        52    78 
 Vt.              67           37               51            58  data not    73 
                                                                 available 
 Va.a             84           74               77            79        84    92 
 Wash.            27           29               38            44        39    66 
 W.Va.            68           33               53            64        68    70 
 Wisc.            50           51               45            56        53    80 
 Wyo.             15           21               25            29        24    42 
 Number of states              14               28            37        41    49 
 where students with                                                       
 limited English                                                           
 proficiency had                                                           
 lower proficiency                                                         
 levels than this                                                          
 group (bolded                                                             
 numbers)                                                                  

Source: Consolidated State Performance Reports, 2003-04 school year.

Notes: Bolded numbers indicate when the percentage of students in a group
achieving proficient scores is greater than the percentage of students
with limited English proficiency achieving proficient scores.

Student groups are not mutually exclusive, so that results for a student
who is both Hispanic and economically disadvantaged appear in both groups.

New York and Tennessee did not provide assessment data in their State
Consolidated Performance Reports for the 2003-2004 school year.

aMost states reported assessment data for students in the fourth grade.
States marked with this superscript reported on some other grade at the
elementary school level, usually either third grade or fifth grade.

bThe percentages for Arkansas do not include those students with limited
English proficiency or students with disabilities who were considered
proficient based on the state's portfolio assessment.

Factors beyond Student Performance Influence Progress Measures Reported by
States

Factors beyond student performance can influence the number of states,
districts, and schools meeting progress goals for students with limited
English proficiency. One factor that can affect a state or district's
ability to meet progress goals for this student group is the criteria
states use to determine which students are counted as limited English
proficient. Some states define limited English proficiency so that
students may be more likely to stay in the group for a longer time, giving
them more of an opportunity to develop the language skills necessary to
demonstrate their academic knowledge on state academic assessments
administered in English. On the basis of our review of state
accountability plans, we found that some states removed students from the
group after they have achieved proficiency on the state's English language
proficiency assessment, while other states continued to include these
students until they met additional academic requirements, such as
achieving proficient scores on the state's language arts assessment. A
number of states measured adequate yearly progress for students with
limited English proficiency by including test scores for students for a
set period of time after they were considered proficient in English,
following Education's policy announcement in February 2004 allowing such
an approach.

How rigorously a state defines the proficient level of academic
achievement can also influence the ability of states, districts, and
schools to meet their progress goals. States with less rigorous
definitions of proficiency are more likely to meet their progress goals
for students with limited English proficiency or any other student group
than states with more stringent definitions. Comparing the performance of
students from different states on a nationally administered assessment
suggests that states differ in how rigorously they define proficiency. For
example, eighth-grade students in Colorado and Missouri achieved somewhat
similar scores on the National Assessment of Educational Progress in
mathematics in 2003.9 Specifically, 34 percent of Colorado students scored
proficient or above on this national assessment compared to 28 percent of
Missouri students. On their own state assessments in 2003, however, 67
percent of Colorado students scored proficient or above, compared to just
21 percent in Missouri.10 These results may reflect, among other things, a
difference in the level of rigor in the tests administered by these
states. However, they may also be due in part to differences in what the
national test measures versus what each of the state tests measure.

9Under NCLBA, states are required to participate in the National
Assessment of Educational Progress for reading and mathematics assessments
in grades four and eight. The results from these assessments provide a
national measure of student achievement and serve as confirmatory evidence
about student achievement on state tests.

The likelihood of a state, district, or school meeting its annual progress
goals also depends, in part, on the proficiency levels of its students
when NCBLA was enacted, as well as how the state sets its annual goals.
States vary significantly in the percentage of students scoring at the
proficient level on their academic assessments, so that states with lower
proficiency levels must, on average, establish larger annual increases in
proficiency levels to meet the 2014 goal. Some states planned for large
increases every 2 to 3 years, while others set smaller annual increases.
States that established smaller annual increases in their initial
proficiency goals may be more likely to meet their progress goals at this
time, compared with states that set larger annual increases.

The use of statistical procedures, such as confidence intervals, can also
affect whether a state, district, or school meets its progress goals.
Education officials said that states use such procedures to improve the
reliability of determinations about the performance of districts.
According to some researchers, such methods may improve the validity of
results because they help to account for the effect of small group sizes
and year-to-year changes in student populations.11 Most states currently
use some type of confidence interval to determine if a state or district
has met its progress goals, according to the Center on Education Policy.12
A confidence interval establishes a range of proficiency levels around a
state's annual progress goal.13 If the percentage of students with limited
English proficiency scoring proficient on a state's academic assessments
falls within that range, that group has made the annual progress goal.

10Example taken from Robert L. Linn, "CRESST Policy Brief 8: Fixing the
NCLB Accountability System," Policy Brief of the National Center for
Research on Evaluation, Standards, and Student Testing, Summer 2005.

11Theodore Coladarci, Gallup Goes to School: The Importance of Confidence
Intervals for Evaluating "Adequate Yearly Progress" in Small Schools, the
Rural School and Community Trust Policy Brief, Oct. 2003, and Thomas J.
Kane and Douglas O. Steiger, "Volatility in School Test Scores:
Implications for Test-Based Accountability Systems," in Diane Ravitch,
ed., Brookings Papers on Education Policy 2002, pp. 235-283. Washington,
D.C.: Brookings Institution.

12Center on Education Policy, "From the Capital to the Classroom: Year 4
of the No Child Left Behind Act," March 2006.

States and Districts We Visited Have Taken Steps to Improve Performance of
Students with Limited English Proficiency

To help students with limited English proficiency progress academically,
state and district officials in our 5 study states reported using somewhat
similar strategies, many of which are also considered good practices for
all students. Among the key factors cited by state and district officials
for their success in working with this group were

           o  strong professional development focused on effective teaching
           strategies for students with limited English proficiency;
           o  school or district leadership that focuses on the needs of
           these students, such as providing sufficient resources to meet
           those needs and establishing high academic standards for these
           students;
           o  "data driven" decisions, such as using data strategically to
           identify students who are doing well and those who need more help,
           to identify effective instructional approaches, or to provide
           effective professional development; and
           o  efforts to work with parents to support the academic progress
           of their children.

These approaches are similar to those used by "blue ribbon"
schools-schools identified by Education as working successfully with all
students to achieve strong academic outcomes. The qualities shared by
these blue ribbon schools include professional development related to
classroom instruction, strong school leadership and a vision that
emphasizes high academic expectations and academic success for all
students, using data to target instructional approaches, and parental
involvement. While many blue ribbon schools have a high percentage of
disadvantaged students, including those with limited English proficiency,
their common approaches help them achieve student outcomes that place them
among the top 10 percent of all schools in the state or that demonstrate
dramatic improvement.

13When using confidence intervals, upper and lower limits around a
district's or state's percentage of proficient students are calculated,
creating a range of values within which there is "confidence" the true
value lies. For example, instead of saying that 72 percent of students
scored at the proficient level or above on a test, a confidence interval
may show that percentage to be between 66 and 78 percent, with 95 percent
confidence.

Officials in all 5 of our study states stressed the importance of
providing teachers with the training they need to work effectively with
students with limited English proficiency. For example, state officials in
North Carolina told us that they are developing a statewide professional
development program to train mainstream teachers to present academic
content material so that it is more understandable to students with
limited English proficiency and to incorporate language development while
teaching subjects such as mathematics and science. In one rural North
Carolina school district where students with limited English proficiency
have only recently become a large presence, district officials commented
that this kind of professional development has helped teachers become more
comfortable with these students and given them useful strategies to work
more effectively with them.

In 4 of our study states, officials emphasized the need for strong school
or district leadership that focuses on the needs of students with limited
English proficiency. For example, officials in a California school
district with a high percentage of students with limited English
proficiency told us that these students are a district priority and that
significant resources are devoted to programs for them. The district
administration has instilled the attitude that students with limited
English proficiency can meet high expectations and are the responsibility
of all teachers. To help maintain the focus on these students, the
district has created an English language development progress profile to
help teachers track the progress of each student in acquiring English and
meeting the state's English language development standards.

In addition, officials in 4 of our study states attributed their success
in working with students with limited English proficiency to using data
strategically, for example, to identify effective practices and guide
instruction. At one California school we visited, officials reviewed test
scores to identify areas needing improvement for different classes and
different student groups and to identify effective practices. In addition,
they reviewed test data for each student to identify areas of weakness. If
test data showed that a student was having trouble with vocabulary, the
teacher would work in class to build the student's vocabulary. Similarly,
officials in a New York school reported that they followed student test
scores over 3 years to set goals for different student groups and identify
areas in need of improvement.

Officials in 3 states we visited also cited the importance of involving
parents of students with limited English proficiency in their children's
education. In Nebraska, for example, a technical assistance agency
implemented a family literacy program to help parents and their children
improve their English, and also to involve parents in their children's
education. The program showed parents how they can help children with
their homework and the importance of reading to their children in their
native language to develop their basic language skills. At a New York
middle school, officials told us that they use a parent coordinator to
establish better communication with families, learn about issues at home
that can affect the student's academic performance, and help families
obtain support services, if needed.

Selected States Considered Language Issues When Developing Academic Assessments,
                  but Validity and Reliability Concerns Remain

For academic assessments in language arts and mathematics, officials in
the 5 states we studied reported that they have taken some steps, such as
reviewing test items to eliminate unnecessarily complex language, to
address the specific challenges associated with assessing students with
limited English proficiency. However, Education recently reviewed the
assessment documentation of 38 states and noted some concerns related to
using these assessments for students with limited English proficiency. Our
group of experts also indicated that states are generally not taking the
appropriate set of comprehensive steps to create assessments that produce
valid and reliable results for students with limited English proficiency.
To increase the validity and reliability of assessment results for this
population, most states offered accommodations, such as providing extra
time to complete the assessment and offering native language assessments.
However, offering accommodations may or may not improve the validity of
test results, as research on the appropriate use of accommodations for
these students is lacking. In addition, native language assessments are
not appropriate for all students with limited English proficiency and are
difficult and expensive to develop.

States Reported Efforts to Improve Validity of Assessment Results for Students
with Limited English Proficiency

Officials in the 5 states we studied reported taking some steps to address
the specific challenges associated with assessing students with limited
English proficiency in language arts and mathematics. Officials in 4 of
these states reported following generally accepted test development
procedures when developing their academic assessments, while a Nebraska
official reported that the state expects districts to follow such
procedures when developing their tests. Test development involves a
structured process with specific steps; however, additional steps and
special attention to language issues are required when developing a test
that includes students with limited English proficiency to ensure that the
results are valid and reliable for these students. As the Standards for
Educational and Psychological Testing notes, for example, the test
instructions or the response format may need to be modified to ensure that
the test provides valid information about the skills of students with
limited English proficiency.

Officials in 2 states and at several testing companies mentioned that they
have been focusing more on the needs of these students in recent years.
Officials in California, New York, North Carolina, and Texas told us that
they try to implement the principles of universal design, which support
making assessments accessible to the widest possible range of students.
This is done by ensuring, among other things, that instructions, forms,
and questions are clear and not more linguistically complex than
necessary. In addition, officials in all 5 states we studied told us they
included students with limited English proficiency in the field testing of
assessments. North Carolina officials reported that they oversample for
students with limited English proficiency to ensure that these students
are adequately represented in the field tests.

Another step officials in some states reported taking is assembling panels
or committees to review test items for bias and testing data for bias
related to a student's English proficient status. For example, Texas and
North Carolina officials reported creating review committees to ensure
that test items are accessible to students with limited English
proficiency. Specifically, when developing mathematics items, these states
try to make the language as clear as possible to ensure that the item is
measuring primarily mathematical concepts and to minimize the extent to
which it is measuring language proficiency. A mathematics word problem
involving subtraction, for example, might refer to fish rather than
barracuda. Officials in 4 of our study states told us they used a
statistical approach to evaluate test items for bias against specific
student groups, and three of these reported using it to detect bias
related to students with limited English proficiency. However, this type
of analysis can only be used when a relatively large number of students in
the specific group is taking the test. Members of our expert group
recommended the use of this technique for states with a large enough
number of students with limited English proficiency; however, one member
noted that this technique may not be appropriate if a state's population
of students with limited English proficiency is diverse but is treated as
homogenous in the analyses.

Some of our study states also reported including experts on limited
English proficiency or English as a second language (ESL) issues in the
development and review of test items, although only 1 reported involving
them in all aspects of test development. In North Carolina, for example,
officials told us that ESL teachers and supervisors are involved in
reviewing all aspects of the test development process, including item
writing, field testing, and operational testing. Some state officials also
told us that they included education staff involved with students with
limited English proficiency in the development of assessments.

Both Education's Peer Reviews and Our Group of Experts Raised Concerns Regarding
State Efforts to Ensure Valid and Reliable Results

Education's recent NCLBA peer reviews of 38 states14 found that 25 did not
provide sufficient evidence on the validity or reliability of results for
students with limited English proficiency, although states have been
required to include these students in their assessments since 1994.15 For
example, peer reviewers found that Alabama's documentation did not include
sufficient evidence on the selection process for committee members to
review test items for bias, noting that no evidence was provided on
whether representatives for students with limited English proficiency were
included. In Idaho, peer reviewers commented that the state did not report
reliability data for student groups, including students with limited
English proficiency. See table 5 for further examples.

14As of July 2006, Education had conducted peer reviews of 50 states and
the District of Columbia. However, detailed peer review notes were
available from only 38 states at the time of our review.

15States were given several years to meet the requirements of the
Improving America's School Act of 1994.

Table 5: Examples of Issues Relating to Assessing Students with Limited
English Proficiency Raised in Education's Peer Review Reports

Validity                   Reliability        Accommodations               
The state's item           The state did not  The state provides a         
development and review     provide any        reasonable list of           
procedures are inadequate  reliability data   accommodations, but does not 
to ensure that the         for each reported  distinguish among those that 
assessments do not reflect subpopulation.     are allowable for students   
unfair irrelevant                             with disabilities, and those 
constructs. For example,                      which are allowable for      
there is no empirical                         students with limited        
evidence that the item                        English proficiency. The     
review process is                             state may wish to provide    
successful in removing                        separate lists of            
barriers due to overly                        accommodations to support    
complex language. Further,                    the selection of appropriate 
a statistical process to                      accommodations that are      
determine bias is                             aligned with instructional   
evaluated only for gender                     approaches for individual    
and race/ethnicity; the                       students. Further, although  
state should consider                         it appears that the state    
using it to evaluate                          has a system in place for    
geographic and demographic                    monitoring the selection of  
diversity, such as                            accommodations for students  
students with limited                         with disabilities, it does   
English proficiency.                          not for students with        
                                                 limited English proficiency. 
The state should conduct a The state does not The state did not provide    
bias review, especially    routinely report   evidence to support the      
for the common items and   subgroup           assertion that               
the alternate assessments  reliability data.  accommodations for students  
for students with limited                     with limited English         
English proficiency and                       proficiency allow for valid  
students with                                 inferences about these       
disabilities.                                 students' knowledge and      
                                                 skills. It does not appear   
                                                 that the state monitors      
                                                 availability of              
                                                 accommodations during test   
                                                 administration. The use of   
                                                 accommodations should be     
                                                 validated on the state's own 
                                                 student population.          

Source: GAO review of Education's peer review notes for 38 states.

Note: This table includes examples from the categories used in Education's
peer review process that determine if a state's assessments are valid,
reliable, fair and accessible, and use appropriate accommodations.

Our group of experts indicated that states are generally not taking the
appropriate set of comprehensive steps to create assessments that produce
valid and reliable results for students with limited English proficiency
and identified essential steps that should be taken. The group noted that
no state has implemented an assessment program for students with limited
English proficiency that is consistent with the Standards for Educational
and Psychological Testing and other technical standards. Specifically, the
group said that students with limited English proficiency are not defined
consistently within and across states, which is a crucial first step to
ensuring the reliability of test results. A reliable test should produce
consistent results, so that students achieve similar scores if tested
repeatedly. If the language proficiency levels of students with limited
English proficiency are classified inconsistently, an assessment may
produce results that appear inconsistent because of the variable
classifications rather than actual differences in skill levels. One expert
noted, for example, that some studies have shown that a student's language
proficiency plays a small role in determining whether a student is
classified as limited English proficient. Inconsistency in defining these
students may be due to variation in how school districts apply state
definitions. For example, according to a 2005 study on students with
limited English proficiency in California, state board of education
guidelines suggest that districts consider a student's performance on the
state's English language proficiency assessment and on the state's
language arts test, a teacher evaluation of the student's academic
performance, and parental recommendations when determining if a student
should or should not continue to be considered limited English proficient.
However, the study noted that districts interpreted and applied these
factors differently.16 Further, it appears that many state assessment
programs do not conduct separate analyses for different groups of limited
English proficient students. Our group of experts indicated that the
reliability of a test may be different for heterogeneous groups of
students with limited English proficiency, such as students who are
literate in their native language and those who are not.

Our group of experts also noted that states are not always explicit about
whether an assessment is attempting to measure skills only (such as
mathematics) or mathematics skills as expressed in English. According to
the group, a fundamental issue affecting the validity of a test is the
definition of what is being measured. Members of the group emphasized that
approaches to ensure valid test results should vary based on which of
these is being measured. For example, North Carolina officials stated that
the state did not offer native language assessments because the state has
explicitly chosen to measure student knowledge in English.

The expert group emphasized that determining the validity and reliability
of academic assessments for students with limited English proficiency is
complicated and requires a comprehensive collection of evidence rather
than a single analysis or review. As one expert noted, "you can't just do
one thing and assume things are valid." In addition, the appropriate
combination of analyses will vary from state to state, depending on the
characteristics of the student population and the type of assessment. For
example, because reliability of test results can vary based on a student's
English proficiency status or a student's native language, states with
more diverse groups of limited English proficient students may need to
conduct additional analyses to ensure sufficient reliability. The group
indicated that states are not universally using all the appropriate
analyses to evaluate the validity and reliability of test results for
students with limited English proficiency. Instead, our experts noted that
states vary in terms of the particular techniques they use for this
purpose, and in the extent to which they collect valid background data.
Members indicated that some states may need assistance to conduct
appropriate analyses that will offer useful information about the validity
of their academic assessments for these students.

16Christopher Jepsen and Shelley de Alth, "English Learners in California
Schools," (San Francisco, California: Public Policy Institute of
California, 2005).

Finally, our group of experts indicated that reducing language complexity
is essential to developing valid assessments for these students, but
expressed concern that some states and test developers do not have a
strong understanding of universal design principles or how to use them to
develop assessments from the beginning to eliminate language that is not
relevant to measuring a student's knowledge of, for example, mathematics.
Members believed that some states may need more information on how to
implement these principles to develop assessments that produce valid
results for students with limited English proficiency.

Accommodations Can Increase Validity of Assessment Results, but Research on
Appropriate Use Is Limited

The majority of states offered some accommodations to try to increase the
validity and reliability of assessment results for students with limited
English proficiency. These accommodations are intended to permit students
with limited English proficiency to demonstrate their academic knowledge,
despite their limited language ability. Our review of state Web sites
found available documentation on accommodations for 42 states. The number
of accommodations offered varied considerably among states. One state, for
example, offered students with limited English proficiency the use of a
bilingual dictionary and a side-by-side English-Spanish version of its
grade 10 mathematics test. Another state listed over 40 acceptable
accommodations, including clarifying test directions in English or the
student's native language, offering extra time, and providing responses
(written or oral) in the student's native language.

Our review found that the most common accommodations offered by these
states were allowing the use of a bilingual dictionary and reading test
items aloud in English (see table 6). In addition, they offered other
accommodations intended to create a less distracting environment for
students, such as administering the assessment to the student in a small
group or individually. Some states also gave students with limited English
proficiency extra time to complete a test to account for their slower
reading speed and information processing time in English. The 5 states we
studied varied in how they established and offered accommodations to
students. For example, Texas officials reported working with its limited
English proficiency focus group to develop a list of allowable
accommodations, which may be offered on a test when they are routinely
used by students in their classrooms. In addition, each school district
has a committee to select particular accommodations based on the needs of
individual students. California officials told us the state provides
guidance to districts on the appropriate use of accommodations. However,
they said that districts might not provide approved accommodations because
of high administrator turnover.

Table 6: Most Frequently Cited Accommodations in 42 States

Accommodation                           Number of states 
Bilingual dictionary                                  32 
Reading items aloud in English                        32 
Small group administration                            29 
Extra time                                            27 
Individual administration                             27 
Separate location                                     25 
Extra breaks                                          25 
Directions in student's native language               24 

Source: GAO review of state documentation.

According to our expert group and our review of the relevant literature,
research is lacking on what specific accommodations are appropriate for
students with limited English proficiency, as well as their effectiveness
in improving the validity of assessment results. A 2004 review of state
policies found that few studies focus on accommodations intended to
address the linguistic needs of students with limited English proficiency
or on how accommodations affect the performance of students with limited
English proficiency.17 In contrast, significantly more research has been
conducted on accommodations for students with disabilities, much of it
funded by Education. Because of this research disparity, our group of
experts reported that some states offer accommodations to students with
limited English proficiency based on those they offer to students with
disabilities, without determining their appropriateness for individual
students. Our experts noted the importance of considering individual
student characteristics to ensure that an accommodation appropriately
addresses the needs of the student. Other researchers have raised similar
issues about the use of accommodations by states.

17Charlene Rivera and Eric Collum. An Analysis of State Assessment
Policies Addressing the Accommodation of English Language Learners. The
George Washington University Center for Equity and Excellence in
Education, Arlington, Virginia: (January 2004).

Education's peer reviews of state academic assessments identified issues
related to accommodations for students with limited English proficiency in
all 38 states reviewed. For example, the reviewers noted that South Dakota
does not clearly indicate whether students with limited English
proficiency were provided accommodations that they do not regularly use in
the classroom. If an accommodation is not used regularly in the classroom,
it may not improve the validity of test results because the student may
not be comfortable with a new procedure. In addition, they noted that
South Dakota does not appear to be monitoring the use of accommodations
and suggested that the state study accommodations to ensure that they are
instructionally appropriate and that they improve the validity and
reliability of the results. In Texas, the reviewers noted that the state
needs to provide information regarding the quality and consistency of
accommodations for students with limited English proficiency-specifically
whether the state routinely monitors the use of accommodations for these
students. In North Carolina, they noted a lack of evidence that the state
has performed research on accommodations. Although conducting such
research could provide useful information on the validity of accommodated
tests, having each state individually study accommodations could be
financially burdensome for them. While research on accommodations for this
population would be useful, it does not have to be conducted directly by
states to be applicable to a state's student population. Further, such
research could involve short-term studies, rather than large-scale,
longitudinal efforts.

Native Language and Alternate Assessments May Improve the Validity of Results
but Are Challenging to Implement

In our survey, 16 states reported that they offered statewide native
language assessments in language arts or mathematics in some grades for
certain students with limited English proficiency in the 2004-2005 school
year. For example, New York translated its statewide mathematics
assessments into Spanish, Chinese, Russian, Korean, and Haitian-Creole. In
addition, 3 states were developing or planning to develop a native
language assessment, and several states allowed school districts to
translate state assessments or offer their own native language
assessments. Our group of experts told us that this type of assessment is
difficult and costly to develop. An assessment provided in a student's
native language is intended to remove language barriers students face in
demonstrating their content knowledge and thereby improve the validity of
test results. Of the 16 states that offered statewide native language
assessments, 4 were able to provide complete data on the number of
students taking native language assessments. These data indicated that
relatively few students took these assessments.

Our group of experts and some state officials also described the
challenges of developing native language assessments that produce valid
results. Members of our expert group and other experts told us that native
language assessments are generally an effective accommodation only for
students in specific circumstances, such as students who are instructed in
their native language or are literate in their native language. In
addition, our experts emphasized that developing valid native language
assessments is challenging, time-consuming, and expensive. Development of
a valid native language assessment involves more than a simple translation
of the original test; in most situations, a process of test development
and validation similar to that of the nontranslated test is recommended to
ensure the validity of the test. In addition, the administration of native
language assessments may not be practicable, for example, when only a
small percentage of limited English students in the state speak a
particular language or when a state's student population has many
languages.

Thirteen states offered statewide alternate assessments (such as reviewing
a student's classroom work portfolio) in 2005 for certain students with
limited English proficiency, based on our review of accountability plans
for all states and the District of Columbia as of March 2006. We also
found that 4 additional states allowed school districts to offer alternate
assessments, while 7 states and the District of Columbia planned to offer
alternate assessments. An official in Wisconsin told us that the state
administers an alternate assessment because developing a native language
assessment for its relatively small Spanish-speaking population would be
impractical and the state does not have bilingual programs in the second
most common language, Hmong (a language that is native to Southeast Asia).
However, our group of experts noted that alternate assessments are
difficult and expensive to develop, and may not be feasible because of the
amount of time required for such an assessment. Members of the group also
expressed concern about the extent to which these assessments are
objective and comparable and can be aggregated with regular assessments.
See figure 4 for information on which states offered native language or
alternate assessments for students with limited English proficiency.

Figure 4: Use of Native Language and Alternate Assessments for Students
with Limited English Proficiency

 Most States Implemented New English Language Proficiency Assessments but Faced
             Challenges Establishing Their Validity and Reliability

With respect to English language proficiency assessments, many states
implemented new tests to address NCLBA requirements, and are working to
align them with newly required state English language proficiency
standards. State and consortia officials reported that states are using
assessments or test items developed by state consortia, customized
assessments developed by testing companies, state-developed assessments,
and off-the-shelf assessments. While a few states already had the required
English language proficiency assessments in place, many states are
implementing them for the first time in spring 2006; as a result, evidence
on their validity and reliability may not be fully developed.

States Are Working with Consortia and Test Developers and Individually to
Develop New English Language Proficiency Assessments

Many states implemented new English language proficiency assessments for
the 2005-2006 school year to meet Education's requirement for states to
administer English language proficiency tests that meet NCLBA requirements
by the spring of 2006.18 These assessments must allow states to track
student progress in learning English; in addition, Education requires that
these assessments be aligned to a state's English language proficiency
standards. According to Education and test development officials, prior to
NCLBA, most states used off-the-shelf English language proficiency
assessments to determine the placement of students in language instruction
programs, but these assessments did not have to be aligned with standards.
Education officials said that because many states did not have tests that
met NCLBA requirements, the agency funded four state consortia to develop
new assessments that were to be aligned with state standards and measure
student progress. Officials in some states told us they have chosen to use
these consortium-developed tests, while officials in other states reported
developing their own tests or continuing to use off-the-shelf tests. Some
states had only recently determined what test they are going to administer
this year, while others may administer a new test in the 2006-2007 school
year. Education officials noted that states' decisions on these tests have
been in flux during this transition year.

In the 2005-2006 school year, 22 states used assessments or test items
developed by one of four state consortia, making this the most common
approach taken by states to develop new English language proficiency
assessments. Each of the four consortia varied somewhat in its development
approach.19 For example, officials in two consortia reported that they
examined all their member states' English language proficiency standards
and reached consensus on core standards for use on the English language
proficiency assessments. They also planned to continue working with member
states in implementing their assessments. For example, one consortium
plans to provide ongoing professional development to help educators
understand the consortium's standards. In contrast, officials in the other
two consortia reported that the consortia disbanded after developing their
assessments. One state official told us that the state hired a contractor
to customize the consortium-developed assessment to more closely align
with state standards. In addition, officials in other states, such as New
Mexico, told us they are using a combination of consortium-developed test
items, along with items developed by another test developer.

18Education officials told us that the agency has approved an extension of
this deadline for 1 state and is currently considering extension requests
from 2 other states.

Fifteen states participated in one of the consortia, but officials in
these states told us they chose not to use the assessments developed by
the consortia in the 2005-2006 school year for a variety of reasons,
including lack of alignment with state standards, the length of the
assessment, and the cost of implementation. For example, Kentucky chose
not to use the consortium assessment because of cost effectiveness
concerns and lack of alignment with state standards. Another state decided
not to use the consortium-developed assessment, as officials were
concerned about its cumbersome nature and associated cost. Officials in
some states told us they plan to use consortium-developed assessments in
the future. For example, Florida officials reported that the state will
use a consortium assessment in the 2006-2007 school year. Appendix V shows
the states that participated in the consortia and which used
consortia-developed assessments in the 2005-2006 school year.

Officials in states that did not use consortia assessments told us that
they used other approaches to develop their English language proficiency
assessments. Eight states worked with test developers to augment
off-the-shelf English language proficiency assessments to incorporate
state standards. For example, Mississippi, South Dakota, and Wyoming are
using versions of an English language proficiency assessment that has been
augmented to align to their respective state standards. Officials in 14
states indicated that they are administering off-the-shelf assessments.
These officials indicated varying degrees of alignment between the
off-the-shelf tests being used and their state's English language
proficiency standards; in 11 of these states, the assessment has not been
fully aligned with state standards.20 Seven states, including Texas,
Minnesota, and Kansas, created their own English language proficiency
assessments. Officials in these states said they typically worked with a
test developer or research organization to create the assessments. See
figure 5 and appendix VI for more detailed information on the English
language proficiency assessments used by each state.

19The four consortia include the World-Class Instructional Design and
Assessment (WIDA) Consortium, State Collaborative on Assessment and
Student Standards (SCASS) Consortium, Mountain West Assessment Consortium
(MWAC), and Pennsylvania Enhanced Assessment Grant (PA EAG) Consortium.

20Although these assessments are not fully aligned to state standards,
Education officials noted that they have not yet provided states guidance
on what level of alignment the agency expects. Thus, some of these may
ultimately meet Education's requirements.

Figure 5: Type of English Language Proficiency Assessment Administered in
2005-2006 School Year

Some officials in our 5 study states and 28 additional states we contacted
to determine what English language proficiency assessment they planned to
use in 2006 pointed to some challenges involving their English language
proficiency assessments. Some of these state officials expressed concerns
about using both their old and new English language proficiency
assessments to measure student progress in learning English. NCLBA
required states to begin tracking student progress in the 2002-2003 school
year, before most states had implemented their new English language
proficiency assessments. In May 2006, Education officials told us that
states must rely on baseline results from their old tests and determine
how results from their old tests relate to results from their new tests in
order to track student progress since 2003, as required by NCLBA. They
noted that states may change their English language proficiency goals
based on results from their new assessments, but they cannot change the
initial baseline established with their old test. In its technical
comments on this report, Education noted that it allows states to make
such determinations in a variety of ways, as long as annual progress is
reported. Officials in some states want to rely solely on data from their
new tests to track student progress. They stated that, unlike their old
tests, their new tests provide more accurate data on student progress
because they are aligned to their English language proficiency standards
and were designed to measure student progress. Officials from other states
questioned the usefulness of conducting studies to determine the
relationship between their old and new tests, especially in states that
had previously used multiple English language proficiency assessments.

Officials in a few of our study states also expressed concern about the
appropriateness of NCLBA's requirement to assess students with limited
English proficiency in kindergarten and the first and second grades. For
example, Texas officials told us traditional tests do not produce good
test results for students this young in part because of their limited
attention spans. In addition, officials in Texas and North Carolina noted
that English proficient students in these grades are not required to be
assessed in the same way.

Many States Are Still in the Process of Establishing the Validity and
Reliability of English Language Proficiency Assessments

Officials in our study states and test developers we interviewed reported
that they commonly apply generally accepted test development procedures in
the development of English language proficiency assessments, but some are
still in the process of documenting the validity and reliability of these
assessments. For example, some evidence needed to confirm the validity and
reliability of the test can be obtained only after the assessment has been
fully administered. One consortium contracted with a research organization
to assess the validity and reliability testing of its English language
proficiency assessment. According to a consortium official, the research
organization performed all of the standard steps that are taken to ensure
high-quality assessments. These included piloting and field testing the
assessment and conducting statistical modeling. An official from another
consortium said that its test vendor is conducting basic psychometric
research and analyzing field test data for evidence of reliability.
California officials noted that the process for developing and ensuring
the validity and reliability of its English language proficiency
assessment is similar to that used for its state academic assessments.

Although states have taken steps toward determining validity, documenting
the validity and reliability of a new assessment is an ongoing process. A
2005 review of the documentation of 17 English language proficiency
assessments used by 33 states in the 2005-2006 school year found that the
evidence presented on validity and reliability was generally
insufficient.21 The report, which was funded by Education, reviewed
documentation for consortium-developed assessments, off-the-shelf
assessments, and custom-developed assessments for evidence of validity,
reliability, and freedom from test bias, among other things. It found that
the technical adequacy of English language proficiency assessments is
undeveloped compared to the adequacy of assessments for general education.
The study noted that none of the assessments contained "sufficient
technical evidence to support the high-stakes accountability information
and conclusions of student readiness they are meant to provide."

In addition, many states are in the process of aligning these assessments
to state English language proficiency standards, which in turn must be
aligned to state content standards. These steps are needed to comply with
NCLBA requirements. Alignment, which refers to the degree to which an
assessment's items measure the content they are intended to measure, is
critical in assuring the validity of an assessment. Officials in some
states22 have expressed uncertainty about how to align their English
language proficiency test with their standards for academic subjects, such
as mathematics and science.23 Officials in 2 states told us that their
English language proficiency assessments are aligned to state language
arts standards but are not aligned to state mathematics standards, meaning
that the assessment may not measure the language needed to succeed in a
mathematics class. Findings from Education's Title III monitoring reviews
of 13 states indicated that 8 states had not yet fully completed
alignment; of these, 5 had not yet linked their English language
proficiency and academic content standards, while 5 had not yet aligned
their English language proficiency assessments with their English language
proficiency standards.24

21Stanley Rabinowitz and Edynn Sato, "Evidence-Based Plan: Technical
Adequacy of Assessments for Alternate Student Populations: A Technical
Review of High-Stakes Assessments for English Language Learners," WestEd
(December 2005).

22The states providing these comments represent more than our 5 case study
states. We also contacted officials in 28 additional states to determine
what English language proficiency assessment they planned to use in 2006.

23Education does not require state English language proficiency tests to
be aligned to state academic standards. However, states' English language
proficiency tests and academic standards are connected, in that Education
requires that state's English language proficiency tests be aligned to
state English language proficiency standards and NCLBA requires that state
English language proficiency standards be aligned with state academic
standards.

Education Has Provided Assistance, but States Reported Need for Additional
                            Guidance and Flexibility

Education has offered states a variety of technical assistance to help
them appropriately assess students with limited English proficiency, such
as providing training and expert reviews of their assessment systems, as
well as flexibility in assessing these students. However, Education has
issued little written guidance on how states are expected to assess and
track the English proficiency of these students, leaving state officials
unclear about Education's expectations. To support states' efforts to
incorporate these students into their accountability systems, Education
has offered states some flexibilities in how they track progress goals for
these students. However, many of the state and district officials we
interviewed told us that the current flexibilities do not fully account
for some characteristics of certain students in this student group, such
as their lack of previous schooling. These officials indicated that
additional flexibility is needed to ensure that the federal progress
measures accurately track the academic progress of these students.

Education Has Provided a Variety of Support on Assessment Issues but Little
Written Guidance on Assessing Students with Limited English Proficiency

Education offers support in a variety of ways to help states meet NCLBA's
requirements for assessing students with limited English proficiency for
both their language proficiency and their academic knowledge. Some of
these efforts focus specifically on students with limited English
proficiency, while others, such as the Title I monitoring visits, focus on
all student groups and on broader compliance issues but review some
assessment issues related to students with limited English proficiency as
part of their broader purposes. The agency's primary technical assistance
efforts have included the following:

24We reviewed reports from Title III monitoring visits of 13 states
conducted between April and October 2005 that were available from
Education as of March 31, 2006.

           o  Title I peer reviews of states' academic standards and
           assessment systems: Education is currently conducting peer reviews
           of the academic assessments that states use in measuring adequate
           yearly progress. During these reviews, three independent experts
           review evidence provided by the state about the validity and
           reliability of these assessments (including whether the results
           are valid and reliable for students with limited English
           proficiency) and make recommendations to Education about whether
           the state's assessment system is technically sufficient and meets
           all legal requirements. Education shares information from the peer
           review to help states address issues identified during the review.
           Education has imposed a deadline requiring that states receive
           peer review approval by June 30, 2006, but only 10 states have had
           their assessment systems fully approved by Education as of that
           date.25

           o  Title III monitoring visits: Education began conducting site
           visits to review state compliance with Title III requirements in
           2005 and has visited 15 states. Education officials reported that
           they plan to visit 11 more states in 2006. As part of these
           visits, the agency reviews the state's progress in developing
           English language proficiency assessments that meet NCLBA
           requirements.

           o  Comprehensive centers: Education has contracted with 16
           regional comprehensive centers to build state capacity to help
           districts that are not meeting their adequate yearly progress
           goals. The grants for these centers were awarded in September
           2005, and the centers provide a broad range of assistance,
           focusing on the specific needs of individual states. At least 3 of
           these centers plan to assist individual states in developing
           appropriate goals for student progress in learning English. In
           2005, Education also funded an assessment and accountability
           comprehensive center, which provides technical assistance to the
           regional comprehensive centers on issues related to the assessment
           of students, including those with limited English proficiency.

           o  Ongoing technical assistance for English language proficiency
           assessments: Education has provided information and ongoing
           technical assistance to states using a variety of tools and has
           focused specifically on the development of the English language
           proficiency standards and assessments required by NCLBA. These
           include:

                        o  a semiannual review of reports states submit to
                        Education and phone calls to state officials focused
                        on state progress in developing their English
                        language proficiency assessments;
                        o  on-site technical assistance to states regarding
                        their English language proficiency assessments;
                        o  an annual conference focused on students with
                        limited English proficiency that includes sessions on
                        assessment issues, such as aligning English language
                        proficiency and academic content standards;
                        o  videoconference training sessions for state
                        officials on developing English language proficiency
                        assessments;
                        o  providing guidance on issues related to students
                        with limited English proficiency on its Web site;
                        o  distributing information through an electronic
                        bulletin board and a weekly electronic newsletter
                        focused on students with limited English proficiency;
                        o  disseminating information through the National
                        Clearinghouse for English Language Acquisition and
                        Language Instruction Educational Programs;
                        o  semiannual meetings and training sessions with
                        state Title III directors; and
                        o  responding to questions from individual states as
                        needed.

           o  Enhanced Assessment Grants: Since 2003,  Education has awarded
           these grants, authorized by NCLBA, to support state activities
           designed to improve the validity and reliability of state
           assessments. According to an Education official, most of the
           grants up to now have funded the English language proficiency
           consortia, although some grants have been used to conduct research
           on accommodations. For grants to be awarded in 2006, Education
           will give preference to projects involving accommodations and
           alternate assessments intended to increase the validity of
           assessments for students with limited English proficiency and
           students with disabilities. 

           o  Title I monitoring visits: As part of its monitoring visits to
           review state compliance with Title I requirements, Education
           reviews some aspects of the academic assessments administered by
           states, but in less detail than during its peer reviews. During
           these visits, for example, states may receive some feedback on how
           the state administers academic assessments to students with
           limited English proficiency and the appropriateness of
           accommodations offered to these students. Education staff also
           reported that they respond to questions about Title I requirements
           from individual states as needed.

           While providing states with a broad range of technical assistance
           and guidance through informal channels, Education has issued
           little written guidance on developing English language proficiency
           assessments that meet NCLBA's requirements and on tracking the
           progress of students in acquiring English. Education issued some
           limited nonregulatory guidance on NCLBA's basic requirements for
           English language proficiency standards and assessments in February
           2003. However, officials in about one-third of the 33 states we
           visited or directly contacted expressed uncertainty about
           implementing these requirements. They told us that they would like
           more specific guidance from Education to help them develop tests
           that meet NCLBA requirements, generally focusing on two issues.
           First, some officials said they were unsure about how to align
           English language proficiency standards with content standards for
           language arts, mathematics, and science, as required by NCLBA. An
           official in 1 state said the state needed specific guidance on
           what Education wants from these assessments, such as how to
           integrate content vocabulary on the English language proficiency
           assessment without creating an excessively long test. In another
           state, officials explained that the state was developing its
           English language proficiency test by using an off-the-shelf test
           and incorporating additional items to align the test with the
           state's English language proficiency and academic standards.
           However, the state discovered that it had not correctly augmented
           the test and will consequently have to revise the test. Officials
           in this state noted that they have had to develop this test
           without a lot of guidance from Education.

           Second, some officials reported that they did not know how to use
           the different scores from their old and new English language
           proficiency assessments to track student progress. For example, an
           official in 1 state said that she would like guidance from
           Education on how to measure student progress in English language
           proficiency using different tests over time. Another official was
           unsure if Education required a formal study to correlate the
           results from their old and new English language proficiency
           assessments, noting that more specific guidance would help them
           better understand Education's requirements. Without guidance and
           specific examples on both of these issues, some of these officials
           were concerned that they will spend time and resources developing
           an assessment that may not meet Education's requirements.

           Education officials told us that they are currently developing
           additional nonregulatory guidance on these issues, but it has not
           been finalized. They also pointed out that they have provided
           extensive technical assistance on developing English language
           proficiency standards and assessments, and have clearly explained
           the requirements to state officials at different meetings on
           multiple occasions. An Education official acknowledged that states
           were looking for more guidance on the degree of alignment required
           between their English language proficiency assessments and
           standards, noting that Education is still considering the issue.
           She stated that the issue would be addressed in the guidance it
           plans to issue in the future.

           With respect to academic content assessments, our group of experts
           reported that some states could use more assistance in creating
           valid academic assessments for students with limited English
           proficiency. While 4 of the 5 states we studied in depth had
           significant experience in, and multiple staff devoted to,
           developing language arts and mathematics assessments, some members
           of our expert group pointed out that the assessment departments in
           other states have limited resources and expertise, as well as high
           turnover. As a result, these states need help to conduct
           appropriate analyses that will offer useful information about the
           validity and reliability of their academic assessments for
           students with limited English proficiency. An Education official
           told us that the agency recently began offering technical
           assistance to states that need help addressing issues raised
           during their peer reviews.

           Our group of experts suggested several areas where states could
           benefit from additional assistance and guidance in developing
           academic assessments for students with limited English
           proficiency. Several members noted the lack of good research on
           what kinds of accommodations can help mitigate language barriers
           for students with limited English proficiency. Several experts
           also believed that some states need more information on how to
           implement universal design principles to develop assessments that
           produce valid results for students with limited English
           proficiency. In addition, some group members pointed out that
           developing equivalent assessments in other languages (that is,
           assessments that measure the same thing and are of equivalent
           difficulty) is challenging and that states need more information
           about how to develop such assessments, as well as examples.

           Education Has Offered Different Accountability Options for Students
			  with Limited English Proficiency, but State Officials Reported
			  Additional Flexibility Is Needed
			  
			  Education has offered states several flexibilities in tracking
           academic progress goals for students with limited English
           proficiency to support their efforts to develop appropriate
           accountability systems for these students. In a February 2004
           notice, Education recognized the existence of language barriers
           that hinder the assessment of students who have been in the
           country for a short time and provided some testing flexibility for
           these students. Specifically, Education does not require students
           with limited English proficiency to participate in a state's
           language arts assessment during their first year in U.S. schools.
           In addition, while these students must take a state's mathematics
           assessment during their first year in U.S. schools, a state may
           exclude their scores in determining whether it met its progress
           goals.26

           Education offered additional flexibility in its February 2004
           notice, recognizing that limited English proficiency is a more
           transient quality than having a disability or being of a
           particular race. Unlike the other NCLBA student groups, students
           who achieve English proficiency leave the group at the point when
           they are more prepared to demonstrate their academic knowledge in
           English, while new students with lower English proficiency are
           constantly entering the group (see fig. 6). Given the group's
           continually changing composition, meeting progress goals may be
           more difficult than doing so for other student groups, especially
           in districts serving large numbers of students with limited
           English proficiency. To compensate for this, Education allowed
           states to include, for up to 2 years, the scores of students who
           were formerly classified as limited English proficient when
           determining whether a state met its progress goals for students
           with limited English proficiency. In addition, Education has
           approved requests from several states to permit students who have
           been redesignated as English proficient to remain in the group of
           students with limited English proficiency until they have achieved
           the proficient level on the state's language arts assessment for 1
           or more years.

25Education has sent letters to the remaining states outlining the issues
that need to be resolved in order for their assessment systems to be
approved.

26On June 24, 2004, Education issued proposed regulations on these
flexibilities for students with limited English proficiency for public
comment, but the regulations have not been finalized as of June 2006.

Figure 6: Movement of Students In and Out of Limited English Proficient
Student Group and Other Student Groups

Several state and local officials in our study states told us that
additional flexibility would be helpful to ensure that the annual progress
measures provide meaningful information about the performance of students
with limited English proficiency. Officials in 4 of the states we studied
suggested that certain students with limited English proficiency should be
exempt for longer periods from taking academic content assessments or that
their test results should be excluded from a state's annual progress
determination for a longer period than is currently allowed. Several
officials voiced concern that some of these students have such poor
English skills or so little previous school experience that the assessment
results do not provide any meaningful information. Instead, some of these
officials stated that students with limited English proficiency should not
be included in academic assessments until they demonstrate appropriate
English skills on the state's English language proficiency assessment.
However, the National Council of La Raza, an Hispanic advocacy
organization, has voiced concern that excluding too many students with
limited English proficiency from a state's annual progress measures will
allow some states and districts to overlook the needs of these students.
Education officials reported that they are developing a regulation with
regard to how test scores for this student group are included in a state's
annual progress measures, but it has not yet been finalized.

With respect to including the scores of students previously classified as
limited English proficient in a state's progress measures for this group
for up to 2 years, officials in 2 of our 5 study states, as well as one
member of our expert group, thought it would be more appropriate for these
students to be counted in the limited English proficient group throughout
their school careers-but only for accountability purposes. They pointed
out that by keeping students formerly classified as limited English
proficient in the group, districts that work well with these students
would see increases in the percentage who score at the proficient level in
language arts and mathematics. An Education official explained that the
agency does not want to label these students as limited English proficient
any longer than necessary and considered including test results for these
students for 2 years after they have achieved English proficiency to be
the right balance. Education officials also noted that including all
students who were formerly limited English proficient would inflate the
achievement measures for the student group.

District officials in 4 of the states we studied argued that tracking the
progress of individual students in this group is a better measure of how
well these students are progressing academically. Officials in one
district pointed to a high school with a large percentage of students with
limited English proficiency that had made tremendous progress with these
students, doubling the percentage of students achieving academic
proficiency. The school missed the annual progress target for this group
by a few percentage points, but school officials said that the school
would be considered successful if it was measured by how much individual
students had improved in their test scores. A district official in another
state explained that many students with limited English proficiency
initially have very low test scores, but demonstrate tremendous
improvement in these scores over time. In response to educators and
policymakers who believe such an approach should be used for all students,
Education initiated a pilot project in November 2005, allowing a limited
number of states to incorporate measures of student progress over time in
determining whether districts and schools met their annual progress goals.
Even using this approach, however, states must still establish annual
goals that lead to all students achieving proficient scores by 2014.27

                                  Conclusions

NCLBA has focused attention on the academic performance of all students,
especially those who have historically not performed as well as the
general student population, such as students with limited English
proficiency. NCLBA requires states to include these students in their
language arts and mathematics assessments and to assess them in a valid
and reliable manner, and states are in various stages of doing so.
Although Education has provided some technical assistance to states, our
group of experts and others have noted the complexity of developing
academic assessments for these students and have raised concerns about the
technical expertise of states to ensure the validity and reliability of
assessment results. Using assessment results that are not a good measure
of student knowledge is likely to lead to poor measures of state and
district progress, thereby undermining NCLBA's purpose to hold schools
accountable for student progress. Further, although most states offered
these students accommodations, research on their appropriateness is
limited. National research on accommodations has informed states'
practices in assessing students with disabilities. Without similar
research efforts, accommodations offered to students with limited English
proficiency may not improve the validity of their test results.

While Education has provided some support and training to states,
officials in a number of states are still uncertain about how to comply
with some of the more technical requirements of the new English language
proficiency assessments required by NCLBA. State officials reported that
they need more guidance from Education to develop these assessments.
States have had to develop many new assessments under NCLBA for both
English language proficiency and academic content, and some states may
lack the technical expertise to develop assessments that produce valid
results for students with limited English proficiency. Without more
specific guidance outlining Education's requirements, states may spend
time developing English language proficiency assessments that do not
adequately track student progress in learning English or otherwise meet
NCLBA's requirements.

27See GAO, No Child Left Behind Act: States Face Challenges Measuring
Academic Growth That Education's Initiatives May Help Address, GAO-06-661
(Washington, D.C.: July 17, 2006) for further information on Education's
pilot project.

Including students with limited English proficiency in NCLBA's
accountability framework presents unique challenges. For example, students
who have little formal schooling may make significant progress in learning
academic skills, but may not achieve proficiency on state academic
assessments for several years. The movement of students into and out of
the group also makes it more difficult for the group to meet state
progress goals, even when these students are making academic progress.
Education has addressed some of the unique characteristics of this student
group and provided some flexibility in how states and districts are held
accountable for the progress of these students. However, these current
flexibilities may not fully account for the characteristics of certain
students with limited English proficiency, such as those who have little
previous formal schooling.

                      Recommendations for Executive Action

We recommend that the Secretary of Education

           1. Support additional research on appropriate accommodations for
           students with limited English proficiency and disseminate
           information on research-based accommodations to states.
           2. Determine what additional technical assistance states need with
           respect to assessing the academic knowledge of students with
           limited English proficiency and to improve the validity and
           reliability of their assessment results (such as consultations
           with assessment experts and examples of assessments targeted to
           these students) and provide such additional assistance.
           3. Publish additional guidance with more specific information on
           the requirements for assessing English language proficiency and
           tracking the progress of students with limited English proficiency
           in learning English.
           4. Explore ways to provide additional flexibilities to states in
           terms of holding states accountable for students with limited
           English proficiency. For example, among the flexibilities that
           could be considered are

                        o  allowing states to include the assessment scores
                        for all students formerly considered to have limited
                        English proficiency in a state's annual progress
                        results for the group of students with limited
                        English proficiency,

                        o  extending the period during which the assessment
                        scores for some or all students with limited English
                        proficiency would not be included in a state's annual
                        progress results, and

                        o  adjusting how states account for recent immigrants
                        with little formal schooling in their annual progress
                        results.

                                Agency Comments

We provided a draft of this report to Education for review and comment.
The agency provided comments, which are reproduced in appendix VII.
Education also provided technical clarifications, which we incorporated
when appropriate. Education agreed with our first three recommendations.
The department noted that it has conducted some research on the
effectiveness of accommodations and is currently working with its National
Research and Development Center for Assessment and Accountability to
synthesize the existing research literature on the assessment of students
with limited English proficiency. Education also explained that it has
begun the process of identifying the additional technical assistance needs
of states with respect to academic assessments; specifically, it will have
its Assessment and Accountability Comprehensive Center conduct a needs
assessment this fall to determine specific areas in which states need
assistance and will provide technical assistance to address those areas.
In addition, the department stated that it is exploring ways to help
states assess English language proficiency.

Education did not explicitly agree or disagree with our fourth
recommendation. Instead, the agency commented that it has explored and
already provided various types of flexibility regarding the inclusion of
students with limited English proficiency in accountability systems.
Further, Education noted that it is in the process of completing a
regulation on flexibility for these students. However, the department also
emphasized that all students with limited English proficiency must be
included in school accountability systems to improve both instruction and
achievement outcomes. Through our recommendation, we encourage the
department to continue its efforts.

We are sending copies of this report to the Secretary of Education,
relevant congressional committees, and other interested parties. We will
make copies available to others upon request. In addition, the report will
be available at no charge on GAO's Web site at http://www.gao.gov .

If you or your staff have any questions or wish to discuss this report
further, please contact me at (202) 512-7215 or at [email protected] .
Contact points for our Offices of Congressional Relations and Public
Affairs may be found on the last page of this report. Other contacts and
major contributors are listed in appendix VIII.

Marnie S. Shaul
Director, Education, Workforce, and Income Security Issues

Appendix I: GAO's Group of Experts on Assessing the Academic Knowledge
of Students with Limited English Proficiency

On January 20, 2006, GAO, with the assistance of the National Academy of
Sciences, convened a group of experts in Davis, California, to discuss
issues related to assessing the academic knowledge of students with
limited English proficiency. Specifically, we asked the group to discuss
the following questions:

           o  To meet the requirements of the No Child Left Behind Act
           (NCLBA), what steps should states take to ensure the validity and
           reliability of language arts and mathematics assessments for
           students with limited English proficiency?

           o  What steps should states take to ensure that students with
           limited English proficiency receive appropriate accommodations on
           language arts and mathematics assessments?

           o  Given NCLBA's accountability framework, what is the most
           appropriate way to hold schools and districts accountable for the
           performance of students with limited English proficiency?

           o  How can the U.S. Department of Education assist states in their
           efforts to meet NCLBA's assessment and accountability requirements
           for students with limited English proficiency?

Group members who were selected had significant technical and research
expertise in assessments issues. Some members had technical expertise on
general assessment issues, while others had specifically conducted
assessment research focused on students with limited English proficiency.
The members of our expert group are listed below:

Dr. Jamal Abedi University of California, Davis

Dr. Stephen Dunbar The University of Iowa

Dr. Richard Duran University of California, Santa Barbara

Dr. Steven Ferrara American Institutes for Research

Dr. Patricia Gandara University of California, Davis

Dr. Edward Haertel Stanford University

Dr. Rebecca Kopriva University of Maryland

Dr. Stanley Rabinowitz WestEd

Dr. Charlene Rivera The George Washington University

Dr. Rebecca Zwick University of California, Santa Barbara

Appendix II: Determining Adequate Yearly Progress for Student Groups

NCLBA requires states to report adequate yearly progress (AYP) results at
the state level for each of the required student groups, including
students with limited English proficiency. The law also requires
Education, starting in the 2004-2005 school year, to make an annual
determination about whether states have made adequate yearly progress for
each student group.1 Education has issued some general regulations
regarding state-level adequate yearly progress. However, Education has not
yet collected any such state-level adequate yearly progress results and
has not issued any guidance on how states should determine whether a
student group has made adequate yearly progress. As a result, some states
have not yet made adequate yearly progress determinations for student
groups at the state level.

In order for a student group, such as students with limited English
proficiency, to make adequate yearly progress, it must make a number of
different goals. Specifically:

           o  At least 95 percent of students in the group must take the
           state's language arts and mathematics assessments, and

           o  The student group must meet the progress goals established by
           the state for both language arts and mathematics proficiency or

           o  The percentage of students who did not achieve proficient
           scores must have decreased by at least 10 percent from the
           previous year, and the student group must also meet the progress
           goals established by the state for its other academic indicator
           (graduation rate for high schools and usually attendance rate for
           other schools).

Figure 7 illustrates the basic decision process for determining adequate
yearly progress for a student group.

1NCLBA also requires Education to annually determine whether states have
met their Title III goals related to increases in students making progress
in learning English and attaining English proficiency.

Figure 7: Process for Determining Adequate Yearly Progress for a Student
Group

Because states have different assessment systems, they use different
methods for determining adequate yearly progress. A state can have an
assessment system that allows it to create the same progress goal for
mathematics and language arts for all grades, despite using different
tests in each grade. In this case, the state could review data for all
students in a student group across the state to determine if the group met
its annual progress goals. A state can also establish different progress
goals for different grades or groups of grades, depending on the
particular test being used. In this case, according to an Education
official, a state would have to meet all the proficiency and participation
goals for all the different grades or groups of grades in order to make
adequate yearly progress.

Appendix III: Percentage of Districts Making AYP Goals for Mathematics
for Students with Limited English Proficiency

Notes: Data are for school year 2003-2004.

We requested district-level achievement data from 20 states, and 18 states
responded to our request.

When districts reported proficiency data for different grades or groups of
grades, we determined that the percentage of students with limited English
proficiency met a state's mathematics progress goal if the student group
met the goal for all grades reported.

Results from charter schools are included when a charter school is its own
school district or part of a larger school district.

Hawaii only has one school district.

Appendix IV: Proficiency Scores on Mathematics Tests for All Students
and Students with Limited English Proficiency

Notes: New York and Tennessee did not provide assessment data in their
state consolidated performance reports for the 2003-2004 school year.

The results for Arkansas do not include those students with limited
English proficiency who were considered proficient based on the state's
portfolio assessment.

The total student population includes students with limited English
proficiency.

aMost states reported assessment data for students in the fourth grade.
States marked with this superscript reported on some other grade at the
elementary school level, usually either third grade or fifth grade.

Appendix V: Enhanced Assessment Consortia Participation 

                                  State                                       
                                  Collaborative                               
                World-Class       on Assessment   Mountain                    
                Instructional     and Student     West         Pennsylvania   
                Design and        Standards       Assessment   Enhanced
                Assessment (WIDA) (SCASS)         Consortium   Assessment
                Consortium        Consortium      (MWAC)       Grant (PA EAG)
Assessment   Assessing         English         MWAC         Comprehensive  
                Comprehension and Language                     English        
                Communication in  Development                  Language       
                English           Assessment                   Learning       
                State-to-State    (ELDA)                       Assessment     
                for English                                    (CELLA)        
                Language Learners                              
                (ACCESS for ELLs)                              
Consortia                                                   
states                                                      
Using        Alabama Delaware  Iowa Louisiana  Idahoa New   Tennessee      
assessment   District of       Nebraska Ohio   Mexicoa      
in 2005-2006 Columbia Georgia  South Carolina  Michigan a b 
school year  Illinois Maine    West Virginia                
                New Hampshire New                              
                Jersey Oklahoma                                
                Rhode Island                                   
                Vermont Wisconsin                              
Not using    none              Kentucky Nevada Alaska       Florida        
assessment                     North Carolina  Colorado     Maryland       
in 2005-2006                   Texas           Montana      Michiganb      
school year                                    Nevada North Pennsylvania   
                                                  Dakota Utah  
                                                  Wyoming      

Source: Interviews with consortia and state officials.

Note: This table reflects states that participated in the consortia prior
to or during the 2005-2006 school year. Some states are no longer
consortia members.

aUsing test items from consortium-developed assessment.

bParticipated in more than one consortium.

Appendix VI: English Language Proficiency Assessments Used in the
2005-2006 School Year, by State 

State                English Language Proficiency Assessment               
Alabama              Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
Alaska               IDEA Proficiency Test                                 
Arizona              Stanford English Language Proficiency Test            
Arkansas             MAC II (Maculaitis Assessment of Competencies) Test   
                        of English Language Proficiency                       
California           California English Language Development Test          
Colorado             Colorado English Language Assessment                  
Connecticut          LAS (Language Assessment System) Links                
Delaware             Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
District of Columbia Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
Florida              Various off-the-shelf testsa                          
Georgia              Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
Hawaii               LAS (Language Assessment System) Links                
Idaho                Mountain West Assessment Consortium test items        
Illinois             Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
Indiana              LAS (Language Assessment System) Links                
Iowa                 English Language Development Assessment (SCASS)       
Kansas               Kansas English Language Proficiency Assessment        
Kentucky             2004 IDEA Proficiency Test or Language Assessment     
                        Scales (LAS)a                                         
Louisiana            English Language Development Assessment (SCASS)       
Maine                Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
Maryland             LAS (Language Assessment System) Links                
Massachusetts        Massachusetts English Proficiency Assessment          
Michigan             English Language Proficiency Assessmentb (includes    
                        Mountain West Consortium test items)                  
Minnesota            Test of Emerging Academic English, Minnesota Student  
                        Oral Language Observation Matrix, and checklist for   
                        reading and writing for K-2 students                  
Mississippi          Stanford English Language Proficiency Test            
Missouri             MAC II (Maculaitis Assessment of Competencies) Test   
                        of English Language Proficiency                       
Montana              Iowa Test of Basic Skills, Woodcock-Munoz Language    
                        Survey (English), or other state-approved testa       
Nebraska             English Language Development Assessment (SCASS)       
Nevada               LAS (Language Assessment System) Links                
New Hampshire        Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
New Jersey           Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
New Mexico           New Mexico English Language Proficiency Assessment    
                        (includes Mountain West Consortium test items)        
New York             New York State English as a Second Language           
                        Achievement Test                                      
North Carolina       IDEA Proficiency Test                                 
North Dakota         2004 IDEA Proficiency Test, Woodcock-Munoz Language   
                        Survey (English), and Language Assessment Scales      
                        (LAS)a                                                
Ohio                 English Language Development Assessment (SCASS)       
Oklahoma             Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
Oregon               English Language Proficiency Assessmentb              
Pennsylvania         Stanford English Language Proficiency Test            
Rhode Island         Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
South Carolina       English Language Development Assessment (SCASS)       
South Dakota         Dakota English Language Proficiency assessment        
Tennessee            Comprehensive English Language Learning Assessment    
                        (PA EAG)                                              
Texas                Texas English Language Proficiency Assessment System; 
                        consists of Reading Proficiency Tests in English and  
                        Texas Observation Protocols                           
Utah                 2004 IDEA Proficiency Test                            
Vermont              Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
Virginia             Stanford English Language Proficiency Test            
Washington           Washington Language Proficiency Test                  
West Virginia        English Language Development Assessment (SCASS)       
Wisconsin            Assessing Comprehension and Communication in English  
                        State-to-State for English Language Learners (WIDA)   
Wyoming              Wyoming English Language Learner Assessment           

Source: Interviews with consortia and state officials.

aState allows school districts to individually choose tests.

bAssessments are not the same; Oregon's is a state developed assessment,
while Michigan's is a combination of an augmented version of the Stanford
English Language Proficiency Test and test items from the Mountain West
Assessment Consortium.

Appendix VII: Comments from the Department of Education 

Appendix VIII: GAO Contacts and Acknowledgments

GAO Contacts

Marnie Shaul, (202) 512-7215, [email protected] and Cornelia Ashby (202)
512-7215, [email protected]

Staff Acknowledgments

Harriet Ganson (Assistant Director) and Michelle St. Pierre
(Analyst-in-Charge) managed all aspects of this assignment. Shannon Groff,
Eileen Harrity, and Krista Loose made significant contributions to this
report. Katie Brillantes contributed to the initial design of the
assignment. Carolyn Boyce, John Mingus, and Lynn Musser provided key
technical support, James Rebbe provided legal support, and Scott Heacock
assisted in message and report development.

Related GAO Products

No Child Left Behind Act:, States Face Challenges Measuring Academic
Growth That Education's Initiatives May Help Address. GAO-06-661 .
Washington, D.C.: July 17, 2006.

No Child Left Behind Act: Improved Accessibility to Education's
Information Could Help States Further Implement Teacher Qualification
Requirements. GAO-06-25 . Washington, D.C.: November 21, 2005.

No Child Left Behind Act: Education Could Do More to Help States Better
Define Graduation Rates and Improve Knowledge about Intervention
Strategies. GAO-05-879 . Washington, D.C.: September 20, 2005.

No Child Left Behind Act: Most Students with Disabilities Participated in
Statewide Assessments, but Inclusion Options Could Be Improved. GAO-05-618
. Washington, D.C.: July 20, 2005.

Head Start: Further Development Could Allow Results of New Test to Be Used
for Decision Making. GAO-05-343 . Washington, D.C.: May 17, 2005.

No Child Left Behind Act: Education Needs to Provide Additional Technical
Assistance and Conduct Implementation Studies for School Choice Provision.
GAO-05-7 . Washington, D.C.: December 10, 2004.

No Child Left Behind Act: Improvements Needed in Education's Process for
Tracking States' Implementation of Key Provisions. GAO-04-734 .
Washington, D.C.: September 30, 2004.

(130478)

GAO's Mission

The Government Accountability Office, the audit, evaluation and
investigative arm of Congress, exists to support Congress in meeting its
constitutional responsibilities and to help improve the performance and
accountability of the federal government for the American people. GAO
examines the use of public funds; evaluates federal programs and policies;
and provides analyses, recommendations, and other assistance to help
Congress make informed oversight, policy, and funding decisions. GAO's
commitment to good government is reflected in its core values of
accountability, integrity, and reliability.

Obtaining Copies of GAO Reports and Testimony

The fastest and easiest way to obtain copies of GAO documents at no cost
is through GAO's Web site ( www.gao.gov ). Each weekday, GAO posts newly
released reports, testimony, and correspondence on its Web site. To have
GAO e-mail you a list of newly posted products every afternoon, go to
www.gao.gov and select "Subscribe to Updates."

Order by Mail or Phone

The first copy of each printed report is free. Additional copies are $2
each. A check or money order should be made out to the Superintendent of
Documents. GAO also accepts VISA and Mastercard. Orders for 100 or more
copies mailed to a single address are discounted 25 percent. Orders should
be sent to:

U.S. Government Accountability Office 441 G Street NW, Room LM Washington,
D.C. 20548

To order by Phone: Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202)
512-6061

To Report Fraud, Waste, and Abuse in Federal Programs

Contact:

Web site: www.gao.gov/fraudnet/fraudnet.htm E-mail: [email protected]
Automated answering system: (800) 424-5454 or (202) 512-7470

Congressional Relations

Gloria Jarmon, Managing Director, [email protected] (202) 512-4400 U.S.
Government Accountability Office, 441 G Street NW, Room 7125 Washington,
D.C. 20548

Public Affairs

Paul Anderson, Managing Director, [email protected] (202) 512-4800 U.S.
Government Accountability Office, 441 G Street NW, Room 7149 Washington,
D.C. 20548

www.gao.gov/cgi-bin/getrpt? GAO-06-815 .

To view the full product, including the scope
and methodology, click on the link above.

For more information, contact Marnie S. Shaul at (202) 512-7215 or
[email protected].

Highlights of GAO-06-815 , a report to congressional requesters

July 2006

NO CHILD LEFT BEHIND ACT

Assistance from Education Could Help States Better Measure Progress of
Students with Limited English Proficiency

The No Child Left Behind Act of 2001 (NCLBA) focused attention on the
academic achievement of more than 5 million students with limited English
proficiency. Obtaining valid test results for these students is
challenging, given their language barriers. This report describes (1) the
extent to which these students are meeting annual academic progress goals,
(2) what states have done to ensure the validity of their academic
assessments, (3) what states are doing to ensure the validity of their
English language proficiency assessments, and (4) how the U.S. Department
of Education (Education) is supporting states' efforts to meet NCLBA's
assessment requirements for these students. To collect this information,
we convened a group of experts and studied five states (California,
Nebraska, New York, North Carolina, and Texas). We also conducted a state
survey and reviewed state and Education documents.

What GAO Recommends

GAO recommends that the Secretary of Education (1) support research on
accommodations, (2) identify and provide technical support states need to
ensure the validity of academic assessments, (3) publish additional
guidance on requirements for assessing English language proficiency, and
(4) explore ways to provide additional flexibility for measuring annual
progress for these students. Education generally agreed with our
recommendations.

To view the Spanish translation of this highlights page, please see
http://www.gao.gov/highlights/ d061111high.pdf .

In the 2003-2004 school year, state data showed that the percentage of
students with limited English proficiency scoring proficient on a state's
language arts and mathematics tests was lower than the state's annual
progress goals in nearly two-thirds of the 48 states for which we obtained
data. Further, our review of data 49 states submitted to Education showed
that in most states, these students generally did not perform as well as
other student groups on state mathematics tests. Factors other than
student knowledge, such as how a state establishes its annual progress
goals, can influence whether states meet their goals.

For their academic assessments, officials in our five study states
reported taking steps to follow generally accepted test development
procedures and to ensure the validity and reliability of these tests for
students with limited English proficiency, such as reviewing test
questions for bias. However, our group of experts expressed concerns about
whether all states are assessing these students in a valid manner, noting
that some states lack the resources and technical expertise to take
appropriate steps to ensure the validity of tests for these students.
Further, Education's peer reviews of assessments in 38 states found that
25 states did not provide adequate evidence to ensure the validity or
reliability of academic test results for these students. To improve the
validity of these test results, most states offer accommodations, such as
a bilingual dictionary. However, our experts reported that research is
lacking on what accommodations are effective in mitigating language
barriers. A minority of states used native language or alternate
assessments for students with limited English proficiency, but these tests
are costly to develop and are not appropriate for all students.

Many states are implementing new English language proficiency assessments
in 2006 to meet NCLBA requirements; as a result, complete information on
their validity and reliability is not yet available. In 2006, 22 states
used tests developed by one of four state consortia. Consortia and state
officials reported taking steps to ensure the validity of these tests,
such as conducting field tests. A 2005 Education-funded technical review
of available documentation for 17 English language proficiency tests found
insufficient documentation of the validity of these assessments' results.

Education has offered a variety of technical assistance to help states
assess students with limited English proficiency, such as peer reviews of
states' academic assessments. However, Education has issued little written
guidance to states on developing English language proficiency tests.
Officials in one-third of the 33 states we visited or directly contacted
told us they wanted more guidance about how to develop tests that meet
NCLBA requirements. Education has offered states some flexibility in how
they assess students with limited English proficiency, but officials in
our study states told us that additional flexibility is needed to ensure
that progress measures appropriately track the academic progress of these
students.
*** End of document. ***