State and District Assessments as an Avenue to Equity and Excellence for English Language Learners with Disabilities

LEP Projects Report 2

Published by the National Center on Educational Outcomes

Prepared by Martha Thurlow and Kristin Liu

September 2001

This report is based on a paper presented at the Harvard Civil Rights Conference, Harvard University, Cambridge, Massachusetts, November 6, 2000.

Any or all portions of this document may be reproduced and distributed without prior permission, provided the source is cited as:

Thurlow, M., & Liu, K. (2001). State and District Assessments as an avenue to equity and excellence for English language learners with disabilities (LEP Projects Report 2). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes. Retrieved [today's date], from the World Wide Web: http://education.umn.edu/NCEO/OnlinePubs/LEP2.html

Executive Summary

Standards-based education has swept the country during the past decade, pushing and seeping its way into state after state across the nation. Based on the noble tenets of high standards, higher expectations, and improved instruction, politicians have taken on the battle cry for more assessments – both state and district assessments – to measure what students know and are able to do. At the same time, there are significant numbers of students for whom the educational system does not seem to be working as it should.

As in the past, the students who are most likely to fail to thrive in current educational environments are those of color, those who are poor, and those who are English language learners (ELLs), also referred to as limited English proficient (LEP) students. Why then, does the push for standards-based reform continue – even among those who are advocating for children most at risk, including those with disabilities and those who have limited English proficiency (Citizens’ Commission on Civil Rights, 1998)? And, why would we suggest that state and district assessments that hold schools accountable for student learning can serve as an avenue to equity and excellence for LEP students with disabilities?

It is impossible to begin a discussion of state and district assessments without raising concerns about assessments that have significant consequences for students – graduation exams and exams that determine whether students are allowed to move from one grade to the next. These high stakes assessments for students are particularly controversial whenever low student performance is directly related to poorer quality educational opportunities. For this reason, it is important for system accountability assessments – those that hold the educational system accountable and assign consequences to schools, administrators, or educators – to precede student high stake assessments. In those states and districts where high stakes assessments for students already exist, policymakers and educators must be held responsible for ensuring that the assessment system is appropriate – comprised of multiple measures (i.e., more than one kind of assessment), with accommodations policies that provide a wide range of accommodations, and appeals procedures for students who need alternative ways to demonstrate their knowledge and skills.

We want to focus our discussion on state and district assessments designed to hold the educational system accountable for the performance of students, and to demonstrate how these assessments are an important avenue to equity and excellence for English language learners with disabilities. To support our position, we first lay out some of the original ideas behind standards-based education. We indicate how current federal legislation for Title I of the Elementary and Secondary Education Act (reauthorized in 1994 as the Improving America’s Schools Act) and for special education services (Individuals with Disabilities Education Act - IDEA 97) reinforce the implementation of standards-based education for all students. After describing the characteristics of current standards-based assessment (including the important distinction between high stakes for students and high stakes for the schools), we identify some of the commonly projected benefits of standards-based assessments, particularly those in which the system is held accountable rather than the student.

Next we examine what we know about English language learners with disabilities, both in terms of their numbers across states, and in terms of their performance. We describe the information we can glean from our national report card – the National Assessment of Educational Progress. We look at information in state-level reports and on current state and district Web sites.

Finally, we look at data from a special project in which we examined, in detail, data from students with disabilities and English language learners. Based on the data that now exist and what could exist, we make several recommendations about ways to move forward to ensure that English language learners with disabilities actually reap the benefits that can be obtained from state and district level assessments.

Principles of Standards-Based Education

In the late 1980s, all 50 governors and the president convened to set the pathway for standards-based education. At the educational summit of 1989, all present agreed on the importance of a strong national education agenda focused on goals that would improve the global competitiveness of tomorrow’s workers. Chief among the goals was one that pushed for high rigorous standards, both content standards – to define what students should know and be able to do – and performance standards – to define how well students had to perform. Following quickly on the heels of standards was the recognition that it is necessary to measure progress toward meeting standards; data were needed to assist the system in recognizing whether students were meeting, or a least making progress toward meeting, the standards that had been defined for them.

Part of the rationale behind standards-based education is the belief that one way to drive better opportunities to learn is to ensure that the public knows how students are performing in relation to standards. For decades, states and districts relied primarily on norm-referenced tests to document student performance and growth. These tests, however, are designed to spread the scores of students and to allow for normative comparisons, not to judge whether students have met specific standards. Evidence that the improvements that had been made under the old system were not sufficient have come from several international studies in which students in the United States performed at levels comparable to many third world countries, and way below the levels of those countries with which the U.S. wanted to be economically competitive (e.g., Japan, Korea). Evidence of insufficient levels of performance also came from the business community, with anecdotal evidence that the graduates of high schools did not have the basic skills needed for entry jobs in most companies, as well as from higher education, where the need for remedial courses for incoming freshman had sky-rocketed.

In the early days of the educational reform movement, there was much discussion of authentic assessment as a way to obtain valid information about students’ knowledge and skills. Authentic assessment referred to a broad array of measurement approaches, including performance assessments and portfolio assessments. Over time, these approaches have faded considerably, and have been replaced by essay questions, certainly more performance-like than multiple-choice tests, but nowhere near the original conception of authentic assessment. This diversion, hopefully temporary, creates significant challenges for students from diverse backgrounds, particularly if they are English language learners and from diverse cultural backgrounds. Despite these challenges, we still believe that there are significant benefits to be gained from participation in state and district standards-based assessments.

Federal Education Laws

Federal education laws now support the argument that English language learners and students with disabilities will benefit from a standards-based educational system that uses large-scale assessments as accountability tools. Both Title I of the Improving America’s Schools Act (the reauthorization of the Elementary and Secondary Education Act) and the 1997 amendments to the Individuals with Disabilities Education Act (IDEA) require the participation of all students in state and district assessments. Title I, which clearly defines “all students” as including students with disabilities and students with limited English proficiency, requires (1) the participation of all students in the grades being assessed (which, for Title I purposes must include at a minimum, assessments of reading and mathematics at the elementary, middle, and secondary school levels); (2) the provision of reasonable adaptations and accommodations for students with diverse learning needs; (3) the assessment of limited English proficient students in the language and form most likely to yield accurate and reliable information on what they know and can do in areas other than English; and (4) the disaggregation of results within each State, local educational agency, and school by (a) gender, (b) each major racial and ethnic group, (c) English proficiency status, (d) migrant status, (e) students with disabilities compared to nondisabled students, and (f) economically disadvantaged students compared to students who are not economically disadvantaged.

Title I does not permit states to exempt any student subgroup from their assessment systems, and states must implement an auditing and record-keeping system to document which students are not assessed. States are required to explain how they will reduce the number of exemptions, and to examine whether intended effects are achieved by policies designed to increase student participation rates. The intent of all these provisions of the law are to spur educational reform for all students, not just a select few, as is clear in the following:

The intent of these requirements is to: 1) ensure that all students are held to the same high standards and appropriately assessed against those standards; and 2) ensure that all students are part of the indicators used to hold schools accountable. (U.S. Department of Education, 1996, p. 60)

IDEA has similar requirements. Students with disabilities are to participate in state and district assessments, with appropriate accommodations as necessary. Further, states and districts are to develop and implement alternate assessments for those students with disabilities unable to participate in state and district assessments. The number of students with disabilities in the general assessment and the number in the alternate assessment are to be reported, along with information on the performance of these students in each assessment, with the same frequency and in the same detail as for other students. These requirements are reinforced in the requirements for Individualized Education Programs (IEPs), and in general performance goals and indicators submitted bi-annually to the U.S. Secretary of Education. Together, these federal laws represent a significant commitment to holding schools accountable for the performance of all students.

Benefits of Standards-Based Assessments

Besides the dramatic evidence for the need for reform, and now the requirements of education laws, many benefits of having all students in the school accountability assessment system have been identified. One benefit of an all-inclusive assessment system is that it gives us a more accurate picture of the status of the educational system. When any group of students is systematically excluded from the measurement system, we have a biased picture of education, particularly if the group that is excluded tends to be lower performing students. This issue has been highlighted in the academic literature (McGrew, Thurlow & Spiegel, 1993), in journals for school boards (Zlatos, 1994), and in the popular press (Why Johnny stayed home, 1997). First, attention was given to the exclusion of students with disabilities, then to the exclusion of English language learners (Rivera, Stansfield, Scialdone & Sharkey, 2000). Only now are we beginning to think about students who are English language learners with disabilities.

There are other benefits of including all students in standards-based assessments. Among them are that participation in the measurement system is a critical piece of benefiting from reforms that are implemented. If groups of students with specific kinds of needs are excluded when assessments are given and results reported, the unique needs reflected in their performance will not be evident when reformers look at assessment results. A concrete example of this occurred in Kentucky during the beginning years of its reform. Kentucky started with principles that pushed forward the inclusion of all students in assessments. When the first set of results came back, they found that students with disabilities had basically zeroed out on the Science test. Students did not even know what a microscope was. With only a little exploring, they found that their students with disabilities had been systematically taken out of science to go to resource rooms! Science opportunity to learn changed dramatically for these youngsters as a result – they were put back into science classes and taught science!

Directly linked to the benefit of being a part of standards-based reforms, and having reforms designed for the students’ needs, is the avoidance of unintended consequences of exclusion from school accountability measures. Researchers have demonstrated, for example, that if a group of students is excluded from an assessment system (such as students with disabilities or English language learners), there is a likely increase in placements in those groups so that more and more low-performing students will not count. Allington and McGill-Franzen dramatically demonstrated this in New York, where a third grade test was used to determine rewards and sanctions for schools (Allington & McGill-Franzen, 1992). Evidence was clear that before the third grade test, there was a dramatic increase in the rate of referral to special education. Also, there was increased retention of students in second grade, probably based on the unfounded belief that giving low performing students one more year in second grade would ensure that their performance would be better when they did make it into the third grade test.

Perhaps even more important than these reasons is the finding that the inclusion of students in standards-based assessments increases expectations for these students; it forces a recognition that all children are expected to learn, which often gets lost when dealing with the challenges of disabilities and non-English background. Intertwined with the higher expectations for students is the recognition that educators working with these students really do have an important role in the education system. Their role is being elevated through the discussion of standards and assessments for all.

The Nature of Standards-Based Assessments: One of the Challenges

Two factors have complicated the notion of standards-based assessments. First, there has been a backslide from the initial educational reform notion of standards-based authentic assessments. Second, there has been increasing pressure to have high stakes for students (e.g., graduation exams, promotion exams), rather than (or in addition to) high stakes for schools.

Authentic assessments have been seen as one way to equalize the assessment situation for all students. In their purest form, authentic assessments maximize the performance of students, in part, by reducing the language load of the assessment process. Yet, authentic assessments are quite difficult to implement, and even more difficult to score in a way that is both reliable and valid.

Similar complications have arisen with respect to standards-based criterion-referenced assessments, which are designed to assess performance relative to standards rather than other students. Criterion-referenced assessments enabled states and districts to better align their assessments with their standards. Yet, over time, there has remained a political interest in being able to compare performance in a state, district, or school to a national norm; thus, states have either retained, gone back to, or added back in a norm-referenced assessment.

With all this shifting, states and districts find themselves in a situation where their assessments are constraining their ability to include all students. Part of this is due to the standardized nature of the assessments being used. When most of these tests were developed, particularly the norm-referenced tests, few English language learners or students with disabilities were included in the assessment development process. Without their presence during development, there was little need for accommodations, thus few accommodations are allowed by the test developers. Several other factors that also impinge on the inclusiveness of assessments were highlighted by the General Accounting Office, which called for states to expand their data collection as well as to “improve the completeness and quality of existing data” (General Accounting Office, 2000).

At the same time that standards-based reform was being pushed, there grew within states a concern about whether students were motivated to do the best that they could do. This concern seemed to have mingled with concerns about students not knowing enough when they graduated from high school. As a result, many states have upped the stakes for students; nearly half require that students pass an exam before they can receive a standard high school diploma (Guy, Shin, Lee & Thurlow, 1999; Heubert & Hauser, 1999), and several states (and many districts) are about to implement exams that determine whether a student is ready to move from one grade to the next (Quenemoen, Lehr, Thurlow, Thompson & Bolt, 2000).

Imposing high stakes for students at the same time that the educational system is still grappling with how to best provide standards-based instruction to all students makes for a very muddy system. It also makes for a system that is not exactly what we think it should be or the way we would like it to be.

Heubert and Hauser (1999) have made an excellent case for holding the education system accountable for student performance before imposing high stakes for students. We agree totally with that viewpoint. Still, we believe that opting out of existing assessment systems wholesale is not a good choice, and that doing so will actually diminish the opportunity for equity and excellence in education for many groups of students. The maxim that we treasure what and who we measure has been verified time and again. It is clearly the case that if you have no data on how students are doing, it is easy to forget their needs, even if the assessment has high stakes for students. This is particularly the case when so much attention is being given to the performance of students on state and district assessments. The critical element in any assessment system is to use data to make good educational choices, to provide a full range of supports to students when they are not doing well so that they are able to show improvements and not be forced to simply give up. Among the critical supports are both curriculum supports and assessment supports (e.g., multiple measures, accommodations, appeals procedures).

Data on English Language Learners with Disabilities

We attempted to cull information on what we know about the participation and performance of English language learners with disabilities from the vast array of national, state, and district level data collection programs. We looked for data that would give us a sense of how many students there are as well as how they are performing. We examined data from the nation’s report card, the National Assessment of Educational Progress (NAEP), data from state reports and Web sites, and data from selected districts. To present a full picture, we looked at both students with disabilities on IEPs, students with limited English proficiency (LEP), and LEP/IEP students. We looked for these with the realization that both terminology and definitional issues would confuse any attempt to make comparisons. Thus, while we are using the labels IEP, LEP, and LEP/IEP here, we know that sometimes a “student with disabilities” label includes both students on 504 plans and students who have IEPs. Similarly, students we refer to as LEP may in different places include students with a non-English language background who have varying degrees of English proficiency, or they may be only those students receiving English as a Second Language (ESL) or Bilingual services.

NAEP Data

NAEP became interested in the extent to which students with disabilities and English language learners participate in its assessments as a result of meetings held in 1994 (August & McArthur, 1996; Ysseldyke, Thurlow, McGrew & Vanderwood, 1994). Following these meetings, NAEP began a series of studies on its exclusion/inclusion criteria and allowing accommodations (Anderson, Jenkins & Miller, 1996; Mazzeo, Carlson, Voekl & Lutkus, 2000; McLaughlin, Vergun, Godlewski & Allen, 1996). A consistent finding of all the NAEP research to date has been that allowing students to use accommodations is a primary way in which to increase their participation rates. This finding, combined with similar findings from other data sources, confirm for educators the importance of accommodations, for students with disabilities, for students with limited English proficiency, and most likely for English language learners with disabilities.

The 1996 NAEP reports were the first to recognize English language learners with disabilities within their sampling plans (O’Sullivan, Reese & Mazzeo, 1997). The numbers of “IEP/LEP” students showing up in NAEP samples were small in most states, with the national average at less than 1% and the states with the highest percentages showing about 2% of students in the sample as being IEP/LEP students. These percentages seem unusually small, given the percentages of IEP and LEP students. For example, according to NAEP, Texas has 11% of its sample in the IEP group and 6% in the LEP group. This is similar to the Office of Special Education Programs estimate of 10.73% IEP students, but in contrast to the National Clearinghouse on Bilingual Education estimate of 12.7% LEP students. Even so, given these kinds of percentages, one might expect to find a greater percentage of students considered to be IEP/LEP students. Those IEP/LEP students who did show up in the NAEP sample were excluded from participating in NAEP, so we know nothing about whether these students could have been included.

State Data

States regularly report on the performance of students in their state assessments. In 1998 we examined state reports for information on the participation and performance of LEP students with disabilities. Table 1 summarizes our findings. While 16 states provided data on the participation or performance of students with disabilities, and 6 provided data on the participation or performance of students with limited English proficiency, only 1 state presented data on IEP/LEP students. However, the state presented only information on the number of such students taking the state test, not on their performance. The numbers were quite small (e.g., 37 IEP/LEP, compared to 8,300 IEP and 1,986 LEP testing in reading; 38 IEP/LEP, compared to 8,260 IEP and 1,994 LEP in math).

Table 1.  Data on IEP, LEP, and IEP/LEP Students Provided in State Reports

Note: The reports from which this information was gathered were collected and examined in 1998; however, the actual date of data included in the reports were from testing that occurred in 1997 or earlier.

Another way to look at state data is to go to state Web sites. We did this by going to the states that were the top five in percentage of the student population with limited English proficiency (Alaska – 26.9%, New Mexico – 23.9%, California – 22.2%, Texas – 12.7%, and Florida – 12.2%). This list was based on information provided on the National Clearinghouse on Bilingual Education Web site (http://www.ncbe.gwu.edu), which was based on 1996-97 data. Table 2 summarizes information on the data available on these states’ Web sites. It is evident from this table that even those states with large populations of LEP students do not necessarily have data on either students with disabilities or LEP students, much less LEP students with disabilities.

Table 2.  Data Availability on Selected State Web Sites

Note:  Data in this table are from 98-99, unless only more recent data were available.
^a    Alaska did have enrollment data for special education, broken down by ethnicity.
^b    California did have enrollment data for special education, broken down by ethnicity; test performance scores were provided by race; performance scores were also broken out by three groups:  All, Not LEP, and LEP.
^c    Florida did have enrollment data for special education, broken down by ethnicity.
^d    Texas did have enrollment data for special education, broken down by ethnicity; also the number of students exiting special education by race; performance scores were broken out by special education status and race/ethnicity, by bilingual education, English as a Second Language program, and migrant status.

Whether paging through reports or surfing the Web, one thing becomes very clear. It is nearly impossible to find data on LEP students with disabilities. Further, it sometimes seems, particularly on Web sites, that states are purposely making it very difficult to find information, even basic information such as the number of students with disabilities who are also of limited English proficiency. Data from Texas, the state with perhaps the most data on LEP students, provide first-hand evidence of the lack of data. In a summary table on the Web site for the Texas Assessment of Academic Skills (TAAS), data on the percentage of students meeting the minimum expectation are disaggregated for limited English proficient students. The table indicates that 50% of LEP students who took the science exam met the science standard, while 89% of students who were not LEP took the exam and met minimum standards. What it does not tell us is how many other LEP students there are who did not take the exam; it is questionable whether it is possible to rely just on the number of students who are exempted from testing to determine that number because states do not usually reveal how many students actually could have taken the test – enrollment data rather than “eligible” students data.

District Data

We also looked at data from selected districts, again, those with large populations of LEP students. According to the National Clearinghouse on Bilingual Education Web site, the districts with the largest populations (in number) were Los Angeles, New York, Chicago, Dade County, and Houston; the five districts with the largest percentage of LEP students were all in California and included (in addition to Los Angeles, with 45.6% LEP population), Santa Ana (69.3%), Glendale (51.9%), Pomona (44.8%), and Garden Grove (42.9%). (The Clearinghouse information on districts was based on the 1993-94 school year.) Data from all of these districts are shown in Table 3. District data also give us some sense of the disarray of some systems when it comes to providing data. On the other hand, sometimes these data are very complete, and once in a while, they are the most accurate data that exist.

Table 3.  Data Availability on Selected District Web Sites

Note:  Data in this table are from 98-99, unless only more recent data were available.
^a    Garden Grove did have enrollment data for LEP students, and broken down by ethnicity.
^b    Houston did have enrollment data for ESL and bilingual programs, as well as number and percent LEP exempt. * Indicates that the number or percentages were only for those students using the Spanish TAAS.
c    Los Angeles did have data on the number of students taking the test, but no data on enrollment.
^d    New York did have data for the district on the web site; however, data for individual schools often had blanks for LEP students.
^e    Pomona had performance scores for students designated as FEP (Fully English Proficient) and for LEP + English only students.

State Assessment Data and What We Can Learn From It

We have, so far, documented how little data we have specific to large-scale assessment and participation and performance for LEP students with disabilities. What kind of data could we have and what would they tell us? Some answers come from the Minnesota Assessment Project (MAP), a four-year federally funded grant awarded by the Office of Educational Research and Improvement to the Minnesota Department of Children, Families and Learning and the National Center on Educational Outcomes. MAP focused on promoting greater inclusion of LEP students and students with disabilities in Minnesota’s standards-based teaching and assessments.

The public perception of Minnesota typically is that there is not a great deal of cultural diversity in this midwestern state. One would assume that relevant data do not exist on LEP students in Minnesota. This is not so. Minnesota has a significant population of LEP students with some unique characteristics that make lessons learned in this state particularly timely and pertinent to educators and policymakers.

The LEP student population in Minnesota is growing rapidly. Minnesota religious organizations and social service agencies have traditionally played a strong role in refugee resettlement to the United States and this accounts for an atypical demographic profile of Minnesota’s LEP students. As political situations around the world create new groups of refugees, numbers of LEP students in Minnesota can rise drastically in short periods of time. The Minnesota Assessment Project found that in 1997 there was such an unusually large increase in the number of LEP students entering Minnesota schools between the time of fall enrollment counts and the spring testing date that state graduation test participation rates calculated for this group showed more than 100% taking the test that year.

Specific school districts within the state also show large gains in the number of LEP students. In the past six years, the LEP population in one of the largest urban districts in the state increased by 183%—from 6,000 students to about 17,000 students—at the start of the 2000 school year. In terms of overall enrollment, the student body of this district is currently 38% LEP and some individual schools within the district sometimes have an enrollment of over 50% LEP students. If the National Clearinghouse on Bilingual Education listing of the school districts with the largest number and percent of ELLs were redone for the 2000-2001 school year, this district might very well appear on the list of the top 20 districts nationwide.

The LEP population in Minnesota is different in make up from that of most other states. Nationwide, Spanish speakers represent the majority of LEP students. In Minnesota, Southeast Asian students from Hmong, Lao, Vietnamese, and Cambodian language backgrounds make up the majority of LEP students. Minnesota also has the largest Somali student group in the country. These students are primarily refugees who may have limited formal schooling in their native language, spotty educational backgrounds with schooling experiences interrupted by war, a lack of literacy in their native language, and high mobility rates in the United States as they settle in one place and later move to be reunited with family members in other places. Added to these characteristics are significant health issues including post-traumatic stress disorder that may make these students more likely to be referred for special education services.

While other states may have fewer numbers of refugee students with limited schooling and literacy, all states are struggling with ways to include them in educational reform movements and to educate them to high standards. Lessons learned in Minnesota can be useful to all states. Minnesota was one of the earliest states to implement large-scale standards-based testing in which there was a concerted effort to include all students, and has longitudinal data dating back to 1996. By looking at what we know about the participation and performance of LEP students and of students with disabilities in Minnesota’s large-scale tests, we can make some inferences about the participation and performance of LEP students with disabilities.

Participation

First, as Table 4 illustrates (Liu & Thurlow, 2000; Thompson et al., 2000, Thompson et al., 1999; Walz, Thompson, Thurlow & Spicuzza, 2000), participation of LEP students in Minnesota’s statewide graduation test at grade 8 and accountability tests at grades 3 and 5 is high. Over the past four out of five years of testing it has ranged from 88% to 100% for the graduation tests of reading and math, and from 89% to 93% for state accountability tests at grades 3 and 5 (Liu & Thurlow, 2000).

Table 4. Participation of Minnesota’s LEP Students and Students with Disabilities in State Assessments 1997-1999

Note:  1996 data are not included here because test participation was not mandatory in the first year.

Participation of students with disabilities in those same tests was lower initially, most likely because the IEPs of some students allowed them at that time to be exempted from testing. For the graduation tests, participation has ranged from 58% to 89% in reading and from 59% to 89% in Math. For the state accountability tests, participation for both reading and math at grades 3 and 5 is consistently 84% to 85%. For both groups, variations in the percent participating from year to year may be partially due to changes in participation requirements for districts, and as previously mentioned, in numbers of incoming LEP students. For example, in 1998 all districts were mandated to give the state developed graduation test for the first time; in earlier years districts had some flexibility in whether the state graduation test or a commercial test was given, thus higher participation rates for students with disabilities can be seen in 1998, compared to 1997.

The Minnesota Assessment Project found that participation rates for students with disabilities varied according to the primary disability classification of the student (Thompson, Thurlow & Spicuzza, 2000; Thompson, Thurlow, Spicuzza, & Parson, 1999). Of all the students in the various disability categories who were tested in 1999, the following categories of students had the lowest participation rates: mild/moderate mental impairment (73% for both reading and math), autism (68% in reading, 66% in math), and moderate/severe mental impairment (<10% for both reading and math). These participation rates are not unexpected given the types of primary disabilities these students had. Based on these findings, LEP students with these four types of disabilities can be expected to have some of the lowest participation rates.

One factor contributing to the possibility of lower participation rates for ELL students with disabilities is a lack of involvement in test decision making by staff who are knowledgeable about second language acquisition issues. Mazzeo and his colleagues documented that NAEP forms documenting students with both limited English and a disability were consistently completed by the special education teacher (O’Sullivan et al., 1997). There was no involvement from the English as a Second Language or Bilingual Education teacher.

Minnesota Assessment Project findings support this lack of ESL/Bilingual teacher involvement in test participation decision making for students they serve and document the fact that the lack of involvement is the most prevalent in large urban districts with high numbers of LEP students. In Minnesota, test participation decisions for ELLs are often made just prior to the test, at a point when it may be too late to order special test forms for accommodated tests. Decisions are often made by a group that does not usually include the individual student or the students’ parents, and often does not include the ESL/Bilingual education teacher who is most familiar with the students’ process of second language acquisition. While test participation decisions for students with IEPs may be made earlier, there are no data to show whether ESL/Bilingual staff are involved in these IEP meetings. Clearly, this finding has implications for the inclusion of LEP students with disabilities in assessments.

Performance

As shown in Table 5 (Liu & Thurlow, 1999, 2000; Thompson et al., 2000; Thompson et al., 1999), the graduation test performance of LEP students in Minnesota’s large-scale assessments is low (Liu & Thurlow, 1999).

Table 5.  Rates of LEP Students and Students with Disabilities Passing Minnesota’s Graduation Test   and Meeting Accountability Test Proficiency Standards 1997-1999

*Note:  1996 data are not included here because test participation was not mandatory in the first year.

Fewer than 25% of LEP students pass either the reading or math test on the first sitting, compared to 59% to 75% of native speakers. Yet these data tell us more than just the expected poorer performance of LEP students who are in the process of learning the academic English that is required on the tests. Besides the finding that initial reading test scores are lower than math scores, continued attention to the data revealed that students made greater gains in the percentage of items correct on reading tests than math tests as they retake them in successive years. When students of all disability categories are grouped together, passing rates on the graduation test are less than 40%, compared to 60% to 75% of their peers who do not have a disability.

These kinds of data give us important information. For example, we see that there are students with limited English proficiency who do pass a reading test written at grade level. The same percentage, or slightly higher, pass a math test written at grade level. There is a growing percentage of LEP students who pass the graduation reading test. Similarly, data from accountability assessments show that students with limited English proficiency do meet reading and math proficiency standards.

We also can look at what happens when those students who did not pass the graduation test are retested. Table 6 shows the percentage of all students (including LEP and students with disabilities) and the percentage of LEP students at different score levels who took the graduation test in both 1997 and 1998 (Liu & Thurlow, 2000; Spicuzza, Liu, Swierzbin, Bielinski & Thurlow, 2000). All of these students failed the test on their first attempt. The table shows the percentages of these students who passed the test on the second attempt, as a function of their level of performance when they were tested the first time.

Table 6.  Percentage of Students Who Did Not Pass the 1997 Graduation Test, but Passed in 1998, by 1997 Performance Level

a   Score groups for reading include students within the following range of percent of items correct:  low=0% to 25% correct, lower middle=26% to 50% correct, upper middle = 56-68% correct, top = 69% to 74% correct.  Students with 75% correct or higher passed the test.

b   Score groups for math include students within the following range of percent of items correct:  low=0% to 25%, lower middle = 26% to 50%, upper middle = 51% to 69%, top = 70% to 74%.  Students with 75% correct or higher passed the test.

It is interesting to note that on the reading test, LEP students in each score group were less likely than all students in the same score groups to pass the test on the second attempt (Spicuzza, et al., 2000). This pattern did not hold true for students taking the math test a second time. On the math test, all students and LEP students had a similar likelihood of passing the test the second time. These data show the need to give extra attention to reading instruction for those LEP students who do not pass. Since the math test was made up of word problems, improved reading skills would also benefit LEP students who had not yet passed the math test.

What we discover from looking at state assessment data of students with disabilities is also informative (Thompson et al., 2000; Thompson et al., 1999). On the 1999 Minnesota graduation tests, the categories of students with disabilities that had some of the lowest performance were students with learning disabilities (29% passed in reading, 24% in math), other health impairments (39% passed in reading, 32% in math) and emotional disabilities (41% passed in reading, 31% in math). This finding is significant because these students could potentially achieve at higher levels, suggesting the need for greater use of appropriate test accommodations. The finding is important because LEP students, particularly those experiencing post-traumatic stress disorder, are most likely to receive special education services under these categories, suggesting that when second language acquisition issues are added in, LEP students with learning disabilities, other health impairments, and emotional disabilities will have the lowest performance. Anecdotal information from some English as a Second Language teachers in Minnesota suggests that there may be some LEP students with evidence of these types of disabilities who are not referred to special education because of concerns about not being able to differentiate second language acquisition and disabilities.

Another way that we can look at LEP data is by language group. When this is done, it becomes clear that not all LEP students perform in the same way on the same tests (Liu, Thurlow, Thompson & Albus, 1999). Table 7 shows, for example, that the performance of students from African Language groups is quite a bit below other language groups. It is important to continue to track these students (perhaps defined by specific language), who may be relatively new immigrants to the state, to determine whether this pattern of performance changes over time. Also evident in the data by language group is that the discrepancy between reading and math performance is much greater in some groups, particularly Vietnamese students. In contrast, students with the Russian language perform nearly equally in reading and math. Systematic study of why these differences might exist and their implications for instruction and other intervention programs is clearly warranted. Care must be taken along each analytical step to truly understand who and what is being measured. For example, in Minnesota, any attempt to look at LEP performance by urban areas compared to suburban or rural will be confounded by the fact that different language groups are concentrated in different areas (e.g., Hmong in urban areas, Hispanic in rural areas, etc.). The importance of carefully defining populations when looking at changes in performance over time has been revealed in recent research on longitudinal trends in the performance of students with disabilities, which revealed that contradictory trends emerge depending on how the population is defined (Thompson et al., 1999).

Table 7.  Number and Percentage of LEP Students from Various Language Groups Passing Read and Math Graduation Tests