NCEO Logo
Bookmark and Share

Technical Report 60

Characteristics of Low Performing Special Education and Non-special Education Students on Large-scale Assessments

Yi-Chen Wu • Kristin K. Liu • Martha L. Thurlow • Sheryl S. Lazarus • Jason Altman • Elizabeth Christian

March 2012

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

Wu, Y. C., Liu, K. K. , Thurlow, M. L., Lazarus, S. S., Altman, J., & Christian, E. (2012). Characteristics of low performing special education and non-special education students on large-scale assessments (Technical Report 60). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes


Table of Contents


Executive Summary

Federal legislation requires that all students participate in state assessments for accountability purposes. Most students with disabilities participate in the general assessment, with or without accommodations. Several states offer an additional assessment option for students with disabilities—the alternate assessment based on modified achievement standards (AA-MAS). In order to take the AA-MAS, students must have an Individualized Education Program (IEP) containing goals aligned with grade level content. Students must also be able to make significant progress, but not be expected to reach grade-level proficiency in the year covered by their IEP.

States that choose to offer an AA-MAS must also develop participation guidelines (subject to federal approval) that IEP teams can use to determine which special education students may participate in the assessment. Previous NCEO reports have demonstrated that the participation guidelines outlined by states with an AA-MAS assessment option differ. Many states target students with disabilities who score at the lowest levels on the regular state assessment as potential AA-MAS participants. Researchers who have examined the characteristics of low-performing students have concluded that low performing students are more often male, from racial or ethnic minority groups, receiving free or reduced price lunch, and receiving special education services. However, many students with low scores on standards-based assessments do not have disabilities.

The purpose of the current report is to investigate whether the characteristics of the lowest performing students in special education differ from the characteristics of the lowest performing students who are not in special education. The investigation in this report used data from low performing students in four states: Alabama, Hawaii, South Dakota, and Wisconsin. Achievement data were disaggregated by three demographic characteristics (gender, race/ethnicity, and income status) for students taking the reading or mathematics assessments in fifth or eighth grade. In addition, we tracked data for each student over three years to identify how students moved in and out of the low performing category (low performing was defined as the tenth percentile and below for this report) across time.

Results revealed that the demographic characteristics of students who performed at the lowest levels on the state assessment over several years were similar regardless of whether students received special education services. In both groups, the lowest performers were more likely to be students of racial or ethnic minority, and students from low-income backgrounds. However, students receiving special education services were more likely than their non-special education peers to score below the 10th percentile for several years in a row. Among students receiving special education, those from racial/ethnic minority backgrounds and low-income families were more likely than their peers to score at the lowest levels on state assessments for several consecutive years. These results suggest widespread issues with low achievement in minority and low income groups that states must address when assessing students for accountability purposes.


Overview

Federal laws require states to include all students, including students in special education, in state assessments used for accountability purposes. Most students with disabilities participate in the general assessment, with or without accommodations (Altman, Thurlow, & Vang, 2010). Students with significant cognitive disabilities may participate in alternate assessments based on alternate achievement standards (AA-AAS) (Quenemoen, 2008). Several states offer an additional assessment option—alternate assessments based on modified achievement standards (AA-MAS) (Albus, Lazarus, Thurlow, & Cormier, 2009; Cortiella, 2007; U.S. Department of Education, 2007).

According to federal regulations, students who participate in an AA-MAS must have an Individualized Education Program (IEP). They also must be able to make significant progress, but not be expected to reach grade-level proficiency within the year covered by their IEP. The regulations require that students who participate in an AA-MAS have access to grade level content and they may be from any disability category (U.S. Department of Education, 2007). The AA-MAS is an optional assessment and many states do not offer it.

Federal regulations require states that offer an AA-MAS to develop participation guidelines (subject to federal approval) that IEP teams can use to determine which special education students may participate in this option. Research has shown that state participation policies differ, and that most states include previous poor performance on state assessments or multiple measures in their participation guidelines for AA-MAS (Lazarus, Hodgson, Price, & Thurlow, 2011).

In an effort to identify who might be potential AA-MAS participants if they were selected from among the lowest performing students, we embarked on a detailed longitudinal analysis of four states' assessment data. The analysis used three years of demographic and performance data. The four states—Alabama, Hawaii, South Dakota, and Wisconsin—are all members of a state consortium that studies ways to identify students who are eligible for an AA-MAS. Some of the analyses shown in this report were included in a previous NCEO Brief (Lazarus, Wu, Altman, & Thurlow, 2010). However, that brief examined the characteristics of all low performing students. It did not disaggregate the data reported to show the characteristics of special education students who may be eligible to take an AA-MAS.

This report examines the characteristics of low performing special education students. This approach is consistent with federal regulations which indicate that students must have an IEP to be eligible to take the AA-MAS. To provide additional context, we compared the data on low performing special education students to those of low performing non-special education students. For this analysis, low performing (LP) students were defined as students who scored at the 10th percentile or below on the statewide assessment in any one of the three years of data we examined. Students who scored at the 10th percentile or below for all three years were identified as persistently low performing (PLP) students. Three research questions guide the investigations presented in this report:

  1. Are the demographic characteristics of persistently low performing special education students different from the characteristics of persistently low performing students who are not in special education?
  2. Are special education students who were low performing in the first year more or less likely to move out of the low performing group than non-special education students? Is there any variation in movement by content area (reading vs. math)?
  3. Are low performing special education students in some demographic subgroups more or less likely to move out of the low performing group than their low performing peers who are not in special education?

Related Research

Since the AA-MAS assessment option was made available to states in 2007, several studies have been conducted to identify the population of students who might qualify to take the AA-MAS. States have found this to be a difficult task because many students with low scores on standards-based assessments do not have identified disabilities (Lazarus & Quenemoen, 2011; Marion, Gong, & Simpson, 2006).

Some of the earliest research on the characteristics of low-performing students was conducted in Colorado. In an effort to learn more about the characteristics of students "in the gap," a study committee in Colorado conducted research focusing on students with IEPs who scored at the lowest levels on the general state assessment, but who were not eligible for the state's alternate assessment for students with significant cognitive disabilities (HB 05-1246 Study Committee, 2005). The study committee found that students "in the gap" were more often American Indian, African American, or Hispanic than "non-gap" students. Male students outnumbered female students "in the gap" by nearly two to one. The committee noted that there was a small percentage of students who scored at the lowest levels on the state assessment, and did not have an IEP. However, these students were mentioned only briefly in the report and their demographic characteristics were not examined.

The characteristics of low-performing students have been studied in other states as well. Hess, McDivitt, and Fincher (2008) examined research conducted in Georgia on the characteristics of students who might qualify for an AA-MAS. In examining the population of students within the state who were "persistently low performing" (defined as those students scoring at the lowest performance level on the regular state assessment for three consecutive years), it was found that these students were more often African American, male, receiving free and reduced lunch, and often had an IEP and/or mild intellectual disabilities.

Research on the characteristics of low performers in the New England Compact states included teacher judgments about this population. It was found that though the group of lowest performers included students with and without disabilities, more than half of the group had IEPs. Additionally, students who performed poorly on standardized tests but received higher teacher judgments were far less likely to have an IEP (Bechard & Godin, 2007). Lazarus and Quenemoen (2011) concluded that the results of these related studies (e.g., Colorado, Georgia, New England Compact states) indicate that the lowest performing students are often from historically underserved populations such as students of ethnic minority background, students of low socioeconomic status, or students in special education.

The Center on Education Policy has issued a series of reports in which they investigated achievement differences among subgroups on state tests. The multi-state results have shown that females tend to outperform males in reading and that the gap is not narrowing (Chudowsky & Chudowsky, 2010). Additional analyses focused on differences between the performance of students from different racial/ethnic backgrounds, students with low socio-economic status (Chudowsky, Chudowsky & Kober, 2009), and students in special education (Chudowsky & Chudowsky, 2009). Performance gains were seen in each of these subgroups at all achievement levels. Achievement gaps narrowed for Latino and African American students to a greater extent than for students from other racial/ethnic groups (Kober, Chudowsky, & Chudowsky, 2010). The authors note that test data for students in special education were imprecise, so interpretations of these data provided only a rough indicator of the achievement trends of these students. Notably, none of these analyses examined students with more than one of these characteristics (e.g., low-income students in special education).

A number of National Center on Educational Outcomes (NCEO) investigations of assessment data reported on state websites showed that students in special education score proficient at different rates across states and that achievement gaps between this population and students without disabilities also vary extensively between states (Albus, Thurlow, & Bremer, 2009; Thurlow, Bremer, & Albus, 2008; VanGetson & Thurlow, 2007). Also, an analysis of assessment performance of students in special education within states compared to the state targets for this subgroup showed that states were making gains but were challenged to keep up with the ambitious targets that they had set (Altman, Rogers, Bremer, & Thurlow, 2010).

These findings raise the question of whether the characteristics of the lowest performing students in special education differ from the characteristics of the lowest performing non-special education students. This question will be investigated in depth in this report. This analysis is also distinctive in its investigation of the movement of students in various demographic subgroups across performance levels.


Method

Procedures

Preparation of Data Sets

The data for the current study are based on students who took the regular statewide assessment in reading or mathematics and had valid scores for three consecutive years. In Alabama, Hawaii, and South Dakota, we selected students who were in grades 5 and 8 during the 2006-07 school year and then collected assessment data for these same students in the previous two years as well (2005-2006 and 2004-2005). In Wisconsin, we followed a similar procedure beginning with fifth and eighth graders in the 2007-2008 school year and following their scores backward for two years (2006-2007 and 2005-2006).

For each state, the older test scores were merged with the scores from the most recent year in order to examine student performance across time. Demographic information such as income-level, race/ethnicity, and gender were based on the characteristics in the most recent year of data.

Identifying Low-performing Students

For the purposes of this report, we identified fifth and eighth grade low performing students whose scale score on state reading or math assessments was at or below the tenth percentile. To make this determination, we examined the cumulative frequency distributions of scale scores for reading and math tests in each year of the available data. Any student who had a score less than or equal to the 10% cut point was classified as low performing for that particular assessment and year. For three years, we tracked the performance of all students who were identified as low performing in the first year of our data set. (For a table comparing the cut scores across states, see Appendix A, Table A1). Students who were not identified as low performing in the first year of three years of data collection were excluded from this analysis.

Defining Variables and Terms

All of the analyses in this report compare low performing non-special education students to low-performing special education students to determine whether there are demographic and performance differences in low performing students between these two groups. We perform the same analyses at grades five and eight and for reading and math assessments.

States' definitions of demographic variables vary considerably. In this report, we only selected the demographic variables available across most, or all, of the states. The research cited in the literature review section of this paper suggested that students who are low performing are more likely to be males, students of color, and low-income students (cf. HB 05-1246 Study Committee, 2005; Hess, McDivitt, & Fincher, 2008). For that reason we focused our analyses on these groups of students. For purposes of clarity we chose to use the terms "white" and "non-white" students for these analyses.

In other instances, we ran data using a variable from state databases as an indicator of a demographic characteristic but we chose not to use the data variable name in our discussions of tables and figures. For example, we used the term "Free/Reduced Lunch," representing students who are eligible for free or reduced price lunches at school, as indicator of students from low-income backgrounds. Our figures and tables present data for students who receive free or reduced lunch but our text explaining those figures refers to low-income students.

Identifying Movement Categories

Some of the analyses in this document examine whether students identified as low performing in year one of our data set moved out of the low performing group in subsequent years. To clearly communicate the results of these movement analyses we use the term "initially low performing" to mean students who were in the bottom 10th percentile the first year of our analyses, but not in the second, or third years. We use the term "persistently low performing" (PLP) to indicate students who were in the bottom 10th percentile all three years of our analyses.

Analyses

Descriptive analyses were performed to explore the characteristics of low performing students in special education and not in special education. We performed crosstabs to provide the information by selected demographic variables, and performance groups at grades five and eight and for reading and math assessments.

Participants

Table 1 shows the total number of non-special education and special education students in each state who had three years of test results for reading and math assessments. The grade levels listed in the table are the grades students were in during the final year of analysis.

Of the four states in this study, Wisconsin had the largest number of students with three years of data. For example, for fifth grade reading 45,467 non-special education students and 6,383 special education students are included in this analysis. South Dakota had the fewest students with three years of data. Of the 5th graders, 6,916 non-special education students had three years of reading assessment scores and 1,014 special education students had three years of reading scores. In each of the four states, the number of students with three years of data for the reading assessment is slightly different from in the math assessment.

Table 1. Number of Grade 5 and 8 Special Education and Non-Special Education Students with Three Years of Data on the State Reading and Math Assessments

   

Alabama

Hawaii

South Dakota

Wisconsin

Grade

Subject

Non-SpEd

SpEd

Non-SpEd

SpEd

Non-SpEd

SpEd

Non-SpEd

SpEd

5

Reading

35706

3578

10377

1182

6916

1014

45467

6383

Math

35666

3579

10372

1179

6967

1028

45577

6551

8

Reading

36410

3829

9954

1335

7793

893

48936

7140

Math

36335

3805

9950

1336

7794

893

48918

7155

Note: The numbers in this table are based on school years 2004-05 through 2006-07 for Alabama, Hawaii, and South Dakota. The data for Wisconsin is based on school years 2005-06 through 2007-08.


Results

Research Question 1: Are the demographic characteristics of persistently low performing special education students different from the characteristics of persistently low performing non-special education students?

As the discussion of Figures 1 through 12 will highlight, the demographic characteristics of persistently low performing students in these four states are more complex than previous research findings have suggested. Findings for gender are mixed and do not show a consistent pattern across tests, across the persistently low performing versus total population of students, or across special education and non-special education students. There is, however, a clear pattern of a greater percentage of non-white and low income students in the persistently low performing groups, regardless of whether those students are in special education or not. The tendency for the persistently low-performing group to have a higher percentage of non-white students is magnified in the non-special education group on both the reading and math assessments. Similarly, our findings suggest that the increased percentage of low income students is magnified in the persistently low-performing non-special education group compared to the special education group.

Gender

Reading. To answer our first research question, we began by examining the gender of persistently low performing students (i.e., below the 10th percentile for three consecutive years) in both non-special education and special education. Figure 1 shows the percentage of males in the persistently low performing non-special education population on the reading test at fifth and eighth grade compared to the percentage of males in the total non-special education population in those same grades (see Appendix A, Table A2 for more detail). Looking at the data in this way allows us to see whether the proportion of males in the persistently low performing non-special education group relates in some way to the proportion of males in the non-special education population. For example, in Alabama approximately 66% of the grade 5 persistently low performing non-special education students were male, compared with about 50% of the students in the total grade 5 non-special education population.

Figure 1. Percentage of Non-Special Education Grades 5 and 8 Males in the Persistently Low Performing Group on the State Reading Test Compared to the Percentage of Males Among all Non-Special Education Students

Figure 1 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

In Figure 1 we see that across states males represented approximately 50% of the total non-special education group taking the reading test (shown in the lighter bars). The percentage of males in the non-special education group (shown in the darker bars) was similar at fifth and eighth grade. The percentage of males in the persistently low performing non-special education group ranged from 54% in Wisconsin at grade 8 to 75% in Hawaii at grade 8 and did show small variations (10% or less) across grades (See Appendix A, Table A2 for more detail).

In all four states there was a larger percentage of males in the persistently low performing non-special education group compared to the total non-special education population taking the reading test. The size of these differences varied. In South Dakota for grade 5 and in Wisconsin at both grades, the size of the difference was small (10% or less). In South Dakota for grade 8, and in Alabama and Hawaii at both grades, the size of the difference was large (more than 10%).

Figure 2 presents similar data to Figure 1 for the special education students. This figure shows, for example, that in Alabama 72% of the grade 5 persistently low performing special education students (darker bar) was male, compared with about 65% of the students in the total grade 5 special education population (lighter bar).

Figure 2. Percentage of Special Education Grades 5 and 8 Males in the Persistently Low Performing Group on the State Reading Test Compared to the Percentage of Males Among all Special Education Students

Figure 2 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

Figure 2 indicates that males represented more than half of the overall special education population taking the reading test at fifth and eighth grades (lighter bars). The percentage of males in the special education population ranged from a low of 65% for South Dakota eighth graders to a high of 72% for Hawaii fifth graders (see Appendix A, Table A3 for detail). The percentage of males in the total special education population was fairly similar at grades five and eight for each state with only small (10% or less) variations evident. The percentage of males in the subgroup of persistently low performing special education students (darker bars) ranged from a low of 63% in South Dakota at both grades to a high of 78% in Hawaii at grade 8. This percentage was similar at grades 5 and 8 with only small differences (10% or less).

The percentage of males varied in the persistently low performing special education student group (darker bars) compared to the total special education group (lighter bars). In two states, Alabama and Hawaii, there was a slightly higher percentage of males in the group of persistently low performing special education students than there was in the total group of special education students. In the other two states, South Dakota and Wisconsin, the percentage of males in the persistently low performing special education students was similar to the percentage of males in the total special education population. The size of these differences was small (10% or less)


Math. Figures 3 and 4 examine mathematics data for the same states at the same grades. Figure 3 presents fifth and eighth grade non-special education data, while Figure 4 presents the special education data (see Appendix A, Tables A4 and A5 for more detail).

Figure 3 shows the percentage of males in the persistently low-performing non-special education population on the math test at fifth and eighth grade compared to the percentage of males in the total non-special education population in those same grades. For example, in Alabama approximately 50% of the students in the grade 5 non-special education group were male (shown by the lighter bar). In comparison, 54% of the grade 5 persistently low performing non-special education students (shown by the darker bar) were male.

Across the four states shown in Figure 3, males represented roughly half, 48% to 50%, of the fifth and eighth grade non-special education population taking the math test (lighter bars). In comparison, the percentage of males in the persistently low performing group of non-special education students (darker bars) ranged from a low of 41% in Wisconsin at both grades to a high of 62% in Hawaii at eighth grade. With the exception of Hawaii, the difference in the percentage of males across grades within a group was typically small (10% or less). Hawaii showed a large difference (more than 10%) from grade 5 to grade 8 in the percentage of males in the persistently low performing non-special education group.

There was no consistent pattern when we compared the percentage of males in the persistently low performing non-special education population (darker bars) to the percentage of males in the total non-special education population (lighter bars). Two states, Alabama and South Dakota, had a greater percentage of males in the persistently low performing non-special education group than in the entire non-special education population. In contrast, Wisconsin had a smaller percentage of males in the persistently low performing population than in the total non-special education population. In the remaining state, Hawaii, the pattern differed across grades. At fifth grade, the percentage of males was slightly smaller in the persistently low performing group in Hawaii than it was in the total non-special education group. In contrast, at eighth grade the percentage of males in the non-special education persistently low performing group was 13% greater than the percentage in the total non-special education population. Other than in Hawaii at eighth grade, most states in Figure 3 did not show a large difference (i.e. greater than 10%) in the percentage of males across grades in the same group.

Figure 4 provides the percentage of males in the persistently low performing special education group on the math test and in the total special education group. The data in this figure indicate that more than half (65% to 72%) of all special education students taking the math test were male (lighter bars). Similarly, more than half (55% to 69%) of persistently low performing special education students (darker bars) were male. Except for in Wisconsin at grade 5, there did not appear to be a sizeable difference (larger than 10%) in the percentage of males in either group across grade levels.

Figure 3. Percentage of Grade 5 and 8 Males in the Non-Special Education Persistently Low Performing Group on the State Math Test Compared to the Percentage of Males Among Non-Special Education Students

Figure 3 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

Figure 4. Percentage of Special Education Grades 5 and 8 Males in the Persistently Low Performing Group on the State Math Test Compared to the Percentage of Males Among all Special Education Students

Figure 4 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

On the math tests the percentage of males in the persistently low performing special education group (darker bars) was either similar to, or smaller than, the percentage in the total special education population (lighter bars) on the math test. Alabama had roughly the same percentage of males in the two groups at each grade. In Wisconsin at grade 5 there was a large (greater than 10%) difference in the percentage of males in the two groups, with a smaller percentage in the persistently low performing special education group (55%) compared to the total population (67%). In the remaining states the difference in the percentage of males across groups within one grade level was small (10% or less).

Race/Ethnicity

Reading. Another demographic variable that we examined was the race/ethnicity of persistently low performing students. The data in these figures are labeled "white" and "non-white students." Figures five and six highlight these data from the reading assessments at grades 5 and 8. Figure 5 presents the non-special education data and Figure 6 presents the special education data (see Appendix A, Tables A6 and A7 for more detail).

Figure 5. Percentage of Non-Special Education Grades 5 and 8 Non-White Students in the Persistently Low Performing Group on the State Reading Test Compared to the Percentage of Non-White Students Among all Non-Special Education Students

Figure 5 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

Figure 5 shows that there was a great deal of variability in the percentage of non-white students in both groups on the reading assessment. In the total non-special education group the percentage of students who were non-white ranged from a low of approximately 13% to 15% in South Dakota to a high of 87% to 88% in Hawaii (lighter bars). Differences across grades were minimal (10% or less). The range in the percentage of non-white students in the persistently low performing non-special education group was slightly smaller with a low of 52% non-white students at grade 5 in South Dakota to a high of 96% in Hawaii at both grades 5 and 8 (darker bars). There were some differences in the percentage of males across grades for this population, but the size of the differences varied from small (10% or less) in Alabama, Hawaii, and South Dakota, to large (greater than 10%) in Wisconsin. When examining Figure 5, we caution the reader to keep in mind that small group sizes can affect the appearance of the magnitude of changes in student demographics.

Persistently low performing non-special education students in all states were more likely to be non-white compared to the total population of non-special education students. In Alabama, South Dakota, and Wisconsin there were substantially larger differences (30% or larger) in the percentage of non-white students in the persistently low performing group (darker bars) compared to the total non-special education group (lighter bars). In Hawaii there was a small (less than 10%) difference in the percentage of non-white students across the two groups.

Figure 6 provides a look at the percentage of non-white students in the persistently low performing group and in the total population of special education students (see Appendix A, Table A7 for more detail). Figure 6 indicates that the amount of racial and ethnic diversity varied considerably in the total special education population (lighter bars) and the persistently low performing special education group (darker bars) of the four project states. The percentage of non-white students in the total special education population at grades 5 and 8 (lighter bars) ranged from roughly 20% in South Dakota at grade 8 to 87% in Hawaii. This range is similar to the range in the percentage of non-white non-special education students taking the reading test (See Figure 5). In the persistently low performing special education group (darker bars) the percentage of males ranged from 34% in South Dakota at grade 8 to 90% in Hawaii at grade 5.

Figure 6. Percentage of Special Education Grade 5 and 8 Non-White Students in the Persistently Low Performing Group on the State Reading Test Compared to the Percentage of Non-White Students Among all Special Education Students

Figure 6 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

Across states, the percentage of non-white students in the persistently low performing special education group was consistently higher than the percentage of non-white students in the total special education population. In some states such as Hawaii, the size of this difference was relatively small (less than 10%) and in others such as South Dakota it was relatively large (more than 10%). Again, we caution the reader to keep in mind that the size of the total student population and state demographics play a role in the size of this difference.

Comparing bars across grade levels, the percentages in Figure 6 do not highlight any sizeable differences (e.g., 10% or above) between fifth and eighth grade in the percentage of non-white students in the persistently low performing special education students on the state reading assessment.

Math. Figures 7 and 8 present similar data to Figures 5 and 6 but they show data for math instead of reading. Figure 7 includes the non-special education data and Figure 8 includes the special education data (see Appendix A, Tables A8 and A9 for more detail).

Figure 7 shows that there was a great deal of variability across states in the percentage of non-white students in the non-special education group taking the math test. The percentage of non-white students in the total non-special education population (lighter bars) ranged from a low of 13% to 15% in South Dakota to a high of 87% to 88% in Hawaii. The percentage of non-white students in the persistently low performing non-special education group (darker bars) ranged from 58% in South Dakota at grade 8 to about 94% in Hawaii at both grades.

In each state, the percentage of non-white students in the persistently low performing non-special education group (darker bars) was greater than the percentage in the total non-special education population at the same grade. In Alabama, South Dakota, and Wisconsin the size of this difference was quite large (approximately 30%). In Hawaii the difference between the percentages of non-white students in the two groups was fairly small (less than 10%). As with previous figures, the size of these differences should be interpreted cautiously because it relates to the size of the student population in each state and to that state's overall student demographics.

Except for Wisconsin, there were only small differences (<10%) across grades in the percentage of non-white non-special education students who were persistently low performing on the math test.

Figure 7. Percentage of Non-Special Education Grades 5 and 8 Non-White Students in the Persistently Low Performing Group on the State Math Test Compared to the Percentage of Non-White Students Among all Non-Special Education Students

Figure 7 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

Figure 8 shows similar data for the fifth and eighth grade special education population on the math test. As shown in this figure, the percentage of non-white students in the total special education population taking the math test (lighter bars) varied a great deal across the four states. The range in the percentage of non-white special education students stretched from a low of 37% in South Dakota at grade eight to a high of 86% in Hawaii at grade 8. The percentage of non-white students in the persistently low performing group of special education students was also extremely variable. South Dakota, with 37% at grade 8, had the smallest percentage of non-white persistently low performers. Hawaii, with approximately 91% at grade 5, had the largest percentage of non-white persistently low performers in special education. The substantial number of non-white students in special education in Hawaii reflects the overall diversity of the state's total student population.


Figure 8. Percentage of Special Education Grades 5 and 8 Non-White Students in the Persistently Low Performing Group on the State Math Test Compared to the Percentage of Non-White Students Among all Special Education Students

Figure 8 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

For most states there is a general pattern of a larger percentage of non-white students in the persistently low performing special education population (darker bar) compared to the total special education population (lighter bar) on the math test. With the variations in the overall degree of student diversity the size of this difference fluctuated. There was no difference in the percentage of non-white students in the two groups for Hawaii's eighth grade. A small difference (less than 10%) existed for Hawaii's fifth grade. In the remaining three states there were large (greater than 10%) differences between the percentage of non-white students in the persistently low performing special education group and in the total population of special education students.

There was a small difference (10%) across grade levels in the percentage of non-white special education and persistently low performing special education students.

Low-income Background

Reading. The final student characteristic that we investigated in our analyses was income level. Figures 9 and 10 present the percentages of low-income students in the persistently low performing and total populations of special education and non-special education students. The data variable used for these analyses was the percent of students receiving free or reduced lunch at school. Students receive free or reduced lunch at school on the basis of their family's income level. Alabama did not have data on free or reduced lunch status in the state database. Therefore, the column for Alabama is left blank in the figures in this section. Figure 9 presents the non-special education data at grades 5 and 8, and Figure 10 presents the special education data at grades 5 and 8 (see Appendix A, Tables A10 and A11 for more detail).

In Figure 9 the percentage of low-income students in the non-special education population (lighter bars) ranged from 27% for South Dakota and Wisconsin eighth graders to 43% for Hawaii's fifth graders (lighter bars). In the persistently low performing non-special education group (darker bars) the percentage ranged from 50% in South Dakota at eighth grade to 80% in Hawaii at fifth grade. For the most part, differences across grades within either group were small (10% or less) except in Hawaii where there were 12% more low income persistently low performing students in fifth grade compared to eighth grade (i.e., a large difference).

Figure 9. Percentage of Non-Special Education Grades 5 and 8 Low-Income Students in the Persistently Low Performing Group on the State Reading Test Compared to the Percentage of Low-income Students Among all Non-Special Education Students

Figure 9 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

In the figure we can see a consistent pattern of higher percentages of low-income students in the persistently low performing non-special education population (darker bars) compared to the total non-special education population (lighter bars). The differences in the percentage of low income students across the two groups were sizeable (greater than 20%) for the three states with data.

Figure 10 presents the data for fifth and eighth grade special education students on the reading assessment. There were varying percentages of low income students in the special education population overall and in the persistently low performing special education group. Low income students represent slightly less than half (46% to 48%) of the total special education population (lighter bars) in South Dakota and Wisconsin, but more than half of the population (57% to 60%) in Hawaii. In the persistently low performing special education group (darker bars) in South Dakota, low income students made up about half (46% and 55%) of the students while in both Wisconsin and Hawaii they made up greater than 60% of the group. Differences across grade levels in the percentage of low-income persistently low performing special education students were minimal (10% or less) and represented some increases and some decreases.

Figure 10. Percentage of Special Education Grades 5 and 8 Low-Income Students in the Persistently Low Performing Group on the State Reading Test Compared to the Percentage of Low-Income Students Among all Special Education Students

Figure 10 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

In all three states with income data, there was a greater percentage of low income students in the persistently low performing special education group compared to the total special education population. However, the magnitude of the difference varied. In Wisconsin the size of the difference was large (greater than 10%) while in the other two states, South Dakota and Hawaii, it was small (10% or less).

Math. Figures 11 and 12 contain graphs that show the percentages of low-income students in the persistently low performing student groups on the state math test. Figure 11 shows the non-special education group, and Figure 12 shows the special education group (see Appendix A, Tables A12 and A13 for more detail).

Figure 11. Percentage of Non-Special Education Grades 5 and 8 Low-Income Students in the Persistently Low Performing Group on the State Math Test Compared to the Percentage of Low-Income Students Among all Non-Special Education Students

Figure 11 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

Figure 11 highlights the varying percentages of low income students in the total non-special education and persistently low performing non-special education groups on the math test. The percentage of low income students in the total population (lighter bars) ranged from 27% for South Dakota and Wisconsin eighth graders, to 43% for Hawaii fifth graders. In comparison, the percentage of low income students in the subgroup of persistently low performers (darker bars) ranged from 54% in South Dakota at eighth grade to 74% in Hawaii at fifth grade and Wisconsin at eighth grade. Differences within a group across grades were typically small (10% or less) except in Hawaii where there was a large difference (greater than 10%) with fewer low income students in the persistently low performing group at grade 8 compared to grade 5.

In each of the three states with data on family income, the percentage of low income students in the persistently low performing non-special education group was substantially higher, 26% to 47% more, than the percentage of low-income non-special education students overall.

Figure 12 shows data for all special education students and persistently low performing special education students taking the state math test at grades 5 and 8.

Figure 12. Percentage of Special Education Grades 5 and 8 Low-Income Students in the Persistently Low Performing Group on the State Math Test Compared to the Percentage of Low-Income Students Among all Special Education Students

Figure 12 Bar Chart

*PLP: Persistently Low Performing students were students who scored at the 10th percentile or below for three consecutive years.

On the math test, the percentage of low-income students in the total special education population (lighter bars) varied from just under 50% in South Dakota and Wisconsin to over 57% to 60% in Hawaii. These percentages did show small differences (10% or less) across grades in two of the three states with income data. In the subgroup of persistently low performing special education students, those from low income backgrounds represented between 51% (South Dakota, grade 8) and 68% (Hawaii grade 5) of the group. Again, there were small differences across grades (10% or less).

Across the three states with data, there was a larger percentage of low-income students in the persistently low performing group of special education students (darker bars) than there was in the total special education population (lighter bars). The size of this difference varied from small (10% or less) in Hawaii at both grades and South Dakota at grade 8 to large (more than 10%) in South Dakota at grade 5 and Wisconsin at both grades.

Conclusion for Research Question 1

The discussion of Figures 1-12 highlighted that there are more nuances to the demographic data on persistently low performers in special education and non-special education settings than previous research has suggested. If we use the percentage of students with a particular demographic characteristic in the total population as our point of comparison for the persistently low performers we see that there are mixed findings on whether or not there are more males in the persistently low performing subgroup of either non-special education or special education students. Patterns varied for males on the reading test versus the math test and for special education students versus non-special education students. In comparison, there were clear indications that persistently low performers, in both the special education and non-special education groups, were more likely to be non-white and low-income than their higher performing peers. These results were true on the reading and the math test. There were few consistent differences in the demographic make-up of the groups across grade levels.

Although we did not specifically compare variations in the demographic characteristics of a group taking the math test versus the reading test, we did note that for gender, there were different patterns of gender representation across the two tests for the non-special education group.

 

Research Question 2: Are special education students more or less likely to move out of the persistently low performing group than non-special education students? Is there any variation in movement by content area?

As Tables 2 and 3, along with Figures 13 through 16, will demonstrate, low performing special education students were more likely to be identified as persistently low performing across three years than their non-special education peers at the same grade. There were no major differences in the percentages of students moving out of the low performing groups for mathematics compared to reading and patterns were similar across grades. These findings are described in more detail in the sections that follow.

Overall Movement Out of the Low-performing Category

Table 2 shows the total number of all fifth and eighth grade students who were identified as low performing (i.e., at or below the 10th percentile) in the first year in our data set. We grouped the students into what we call "movement categories" according to the years in which they were identified as low performing.

In the left-hand column of the table, Persistently Low Performing (PLP) students are those who were low performing in each of the three years in our data set. In the middle two rows under "Movement Category" for each grade are students who were identified as low performing twice within three consecutive years, including the first year. In the last row in each grade, "Initially low performing" students are those who were identified as low performing only in the first year of our data set. The second and third years these students obtained a score above the 10th percentile.

Table 2. Total Number and Percentage of Fifth Grade and Eighth Grade Low Performing Students in Different Movement Categories in the Reading Assessment

Grade

Movement Category

AL

HI

SD

WI

Count

%

Count

%

Count

%

Count

%

5

Persistently low performing

1,778

48.4

566

47.1

344

40.1

2,200

42.0

Low performing in year 1 and 3

333

9.1

145

12.1

96

11.2

584

11.1

Low performing in year 1 and 2

598

16.3

196

16.3

182

21.2

976

18.6

Initially low performing

962

26.2

295

24.5

236

27.5

1,479

28.2

Total

3,671

100.0

1,202

100.0

858

100.0

5,239

100.0

8

Persistently low performing

1,740

43.8

478

42.2

377

41.1

2,757

48.3

Low performing in year 1 and 3

343

8.6

144

12.7

114

12.4

689

12.1

Low performing in year 1 and 2

824

20.8

200

17.7

186

20.3

873

15.3

Initially low performing

1,064

26.8

310

27.4

240

26.2

1,389

24.3

Total

3,971

100.0

1,132

100.0

917

100.0

5,708

100.0

In Table 2 students who were persistently low performers (i.e., low performing for three years) made up the largest percentage of all low performing students on the reading test (from 40.1% to 48.4%). Students who were initially low performing in year one and scored higher than the tenth percentile in years two and three represented the next largest group (24.3% to 28.2%). There were no obvious differences across grade levels in the percentage of students who fell into a particular movement category on the reading assessment.

We also looked at the total number of students in each state who fell into a particular performance category in math (see Table 3) to see whether there were any differences across grade levels.

Table 3. Total Number and Percentage of Fifth Grade and Eighth Grade Low Performing Students in Different Movement Categories in the Math Assessment

Grade

Movement Category

AL

HI

SD

WI

Count

%

Count

%

Count

%

Count

%

5

Persistently low performing

1,616

40.6

531

43.8

337

42.2

2,071

39.9

Low performing in year 1 and 3

393

9.9

172

14.2

109

13.7

603

11.6

Low performing in year 1 and 2

733

18.4

196

16.2

131

16.4

931

17.9

Initially low performing

1,239

31.1

314

25.9

221

27.7

1,587

30.6

Total

3,981

100.0

1,213

100.0

798

100.0

5,192

100.0

8

Persistently low performing

1,201

28.2

450

39.4

402

41.0

2,630

46.0

Low performing in year 1 and 3

600

14.1

143

12.5

119

12.1

672

11.7

Low performing in year 1 and 2

1,162

27.3

264

23.1

198

20.2

975

17.0

Initially low performing

1,293

30.4

285

25.0

262

26.7

1,444

25.2

Total

4,256

100.0

1,142

100.0

981

100.0

5,721

100.0

As Table 3 shows, in math there was some variability in the percentage of students within a particular performance group across grades and states, but persistently low performing students still represented the largest group of low performers in most cases. Similar to reading, the second largest group of low performers was typically those who were initially low performing in year one of the data and who scored above the tenth percentile in years two and three. The one exception to this pattern was the state of Alabama, where a large difference (greater than 10%) was observed in the percentage of persistently low performing students at grade eight compared to grade five. In the same state a small difference (10% or less) was observed in the percentage of low performing in years one and two students.

Movement on the Reading Test

Figure 13 uses stacked bar charts to represent the percentage of non-special education and special education fifth graders in each state and the movement category into which they fell on the state reading test (see Appendix B, Table B1 for more detail). This figure includes all fifth grade students who were identified as low performing (i.e., at or below the 10th percentile) in year one (2005-2006 in Wisconsin; 2004-2005 in all other states) of our data set. The non-special education bar is on the left side of each column and the special education bar is on the right side. The darkest colored tier, at the bottom, shows the percentage of students in either general or special education who were persistently low performing (i.e., below the 10th percentile each of three years). The second tier from the bottom is the percentage of students who were low performing in years one and three only. The third tier from the bottom shows the students who were low performing in years one and two only, followed by the students who were initially low performing (i.e., low performing only in year one).

Figure 13. Percentage of Fifth Grade Non-Special Education and Special Education Low Performing Students in Each Movement Category on the Reading Test

Figure 13 Bar Chart

The left-hand bar of each state in Figure 13, shows that on the state reading assessment a higher overall percentage of fifth grade special education students were identified as persistently low performing compared to their non-special education peers (right-hand bar in each state). Across the four states, 27% to 32% of low performing non-special education fifth graders on the reading test were persistently low performing. In contrast, 52% to 66% of low performing fifth grade special education students (the right hand bar in each column) were persistently low performing.

Of the students who initially scored below the tenth percentile, non-special education students were more likely to move out of the low-performing category over time in comparison to the special education students. These students are represented by the top two tiers of each bar. Adding together these two tiers for each state, a range of 57% to 60% of non-special education students who were initially low performing were no longer low performing by year three on the reading test. They scored above the 10th percentile in years two and three. In contrast, only 26% to 38% of fifth grade special education students who were initially low performing (see the top two tiers of the right hand bar in each column) were no longer low performing by year three.

Figure 14 shows comparable data for all eighth grade students who took the reading assessment and who were identified as low performing in year one of our data set (see Appendix B, Table B1 for more detail). Similar to Figure 13, each state's column includes a bar for non-special education students on the left side, and a bar for special education students on the right side. The bottom tier of each bar represents the students who continued to be identified as low performing in each of the three years (i.e., persistently low performing). The second and third tiers represent the percentage of students identified as low performing in only two out of the three years. The top tier of each bar represents the percentage of students who were identified as low performing only in year one (i.e., initially low performing). By years two and three these students scored at or above the 10th percentile on the reading assessment.

Figure 14. Percentage of Eighth Grade Non-Special Education and Special Education Low Performing Students in Each Movement Category on the Reading Test

Figure 14 Bar Chart

Figure 14 indicates that for eighth graders taking the state reading assessment, a greater percentage of special education students were identified as persistently low performing compared to their non-special education peers. Twenty-three percent to 32% of non-special education students who were low performing in the first year of the data set (see the left hand bar in each column) were persistently low performing across three years. In contrast, 56% to 61% of eighth grade special education students (see the right hand bar in each column) were persistently low performing.

Of the eighth grade students who initially scored below the tenth percentile, special education students were less likely than their non-special education peers to score above the tenth percentile in reading in later years. Fifty-five percent to 62% of eighth grade non-special education students who were low performing (see left-hand bars in each column) in year one moved out of the low-performing group by year three on the reading test. In contrast, only 28% to 34% of eighth grade special education students who were initially identified as low performing (see the right hand bars in each column) moved out of the low performing group by year three.

Movement on the Math Test

Figures 15 and 16 show the percentage of non-special education and special education fifth and eighth graders in the various movement categories on the state math test (see Appendix B, Table B2 for more details). Figure 15 presents all fifth grade students who took the state math assessment and were identified as low performing in year one of our data set. The tiers within each bar indicate the percentage of those students who either remained low performing in each subsequent year, or moved out of the low-performing group at least temporarily. Each state's column includes a bar for non-special education students on the left side, and a bar for special education students on the right side. The bottom tier of each bar represents the students who were persistently low performing. The two middle tiers represent the percentage of students identified as low performing twice within three consecutive years, including the first year of data collection. The top tier of each bar represents the percentage of students who were identified as low performing only in year one. In years two and three these students scored above the 10th percentile on the math test.

Figure 15 shows that on the fifth grade math test a greater percentage of low performing special education students were persistently low performing compared to their non-special education peers. Between 26% and 32% of fifth grade non-special education students who were initially identified as low performing in math (the left-hand bar in each column) were persistently low performing. In contrast, 51% to 61% of low performing fifth grade special education students were persistently low performing.

Of the students who were low performing, non-special education students were more likely than special education students to score higher than the tenth percentile at some time during the three-year period. Between 53% and 64% percent of initially low performing non-special education students moved out of the low-performing group by year three (see the top two tiers of the left-hand bar in each state). In contrast, only 27% to 37% of low performing special education students moved out of the low-performing group on the math test by year three.

Figure 15. Percentage of Fifth Grade Non-Special Education and Special Education Low Performing Students in Each Movement Category on the Math Test

Figure 15 Bar Chart

Figure 16 presents data for eighth grade non-special education and special education students who took the state math test.

Figure 16 shows that on the eighth grade math test a greater percentage of low performing special education students were persistently low performing over a three year period compared to non-special education students. Twenty-one percent to 35% of eighth grade non-special education students who were initially identified as low performing in math (the left hand bar in each column) were persistently low performing. In contrast, 38% to 56% of low performing special education students were persistently low performing across three years.

Figure 16. Percentage of Eighth Grade Non-Special Education and Special Education Low Performing Students in Each Movement Category on the Math Test

Figure 16 Bar Chart

Examining all of the low performing students who made some kind of movement in their test scores, initially low performing non-special education students were much more likely than their special education peers to score above the tenth percentile in years two or three. For non-special education students, 52% to 65% of the low performing eighth graders in math (the right hand column) in year one had moved out of the low-performing group by year three. In contrast, only 33% to 48% of low-performing students in special education (the left hand column) moved out of the low-performing group by year three.

Conclusion for Research Question 2

The discussion of Tables 5 and 6 and Figures 13-16 indicate that low performing special education students tend to stay low performing (i.e., become persistently low performing) more frequently than the low-performing non-special education students in the same grade level. These results were true for both reading and math and for fifth grade as well as eighth grade.

Although our analyses did not specifically compare the percentages of persistently low performing students on the reading versus the math test, we did note that there appear to be some large differences in the percent of either special education or non-special education persistently low performers across tests.

 

Research Question #3: Are low performing special education students in some demographic groups more or less likely to move out of the low performing group than their peers in non-special education?

For this research question, which contains multi-categorical analyses, we narrowed the focus solely to fifth grade reading data because the data are extremely complex. The fifth grade math data generally show similar patterns (see Appendix C, Tables C4, C5, and C6 for more details).

The discussion of Figures 17 through 19 indicate that low-performing special education students who were male, non-white and low income were more likely than their non-special education peers to be identified as persistently low performing on the reading test (i.e., low performing for all three years of the data analysis). In this section we provide more detailed description of the data to support this finding.

Gender

Figure 17 provides data for all fifth grade male students who took the state reading assessment and were identified as low performing in year one of our data set (see Appendix C, Table C1 for more detail). Each state's column includes a bar for non-special education male students on the left side, and a bar for special education male students on the right side. Each bar is then divided into tiers that represent performance levels. The bottom tier of each bar represents the persistently low performing students (i.e., those identified as low performing in each of the three years). The two middle tiers of each bar represent the percentage of students identified as low performing in the first year of the dataset, and in one additional year. Thus, the students in the middle tiers were not low performing in one year of the study. The top tier of each bar represents the percentage of students who were identified as low performing only in year one. After the first year, these students moved out of the low performing group.

Figure 17 shows that on the state reading assessment a greater percentage of low performing special education males were persistently low performing across three years in comparison to their non-special education peers. Twenty-eight percent to 33% of fifth grade low performing non-special education males (the left hand bar in each column), were persistently low performing. In comparison, 51% to 68% of low performing males in special education (the bottom tier of the right-hand bar) were persistently low performing. The percentage of persistently low performing males in special education was higher in Alabama (68%) and Hawaii (64%) compared to Wisconsin (54%) and South Dakota (51%).

Figure 17. Percentage of Fifth Grade Male Non-Special Education and Special Education Low Performing Students in Each Movement Category on the Reading Test

Figure 17 Bar Chart

A larger percentage of low-performing males in the non-special education group, compared to those in special education, moved out of the low-performing group on the reading test by year three. Across states, roughly 57% to 60% of fifth grade non-special education low-performing male students (left-hand bar of each column) scored higher than the 10th percentile by year three. A much smaller percentage of low-performing male students in special education (right-hand bar of each column) moved out of the low-performing group by year three (26% to 39%).

Race/Ethnicity

Figure 18 presents all fifth grade non-white students who took the state reading assessment and were identified as low performing in year one of our data set (see Appendix C, Table C2 for more detail). Each state's column includes a bar for non-white non-special education students on the left side, and a bar for non-white students in special education on the right side. Each bar is then divided into tiers that correspond to the different performance groups. The bottom tier of each bar represents the students who were persistently low performing. The two middle tiers of each bar represent the percentage of students identified as low performing in the first year of the dataset, and in one additional year, indicating that they were not low performing in one year out of three. The top tier of each bar represents the percentage of students who were identified as low performing only in year one. After the first year, these students were no longer low performing (i.e., achieved at or above the 10th percentile).

Figure 18. Percentage of Fifth Grade Non-White Non-Special Education and Special Education Low Performing Students in Each Movement Category on the Reading Test

Figure 18 Bar Chart

In Figure 18 we can see that, of the fifth grade low-performing students, a larger percentage of non-white students in special education (right-hand bar in each column) were persistently low performing compared to their non-special education peers (left-hand bars). Twenty-eight percent to 40% of low performing non-white non-special education students were persistently low performing. In comparison, the right-hand bar in each state shows that 64% to 69% of low performing non-white students in special education were persistently low performing.

The top two tiers of each bar, representing students moving out of the low performing category after year 1, illustrate that a larger percentage of non-special education students move out of the low performing category over time, in comparison to their special education peers. Of the fifth grade low-performing students, 49% to 59% of non-white non-special education students moved out of the low performing group by year three. In contrast, the right hand bar shows that only 26% to 30% of non-white low-performing students in special education moved out of the low performing group by year three.

Income Level

Figure 19 illustrates the movement of all fifth grade low-income students who took the state reading assessment and were identified as low performing in year one of our data set (see Appendix C, Table C3 for more detail). Similar to previous figures, each state's column includes a bar for non-special education low-income students on the left side, and a bar for low-income students in special education on the right side. The bottom tier of each bar represents the students who continued to be identified as low performing in each of the three years (i.e., persistently low performing). The two middle tiers of each bar represent the percentage of students identified as low performing in the first year of the dataset and in one additional year, indicating that they were not low performing in one year out of three. The top tier represents the percentage of students who were identified as low performing only in year one. After the first year, these students were no longer low performing.

Figure 19. Percentage of Fifth Grade Low-Income Non-Special Education and Special Education Low Performing Students in Each Movement Category on the Reading Test

Figure 19 Bar Chart

Figure 19 shows that, of fifth grade low performing low-income students, a larger percentage in special education (right hand bar in each column) were persistently low performing compared to their non-special education peers (left-hand bar in each column). Thirty-one percent to 33% of fifth grade low performing low income non-special education students were persistently low performing. In contrast, 52% to 66%, of low performing low income students in special education were identified as persistently low performing.

Of the low performing low income students, those who were not in special education were more likely than those in special education to score higher than the tenth percentile by year three. Looking at the top two tiers of the left-hand bar in each state we can see that 54% to 56% of low income non-special education students moved out of the low-performing group by year three compared to 23% to 36% of low income special education students.

Conclusion for Research Question 3

The discussion of Figures 17 through 19 indicates that there are some identifiable patterns of movement in the low performing group on the reading assessments that are different for students from certain demographic categories. Low performing special education students who are male, non-white and low income are be more likely to become persistently low performing. Their non-special education classmates from the same demographic groups are more likely to score in the low performing range once and move out of the group in later years.


Discussion and Conclusions

The results of this study suggest three key findings. First, the demographic characteristics of persistently low performing students were generally similar regardless of whether students received special education services. In both groups, the lowest performers were more likely to be non-white students and students from low-income backgrounds. The data from the four states we studied did not conclusively show a pattern of more persistently low performing males, compared to females, across all content areas and grade levels. Second, of the students who were identified as low performing in the first year of our data, special education students were more likely than their non-special education peers to remain low performing across the three years of the study. Non-special education students were more likely to score higher in subsequent years while the special education students' scores remained relatively unchanged. Third, looking at only the low performing special education students, those who were non-white and from low income backgrounds were more likely than their white, higher-income level peers, to be identified as persistently low performing across three years. The findings for males were less consistent.

There are some important implications of these findings for K-12 standards-based assessment and instruction. The results show that many low performing students on state assessments do not have disabilities, and therefore would not be eligible to take an alternate assessment based on modified achievement standards (AA-MAS). Because low performing non-special education students are not eligible for the AA-MAS, state departments of education should consider other actions to help these students succeed in the grade-level standards-based curriculum. There is a wide-ranging variety of system-level factors that might play into differential performance patterns by particular subgroups of students. These factors include: school poverty levels and resource availability, teacher availability, teacher qualifications and training, student attendance, student health, school violence, an emphasis on remedial instruction rather than grade-level standards-based instruction, a teacher-student culture and relationship gap, and appropriate state assessment design and implementation for a particular population of students (cf. Archibald, 2006; Belfiore, Auld, & Lee, 2005; Cooper & Jackson, 2011; Goodwin, Englert, & Cicchinelli, 2003; Jackson, 2011). It is not within the scope of this report to address all the possible ways in which state departments of education and policymakers might address issues related to low levels of academic achievement by low income and minority students with, and without, disabilities. However, from an assessment fairness perspective (cf. Goodwin et al., 2002) there are three key issues state departments of education could investigate when faced with data showing differential achievement outcomes of this magnitude for certain subgroups of students:

  1. Examine whether existing assessment items function differently for some groups of students compared to others. Test items should be designed to allow the greatest possible number of students to demonstrate their knowledge and skills in the grade-level content. Barriers such as biased test content or inaccessible test item formats should be identified and attended to early.
  2. Examine whether correct test participation and administration policies and procedures are adhered to. Are all students participating in the appropriate state assessments according to state-determined participation criteria? Do students receive the test accommodations that they need as indicated by their IEPs, or more generally by state accommodations policies if accommodations are offered to any student? Are accommodations administered correctly?
  3. Examine whether all students have had sufficient opportunity to learn the grade-level standards-based content that is measured on the state test. Are high academic expectations the norm for all students or only for some? For the lowest performing students on the state assessment, how well are their curricula aligned with grade-level standards? If students are instructed in classrooms where curricula are well-aligned with state standards, what strategies are teachers using to get to know the skills and interests of their low-income, non-white students so that content instruction can build on students' strengths and develop their academic potential? A part of addressing students' opportunity to learn would also include investigation of the progress monitoring and feedback processes educators use to adapt daily instruction and provide targeted interventions for specific students who have learning challenges.

Federal legislation describing the AA-MAS option states that students with disabilities who are being considered as possible candidates for participation in the AA-MAS must have access to grade-level content (U.S. Department of Education, 2007). Further research is needed to get a better understanding of whether low performing students with disabilities have access to grade-level content—though preliminary evidence indicates that some may not have access (see, for example, Altman, Cormier, Lazarus & Thurlow, 2011; Lazarus et al., 2011).Taking steps to ensure that all low performing students with disabilities have access to grade level content is a crucial part of improving their academic outcomes. Improving the opportunity to learn for low performing students with disabilities can also have far-reaching effects that improve academic outcomes for low performing students who are not in special education and vice versa (see Telfer, 2011). Given that students with disabilities may move in and out of special education status repeatedly over time (cf. Ysseldyke & Bielinski, 2002), our data indicate the importance of addressing the instruction and assessment needs of all low performing students simultaneously, regardless of which assessment they take.


References

Albus, D., Lazarus, S. S., Thurlow, M. L., & Cormier, D. (2009). Characteristics of states' alternate assessments based on modified academic achievement standards in 2008 (Synthesis Report 72). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Albus, D., Thurlow, M. L., & Bremer, C. (2009). Achieving transparency in the public reporting of 2006-2007 assessment results (Technical Report 53). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Altman, J. R., Cormier, D. C., Lazarus, S. S., & Thurlow, M. L. (2011). Tennessee special education teacher survey: Training, large-scale testing and TCAP-MAAS administration (Technical Report 61). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Altman, J., Rogers, C., Bremer C., & Thurlow, M. (2010). States challenged to meet special education targets for assessment indicator (Technical Report 55). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Altman, J., Thurlow, M., & Vang, M. (2010). Annual performance report: 2007-2008 state assessment data. Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Archibald, S. (2006). Narrowing in on educational resources that do affect student achievement. Peabody Journal of Education, 81(4), 23-42.

Bechard, S., & Godin, K. (2007). Who are the students in the gaps, what are their attributes, and how do they perform? Newton MA: CAST, The Education Alliance, EDC, INTASC, Measured Progress. Retrieved from http://www.necompact.org/Students_in_the_Gaps.pdf

Belfiore, P., Auld, R., & Lee, D. (2005). The disconnect of poor-urban education: Equal access and a pedagogy of risk taking. Psychology in the schools, 42(8), 855-863.

Chudowsky, N., & Chudowsky, V. (2009). State test score trends through 2007-2008, part 4: Has progress been made in raising achievement for students with disabilities? Retrieved from Center on Education Policy website: http://www.cep-dc.org/publications/index.cfm?selectedYear=2009

Chudowsky, N., & Chudowsky, V. (2010). State test score trends through 2007-2008, part 5: Are there differences in achievement between boys and girls? Retrieved from Center on Education Policy website: http://www.cep-dc.org/publications/index.cfm?selectedYear=2010

Chudowsky, N., Chudowsky, V., & Kober, N. (2009). State test score trends through 2007-2008, part 3: Are achievement gaps closing and is achievement rising for all? Retrieved from Center on Education Policy website: http://www.cep-dc.org/publications/index.cfm?selectedYear=2009

Cooper, E., & Jackson, Y. (2011). The ‘fierce urgency of now': Narrowing the race and culture gaps between students and teachers. Education Week, 30(18), 22-23.

Cortiella, C. (2007). Learning opportunities for your child through alternate assessments: Alternate assessments based on modified academic achievement standards. Minneapolis MN: University of Minnesota, National Center on Educational Outcomes.

Colorado Department of Education. (2005, December). Assessing "students in the gap" in Colorado: Report from the HB 05-1246 study committee. Denver, CO: Author.

Goodwin, B., Englert, K., & Cicchinelli, L. (2003). Comprehensive accountability systems: A framework for evaluation. Aurora, CO: Mid-continent Research for Education and Learning.

Hess, K., McDivitt, P., & Fincher, M. (2008). Who are the 2% students and how do we design test items and assessments that provide greater access for them? Results from a pilot study with Georgia students. Paper presented at the CCSSO National Conference on Student Assessment Orlando, FL. Retrieved from http://www.nciea.org/publications/CCSSO_KHPMMF08.pdf

Jackson, Y. (2011). The pedagogy of confidence: Inspiring high intellectual performance in urban schools. New York: Teachers College Press, Columbia University.

Kober, N., Chudowsky, N., & Chudowsky, V. (2010). State test score trends through 2008-09, part 2: Slow and uneven progress in narrowing gaps. Retrieved from Center on Education Policy website: http://www.cep-dc.org/publications/index.cfm?selectedYear=2010

Lazarus, S. S., Hodgson, J. R., Price, L. M., & Thurlow, M. L. (2011). States' participation guidelines for alternate assessments based on modified academic achievement standards (AA-MAS) in 2010 (Synthesis Report 82). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Lazarus, S. S., & Quenemoen, R. (2011). Instructing and assessing students who may be candidates for alternate assessments based on modified achievement standards (AA-MAS). In M. Russell & M. Kavanaugh (Eds.), Assessing students in the margins (pp.215-233). Charlotte, NC: Information Age Publishing.

Lazarus, S. S., Wu, Y. C., Altman, J., & Thurlow, M. L. (2010). NCEO brief: The characteristics of low performing students on large-scale assessments. Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Marion, S., Gong, B., & Simpson, M. A. (2006, Feb. 6). Mining achievement data to guide policies and practices on assessment options. Teleconferences on Making Good Decisions on NCLB Flexibility Options. Minneapolis: National Center on Educational Outcomes. Available at: http://www.cehd.umn.edu/nceo/Teleconferences/tele11/default.html

Quenemoen, R. (2008). A brief history of alternate assessments based on alternate achievement standards (Synthesis Report 68). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Telfer, D. (2011). Moving your numbers: Five districts share how they used assessment and accountability to increase performance for students with disabilities as part of district-wide improvement. Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Thurlow, M., Bremer, C., & Albus, D. (2008). Good news and bad news in disaggregated subgroup reporting to the public on 2005-2006 assessment results (Technical Report 52). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

U.S. Department of Education (2007, April 9). Final Rule 34 CFR Parts 200 and 300: Title I-Improving the Academic Achievement of the Disadvantaged; Individuals with Disabilities Education Act (IDEA). Federal Register. 72(67), Washington DC: Author.

VanGetson, G. R., & Thurlow, M. L. (2007). Nearing the target in disaggregated subgroup reporting to the public on 2004-2005 assessment results (Technical Report 46). Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes.

Ysseldyke, J., & Bielinski, J. (2002). Effect of different methods of reporting and reclassification on trends in test scores for students with disabilities. Exceptional Children, 68(2), 189-201.


© 2013 by the Regents of the University of Minnesota.
The University of Minnesota is an equal opportunity educator and employer.

Online Privacy Statement
This page was last updated on January 03, 2013

NCEO is supported primarily through a Cooperative Agreement (#H326G050007, #H326G110002) with the Research to Practice Division, Office of Special Education Programs, U.S. Department of Education. Additional support for targeted projects, including those on LEP students, is provided by other federal and state agencies. Opinions expressed in this Web site do not necessarily reflect those of the U.S. Department of Education or Offices within it.