ELL test-takers trail far behind the other major racial/ethnic groups of test-takers partly because they are concentrated in schools that report the assessment test scores of ELL students and those schools tend to be low-achieving schools. Average proficiency rates in math are lower at ELL reporting schools not only for ELL students, but also for white and black students who attend those schools. In the NLSLSASD, the average proficiency rate for white test-takers was lower if they were educated in ELL reporting schools rather than in schools that had below the minimum number of ELL test-takers required to report the school ELL proficiency rate (in other words, a public school with few ELL test-takers) (Figure 7). The average black test-takers’ proficiency rate was also lower if they were educated at ELL reporting schools (except in Florida) (Figure 8).

It is impossible to determine whether ELL students had higher mathematics proficiency rates if they attended public school that did not have enough ELL students to report ELL assessment test scores because, by definition, ELL proficiency rates in math were not reported for public schools that were not ELL reporting schools. But the school achievement data certainly suggest that ELL math achievement is positively associated with schools that have larger numbers of white students.

Some ELL reporting schools have so few white students that the schools do not report assessment test results for white students. Figure 9 shows that ELL proficiency rates were higher at ELL reporting schools that had sufficient numbers of white students to report the white achievement results.

In the five states examined, ELL math achievement follows a consistent pattern. ELL student math proficiency rates tend to be highest at ELL reporting schools with sufficient numbers of white students to report the white results (Table 2). ELL proficiency rates are lower at ELL reporting schools with neither enough white students to report the white results nor sufficient black students to report the black results. Finally, ELL math achievement tends to be lowest at ELL reporting schools with sufficient black students to report the black results (but not enough white students to report the white results).

Other achievement data in addition to the state assessments in the NLSLSASD support the assertion that public schools with concentrations of ELL students tend to be low-achieving schools. Table 3 reports math achievement results from the 2007 National Assessment of Education Progress. The NAEP uses representative samples of students and is the basis for the well-known “Nation’s Report Card” (NCES, 2007a). Table 3 reports the average scale score (on a scale of 0 to 500) for both white and black fourth- and eighth-graders. NAEP categorizes the public school by the percentage of its student enrollment identified as LEP, or limited English proficient. White and black math achievement tends to be lower at public schools whose LEP enrollment exceeds 5 to 10 percent.

Returning to Figures 5 and 6, there are two basic reasons that the ELL achievement gaps shrink when ELL test-takers attend the same schools as white and black test-takers. First, the math achievement scores of white and black students decline if they attend schools that report ELL test scores, shrinking the gap. Second, measured ELL math achievement tends to improve when ELL students are educated at schools that have sufficient numbers of white students (reducing the difference).