Before Delaware began hailing the dramatic gains its 4th and 8th graders made on the 2002 National Assessment of Educational Progress in reading, a team of statisticians and testing experts was quietly investigating whether the scores were valid.
Officials with the National Center for Education Statistics and a private contractor spent days scrutinizing data from the state’s test-takers, as they did with several other states that showed major changes since 1998. Among the reasons for poring over the data was the rise in the rates of children with disabilities and limited mastery of English who were excluded from the test. In the end, the NCES concluded that Delaware had reason to celebrate its rising scores.
Delaware saw the most improvement in reading achievement in the nation in the two grades for which state data were collected. Among public school students, the average score for 4th graders rose 17 points, to 224 on a 500-point scale, and climbed 13 points among 8th graders, to 267, bringing the state’s NAEP scores above the national average for the first time. A larger proportion of students are demonstrating proficiency in reading as well.
Some 20 states, in fact, showed significant improvement on the NAEP scores released here last month. But federal officials are advising caution in interpreting the results, which in some places could be tempered by large increases in the number of students who were excluded from taking the reading test because of disabilities or limited English proficiency.
“Since students with disabilities or limited-English-proficient students tend to score below average on assessments, excluding students with special needs may increase a jurisdiction’s scores,” the NAEP report says. “Conversely, including more of these students might depress score gains.”
In Delaware, for example, 8 percent of the students identified as having disabilities or limited command of English did not take the test last year. In 1998, however, just 1 percent of such students weren’t tested.
But in California, a state in which sizable numbers of students are not proficient English-speakers, just 5 percent of those 4th graders were omitted from the 2002 assessment, significantly fewer than the 14 percent in 1998. The state’s average score at that grade level rose several points, but the gain is not considered statistically significant.
NCES officials said there was currently no indication, though, that the exclusion rates had significantly affected the scores for states overall. And Peggy G. Carr, the associate commissioner of the NCES, said no consistent correlation exists between states’ scores and their exclusion rates.
Since 1998, students with disabilities and those with limited English proficiency have been allowed to take NAEP tests with accommodations—such as added time or large-print text—in an effort to increase their participation rates. But NAEP does not allow the same level of assistance as do some states, such as reading the test to students and translating the test into other languages. So states are left to decide whether a student who would be allowed such help when taking a state test should be excused from the national assessment.
While the change has led to greater diversity among test-takers, local decisions about who should and should not be tested make comparisons of achievement between states more complex.
“As far as state data go, you really do have to make adjustments to interpret state data with any validity,” said David W. Grissmer, a researcher in the Washington office of the RAND Corp. “Accurate trends can be calculated, but they can’t be read from the raw data.”
Other observers, however, have questioned whether state trends on the test can be considered accurate when exclusion rates vary significantly between administrations of the reading NAEP.
Richard G. Innes, a Kentucky education activist, argues that states “buy” a point on their scale scores for each additional percent of students they exclude, giving them an unfair advantage over states with low exclusion rates.
Studying the potential effects of exclusion rates and accommodations is a “high priority” of the board that oversees NAEP, said Arnold A. Goldstein, a NAEP project director for the NCES, which manages the national assessment for the U.S. Department of Education. The NCES has been formulating alternative interpretations of the reading test results to estimate how state averages would differ if the excluded students had taken the test. So far, those studies indicate that scores would vary only slightly.
Nationwide, reading achievement among 4th graders— including the lowest-performing of those students—is showing signs of progress after a decade of state and federal initiatives to improve instruction in the early grades.
But the performance of older students on NAEP is not so promising, with the scores of 8th graders stagnating over the past four years and those of 12th graders declining, according to the latest results.
The assessment, often called “the nation’s report card,” was given January through March of last year to more than 270,000 students, including a nationally representative sample of 4th, 8th, and 12th graders. Also included was a sample of 4th and 8th graders tested in most states, as well as other U.S. jurisdictions. The reading NAEP was last given at all three grade levels in 1998, and only to 4th graders in 2000.
The test of reading comprehension required students to read passages that included fictional stories, newspaper articles, and text from print materials they might encounter every day, such as bus schedules. They were then asked questions that required multiple-choice or short written answers.
In 2002, 4th graders nationwide scored an average 219 on the 500-point scale, a 6-point increase over the 2000 test, but just a 2-point improvement—not considered statistically significant—since this test was first given in 1992. Eighth graders scored an average 264 points in 2002, essentially the same as in 1998. High school seniors turned in an average score of 287, a small, but statistically significant drop from 1998, and a 5- point drop from a decade ago.
Still, the lowest-performing 4th graders made an 11-point leap, the largest in performance among demographic groups, since 2000. The average score for 8th graders at the bottom climbed 4 points in 2002. High school seniors turned in lower scores across the board.
“The results for black, white, and Hispanic, and poor children ... are up because of a very intentional effort in some states across the nation at the very early grades,” said Mark D. Musick, a member of the National Assessment Governing Board, or NAGB. “Could these results be a blip? Perhaps, but I see them as a turnaround.”
The proportion of students who can demonstrate what the governing board has defined as “basic” skills in reading comprehension rose at the two lower grade levels as well. Almost two-thirds of 4th graders and three-fourths of 8th graders reached the “basic” level, or demonstrated “partial mastery of the knowledge and skills that are fundamental for proficient work at a given grade,” according to the report.
About a third of 4th graders and 8th graders scored at the “proficient” level, meaning they could show mastery over the challenging subject matter on the test. The percentage of 12th graders that could demonstrate basic skills dipped from 76 percent in 1998 to 74 percent last year.
The gap in achievement between black and Hispanic 4th graders, at the lower end of the performance scale, and their white peers has narrowed somewhat in the past few years. While white students’ scores rose 5 points last year, to an average 229, black students’ performance improved by 9 points, to 199, and Hispanic students’ scores increased 11 points, to 201, since 2000.