The 2011 results are out on the Nation’s Report Card—also known as the National Assessment of Educational Progress (NAEP)—and the headlines emphasize the stability in student performance over the past few years in reading and math across the 50 states and Washington, D.C.
To be sure, more states posted gains than losses, and a few showed gains in both the fourth-grade and the eighth-grade in reading 0r math—or, in the laudable case of Hawaii, in both reading and math. But the national gains posted between 2009 and 2011 are statistically significant mainly because of the large sample size—roughly 200,000 students nationally for each of the fourth-grade and eighth-grade assessments in reading and math—not because of the absolute magnitude of the gains, which was quite small (on average, one point on the NAEP scale of 0 to 500).
Trends over the two-year period separating one NAEP administration from the next don’t tell us much, and as tempting as it may be, it’s inappropriate to treat trends in NAEP results as a quantitative referendum on particular education policies. (Yes, I’m talking to you, No Child Left Behind.) Many forces shape NAEP results, and they are best interpreted as a broad indicator of the status of the educational system, and of whether trends over time suggest that student outcomes are improving, declining, or staying about the same. There is, for example, clear evidence of substantial growth in math performance among U.S. fourth- and eighth-graders from the early 1990s to 2011. Conversely, scores in reading over the same period have been relatively flat, even though the 2011 figures match the highest scores ever recorded for both fourth- and eighth-graders.
Although I caution against the use of NAEP to evaluate the success of No Child Left Behind, the data nevertheless may inform one of the underlying principles driving the NCLB accountability system. A core idea behind NCLB is the need for a system of incentives to propel schools to ensure that all students achieve proficiency in the mandated testing grades and subject areas (especially reading and math in grades 3-8).
Recognizing that student performance varies by demographic subgroup, NCLB seeks to require schools to close achievement gaps among different subgroups over a fixed period of time. For this to happen, the average performance of students in all subgroups would need to improve each year, but the gains made by students in groups that started out behind would need to be even steeper. The fact that groups start at different levels obliges schools to seek to accelerate the learning of students in lower-performing groups, such as students eligible for free or reduced-price lunch, or Black and Hispanic students. It’s no secret that there have long been substantial differences in the average performance of students from varying social and economic backgrounds, both on the NAEP assessments and the state tests that NCLB required.
Let’s take the performance of Black and White students as an example. It’s possible to reduce the Black-White achievement gap in several ways. The least palatable version of doing so would be for the gap to close because all students’ performance declines over time, but the performance of White students declines faster than that of Black students. No one would be happy with that pattern, because achievement would be falling over time. A slight variant would be a decline in White performance coupled with no change in Black performance; but any configuration that involves stagnation for any group is likely to be viewed as antithetical to systemic progress. And even closing the gap via gains in Black performance coupled with stability in White performance is problematic and counter to NCLB’s aspirational goal of universal student proficiency.
The only configuration that is consistent with the logic of NCLB, then, is one in which we observe growth among White students, but even more growth among Black students. That is, the gains over time among Black students have to outpace the gains of White students beyond what might have occurred just by chance—and we have to be relatively confident that there have been genuine gains for White students over time as well.
To see whether this has been happening anywhere in the United States, I looked at changes in the performance of White, Black and Hispanic students between 2003 and 2011 on the NAEP fourth- and eighth-grade reading and math assessments for each of the 50 states and the District of Columbia. Some states don’t have a measurable achievement gap because there aren’t enough Black or Hispanic students in the NAEP sample to provide a stable estimate of student performance. (The District of Columbia, meanwhile, doesn’t have enough White students for a NAEP performance estimate.) The timeframe of 2003 to 2011 is arbitrary; I wanted a span of years to allow for enough time to elapse to see a trend, and it’s just a coincidence that 2003 to 2011 happens to be the “prime-time” years of NCLB, as we’re not evaluating it here. (A comprehensive evaluation of NCLB would require at least two or three blog posts.)
Here’s what I found. For each of the four categories below, I report all of the states that have made progress in narrowing racial achievement gaps along the lines mandated by NCLB.
Fourth-grade reading. New Jersey’s White fourth-graders gained 4 points from 2003 to 2011—a statistically significant gain—but its Black fourth-graders gained 16 points over the same period, significantly narrowing the Black-White in fourth-grade reading achievement from 35 points to 23 points. There was, however, no decline in the Hispanic-White achievement gap in New Jersey between 2003 and 2011, as the 4-point gain of Hispanic students matched that of White students. Thus, the pattern required by NCLB is observed in only one of the nation’s 50 states—and with just one of the two subgroup gaps I’m examining.
Eighth-grade reading. Idaho’s White students gained 4 points from 2003 to 2011, a significant gain, but since its Hispanic students gained 11 points over the same time, the Hispanic-White gap shrunk significantly from 24 points to 17 points. (Idaho doesn’t have enough Black students for a NAEP performance estimate.) And in Nevada, White students gained 7 points on the NAEP reading scale, but Black students gained 17 points, which narrowed the Black-White gap from 29 points to 19 points. But the 10-point gain of Hispanic students in Nevada from 2003 to 2010 was not significantly different from the 7-point gain of White students, so that gap did not narrow significantly. At best, then, two states show the NCLB aspirational pattern of gains for White students over time coupled with steeper gains for Black or Hispanic students.
Fourth-grade math. Two states recorded gains in the performance of White students and even higher gains for minority students. In Arkansas, Black students gained 13 points, whereas White students gained 6 points, narrowing the Black-White gap from 31 points to 25 points. And in Rhode Island, Hispanic students gained 17 points, outpacing the 10-point gain for White students, and shrinking the Hispanic-White gap from 32 points to 25 points. There is, however, no reliable evidence of a narrowing of the Hispanic-White gap in Arkansas or of the Black-White gap in Rhode Island. Two states, and one subgroup in each state, display the desired pattern of performance growth.
Eighth-grade math. Two states, Illinois and Indiana, showed significant gains between 2003 and 2011 for White students and even steeper gains for both Black and Hispanic students. An additional five states recorded gains for White students and sharper gains for one minority subgroup. In Delaware, Idaho, Massachusetts and New Mexico, the double-digit gains of Hispanic students outpaced those of White students, whereas in Kansas, the 17-point gain of Black students far outstripped the significant 3-point gain achieved by White students.
Note how little overlap there is among the states across these four cases of fourth- and eighth-grade reading and math. Few states show the pattern of accelerated gains for minority students and significant gains for White students, and no state does so consistently for both subjects and both grade levels.
What this implies is that the Nation’s Report Card provides no evidence that states can meet the laudable goal of convergence of student-subgroup performance at a significantly higher level of academic proficiency than is currently observed. No state over the past eight years has succeeded in doing this in the way that NCLB demands.
NAEP: It’s not an evaluation of NCLB—but neither is it a friend.