Opinion: Doing the Math on the Department of Education Interim Charter Report
By assuming all poor kids are the same, the DOE can "prove" charters are better.
Governor Christie and acting education commission Chris Cerf contend that charter schools outperform their public counterparts. Their conclusion: more charters mean fewer kids trapped in failing schools.
The evidence for this contention is thin. To bolster its case, the NJ Department of Education (DOE) released in January tables of test results showing that about three-quarters of charters had higher proficiency rates on state tests than their district peers.
Not so fast.
Columnists, respected academics and public school advocates lost no time in pointing out that meaningful performance comparisons must involve students with similar characteristics -- like free lunch eligibility, special education or English learner status.
Failing that, the comparisons cannot be used to decide which schools do the better job.
The department returned on March 11 with much more expansive documentation that -- surprise -- supported the same conclusion concerning the superiority of charters. Accompanying the multiple charts, tables and bar graphs were statements confirming and strengthening the Christie administration’s policy preferences.
Poverty status is at the core of the DOE's contention.
Essentially, the department anchors its argument that charters are pretty much like district schools when it comes to poor kids by dismissing the distinction between "free" and "reduced" lunch eligibility.
To qualify for free lunch, a student’s family must be at 100 percent or less of the federal poverty standard (now $22,350 for a family of four). Reduced lunch is available for families with incomes up to 185 percent of the standard ($41,348 for the same family).
To bolster its argument that free = reduced, the report notes that the new school aid formula combines the two, as do New Jersey reports on No Child Left Behind. It also points out that researchers generally consider the combined numbers "good enough" as a proxy for poverty.
Then, it asserts: "And most important, research shows that concentration of poverty. . . creates unique challenges and most charters in NJ cross a threshold of concentrated poverty that makes these distinctions meaningless (italics added).
No citation is offered.
One way to test the department’s conclusion is to look at the results of the National Assessment of Educational Progress (NAEP), accepted as the gold standard of standardized tests. It adopts international standards in reading and math and scales the tests so that it's possible to track student progress from the fourth to the eighth grade. Its standard for proficiency has remained consistent one test to the next, making its use over time very reliable.
NAEP reports results every two years by various subgroups, including one for students eligible for free lunch and another for those eligible for reduced lunch. In 2009, free lunch students scored a full 28 points behind the national average (204 vs. 232) on fourth grade reading. Reduced lunchers were 16 points back.
This is a very significant difference. To put it in perspective, New Jersey students were second highest on the 4th grade reading test, 12 points ahead of Tennessee, which was 47th. Moreover, gaps of the same magnitude are found on eighth grade reading and both fourth and eighth grade math tests. The gaps have remained stubbornly in place for twenty years. To dismiss such differences as "meaningless" is at least dubious, if not flat-out wrong.
The Department of Education is correct to assert that the poverty rates in charter schools are significant enough to take nothing away from those that perform dramatically well, such as Robert Treat Academy (45 percent free lunch), TEAM (60 percent), and North Star (52.9 percent) in Newark. Nevertheless, that does not justify the claim that charter schools should replace district schools to improve educational outcomes or that anyone should expect similar results in schools with much more concentrated poverty.
NAEP draws one-to-one comparisons for another subgroup, girls and boys. In 1992, girls scored 8 points better than boys on the fourth grade reading test (221 vs. 213). Seventeen years later the gap had closed by a single point. The gap on the eighth grade reading test was nine points in 2009 test (down from 13 in 1992). Incidentally, New Jersey mirrors the national numbers, as expected.
Why is this indisputable gap important to the charter discussion? First, gender is ignored in the DOE report. Second, at three of four of Newark’s consistently highest-performing charter schools, the over-enrollment of girls contributes materially to their improved performance.
At North Star almost three in five students is a girl (59.6 percent); at TEAM, the ratio is only slightly lower at 57.8 percent; and, at Gray it is 57 percent. In the Newark public schools, meanwhile, boys outnumber girls 51.3 percent to 48.7 percent.
Acting commissioner Cerf promises expanded "quantity and quality" of data for greater transparency. He might show a little curiosity about the connection between gender and achievement.
Finally, NAEP confirms the greatest demographic change to occur in recent American history. In 1992, only 7 percent of students taking the fourth grade test were Latino. By 2009 that proportion had almost tripled to 20 percent, most of it through immigration from Mexico, Central America and the Dominican Republic. The result is a sharp increase in the number of students qualifying as English learners.
A review of 30 charter schools operating in Camden, Trenton, Newark, Jersey City, Paterson, Asbury Park and Plainfield enrolling 11,603 students found only 24 English learners (0.2 percent). Their resident districts meanwhile, enrolled a total of 122,770 students of whom 12,977 were English learners (10.5 percent).
There are sensible explanations why there would be fewer English learners in charters than in public schools. None of them, however, explain why the enrollment for English learners is 52 times less than in public schools. By the way, English learners were not even mentioned in the department's Interim Report.
Before the Christie Administration bets just about everything on charter schools, it should conduct a fair and more complete assessment of the performance of similar students. When Stanford undertook a large-scale evaluation in 17 states, it did just that. It found that only 17 percent of charter school students outperformed their district peers, but 37 percent underperformed them. The rest did about the same.
Eighty-three percent doing worse or about the same does not sound like the answer to New Jersey’s educational woes.