From the New York Times:
Triumph Fades on Racial Gap in City Schools
By SHARON OTTERMAN and ROBERT GEBELOFF
Two years ago, Mayor Michael R. Bloomberg and his schools chancellor, Joel I. Klein, testified before Congress about the city’s impressive progress in closing the gulf in performance between minority and white children. The gains were historic, all but unheard of in recent decades.
“Over the past six years, we’ve done everything possible to narrow the achievement gap — and we have,” Mr. Bloomberg testified. “In some cases, we’ve reduced it by half.”
“We are closing the shameful achievement gap faster than ever,” the mayor said again in 2009, as city reading scores — now acknowledged as the height of a test score bubble — showed nearly 70 percent of children had met state standards.
When results from the 2010 tests, which state officials said presented a more accurate portrayal of students’ abilities, were released last month, they came as a blow to the legacy of the mayor and the chancellor, as passing rates dropped by more than 25 percentage points on most tests. But the most painful part might well have been the evaporation of one of their signature accomplishments: the closing of the racial achievement gap.
Among the students in the city’s third through eighth grades, 40 percent of black students and 46 percent of Hispanic students met state standards in math, compared with 75 percent of white students and 82 percent of Asian students. In English, 33 percent of black students and 34 percent of Hispanic students are now proficient, compared with 64 percent among whites and Asians.
“The claims were based on some bad information,” said Michael J. Petrilli, a vice president of the Thomas B. Fordham Institute, a research group that studies education policy. “On achievement, the story in New York City is of some modest progress, but not the miracle that the mayor and the chancellor would like to claim.”
Reducing racial gaps in educational performance has been a national preoccupation for decades. But after substantial progress in the 1970s and ’80s, the effort has largely stalled, except for a brief period from 1999 to 2004, where there were some gains, particularly in reading, according to a report released this month by the Educational Testing Service, which develops standardized tests used across the country.
The achievement gap was also the main thrust of the No Child Left Behind law, which mandated annual testing for all students in grades three through eight and required school systems to track the performance of each racial and ethnic group, with the goal of bringing all children to proficiency by 2014.
New York City’s progress in closing its achievement gap on those tests drew national attention as a possible model for other urban school districts. It won praise from President George W. Bush as evidence that No Child Left Behind was working. In 2007, the city won a prestigious urban education prize from the Broad Foundation, which cited the city’s progress in narrowing the racial achievement gap.
But the latest state math and English tests show that the proficiency gap between minority and white students has returned to about the same level as when the mayor arrived. In 2002, 31 percent of black students were considered proficient in math, for example, while 65 percent of white students met that standard.
La Griffe du Lion explained how politicians and educators can scam the media over racial gaps in test passing rates a decade ago. Let's think about it using a stylized example with made up numbers. Say, you start off with a test in 2002 that 50 percent of black students pass and 84 percent of white students pass. You announce that there's a terrible 34 percentile point racial gap and that Something Must Be Done About It. By 2006, the passing rates rise to 84 percent for blacks and 98 percent for whites. Now, you announce, Our Brilliant Leader has cut the racial gap by 20 percentile points. High fives!
Then, in 2010 some killjoy has a crisis of conscience and the next time the test is given, blacks pass only 50 percent of the time and whites 84 percent of the time. Horrors, the racial gap has expanded from 14 percentage points to 34 percentage points!
Now, if you'd read La Griffe, you'd know that, through all the tumult and press releases and award ceremonies and accusatory articles, nothing actually happened to the racial gap. The authorities and the media just used an innumerate way to measure the racial gap — percentile differences — one that creates false senses of change whenever overall test scores go up or down.
They should have used standard deviations. In this example, the racial gap was one standard deviation in 2002, one standard deviation in 2006, and one standard deviation in 2010.
Why did the passing rates change over time? There could be lots of reasons: the test got easier, then harder; cheating was encouraged, then cracked down upon; the students learned more, then learned less; the teacher got better, then got worse; all sorts of things could have happened to drive up and down the passing rates. But the key overlooked fact is that changes in overall passing rates make it look like the racial gap is changing even when it's not.
It's like how to get around the EEOC's Four-Fifths Rule for sniffing out disparate impact in firemen hiring tests, cities like Chicago have made the test so easy that 96% of whites pass. The racial gap stays around one standard deviation, but the racial gap as measured in percentiles declines because most people pass.
This isn't the whole story in NYC, but it's a big, big part of the story.
Finally, may I just reiterate that a much more feasible and pragmatic goal than the current goal of trying to raise Non-Asian Minority performance by one standard deviation while preventing whites and Asians from improving is to try to raise all four groups' performance by half a standard deviation?