Frank Bass, Nicole Ziegler Dizon and Ben Feller:
States are helping public schools escape potential penalties by skirting the No Child Left Behind law’s requirement that students of all races must show annual academic progress.
With the federal government’s permission, schools aren’t counting the test scores of nearly 2 million students when they report progress by racial groups, an Associated Press computer analysis found.
Minorities – who historically haven’t fared as well as whites in testing – make up the vast majority of students whose scores are being excluded, AP found. And the numbers have been rising.
“I can’t believe that my child is going through testing just like the person sitting next to him or her and she’s not being counted,” said Angela Smith, a single mother. Her daughter, Shunta’ Winston, was among two dozen black students whose test scores weren’t broken out by race at her suburban Kansas City, Mo., high school.
To calculate a nationwide estimate, AP analyzed the 2003-04 enrollment figures the government collected – the latest on record – and applied the current racial category exemptions the states use.
Overall, AP found that about 1.9 million students – or about 1 in every 14 test scores – aren’t being counted under the law’s racial categories. Minorities are seven times as likely to have their scores excluded as whites, the analysis showed.
Less than 2 percent of white children’s scores aren’t being counted as a separate category. In contrast, Hispanics and blacks have roughly 10 percent of their scores excluded. More than one-third of Asian scores and nearly half of American Indian scores aren’t broken out, AP found.
Citizens’ Commission on Civil Rights website.
Carrie Antifinger notes that the loophole snares 33% of Wisconsin minority students.
Andrew Rotherham:
First, a reader of some of the back and forth might end up thinking that the law requires some minimum subgroup or that the feds set the subgroup size. It doesn’t, they don’t. Here are the exact AYP regulations from the Federal Register (pdf) and here is Ed Trust’s explanatory piece. It’s left up to the states although the feds approve the state plans and consequently have approved the various sizes in effect now. Now they’re trying to figure out how to clean up (pdf) some of the mess they’ve created.
This was a very poorly presented story, but it does call attention to the importance of required numbers or “cell sizes” for disaggregation in accountability systems.
There are three valid reasons why student scores may not be utilized in accountability systems. Two of these three inevitably result in minority scores being utilized at a lower rate and the third probably does.
The first is that the cell size needs to be statistically significant to be used as a basis for high stakes sanctions. That AP terms the need for a statistically significant population a “loophole” says more about the AP than the practice, IMO. It’s worth noting that in most cases the results for a group below the minimum number will be reported, it just won’t be used for accountability purposes.
The second is that very small cell sizes won’t even be reported to protect student privacy. Wisconsin’s criteria for supressing results of very small populations is available at http://www.dpi.state.wi.us/lbstat/priv_more.html.
Both of these factors inevitably affect more children from minority groups.
The third factor is that, for accountability purposes, children who have not been in a school or district for a full academic year (FAY) are not considered for accountability purposes. Their scores will be reported in the aggregate, however. In Wisconsin, the state tests focus mostly on the criteria for the previous year. For example, the math test given in the fall of 4th grade will have items designed to assess the objectives expected for 3rd grade. If a student did not attend the school from the 3rd Friday in September of the previous year forward, their scores are not counted for accountability purposes. If children from minority groups tend on average to be more mobile (don’t have data at hand to document this one way or the other), then this provision would cause their scores to be less utilized for accountability purposes. However, they would still be reported.
Much of the phenomenon reported in this story probably occurs for valid reasons. However, some states may choose to go beyond what privacy interests and statistics would suggest, and that would not serve children, the general public, and ultimately the schools well.