|Office of the Chancellor / Public Affairs||
Wednesday, February 25, 2004
Washington Post 2-24-04
Burying a Test Score Scam
I was more than a little curious when I saw a Washington Post story in June 2002 that test scores at Moten Elementary School in the District of Columbia had gone through the roof.
In just one year, the proportion of Moten students scoring proficient or advanced, the two highest categories, on the Stanford 9 math test had jumped from 49 to 88 percent. On the Stanford 9 reading test the increase was even more spectacular, from 50 percent to 91.5 percent.
Moten was a typical Southeast Washington school. More than 90 percent of its 362 students were poor enough to qualify for federal subsidized lunches. It had many hard-working teachers, but in the past they had struggled to get even a majority of its students out of the Stanford 9's bottom testing category, called below basic. Until 2000, its portion of students testing proficient or advanced was like the rest of the schools in its very low-income area, less than 10 percent.
But its scores got better in 2000 and maintained that gain the next year. Then in 2002 it soared to a level never before seen at any District school with so many disadvantaged children. As my Washington Post colleague Justin Blum put it in his June 27, 2002, story "Moten's overall math score on this spring's Stanford 9 test is the second highest among the 106 elementary schools in the D.C. system, and its reading score is the sixth highest."
What was puzzling to many education experts was how little had changed at Moten to explain this sudden leap. The one-hour lunch period had been cut in half to allow more class time, and the principal had put the children in same-sex classes, but similar experiments in other parts of the country had yielded nowhere near such results.
Veteran educators and test-score followers know what comes next in this story. Blum asked school officials in 2002 if there was any indication of cheating, and they said no. I wondered what would happen to Moten's scores in 2003, with all the increased attention to that school. This Feb. 1 Blum reported the new results -- a drop in scores even more stupendous than the original increase. In 2002 only 3.3 percent of Moten students had tested below basic in math and only 2.4 percent had tested below basic in reading. In 2003 the portion of Morten children on that lowest rung of the testing ladder had ballooned to 69 percent in math and 55 percent in reading, approximately where it had been before the climb in Moten scores had first occurred.
Blum's story said: "The significant drop has drawn the attention of school administrators, who call the high results from spring 2002 inconsistent and are considering invalidating them, said William Caritj, the assistant superintendent for accountability and assessment."
"Caritj said an inquiry found no evidence of cheating on the 2002 test. But testing specialists outside the school system said the changes in scores -- coupled with observations of some former Moten teachers -- raise questions about whether cheating occurred. . . ."
"In interviews, three former teachers at Moten before 2002 said they observed a number of testing irregularities during other years when the school's scores rose. They said they suspected that someone was changing students' answer sheets. They also said some students who struggled with basic math and reading were achieving near-perfect scores."
This is, of course, an old story. We have had plenty of cases of phony test scores in the past 30 years, dating from the mid-1970s when newspapers like mine, and education policy makers in general, began reporting annual results as a measure of how well our schools were doing. To be fair, the instances of obvious cheating are still relatively rare. The vast majority of principals and testing officials continue to take their responsibilities very seriously. In America, like other countries, a lot of people will cut corners, but most of us pay our taxes, stop at red lights, remain faithful to our spouses and report test results honestly so that we can live peaceful and productive lives, and sleep well at night.
What bothers me most about Blum's Moten Elementary revelations is not the corruption of the test scores themselves. It is how little energy the D.C. school system had devoted to publicizing this little outrage and doing something about it.
Like most Americans, confirmed in every poll since we began arguing about testing a couple of decades ago, I think regular assessments of student progress like the Stanford 9 are useful parts of the educational process. I can't see any viable alternative to testing children regularly in this way, as the new No Child Left Behind law requires. I know lots of schools that have produced impressive test score increases -- although nothing like Moten's -- by significantly increasing the length of the school day and the school year and helping teachers change corrosive youthful traits like lack of respect for homework and lack of joy in the life of school.
To keep these testing programs intact and effective, it helps to make a big deal out of these rare instances of fraud. But Caritj told Blum the school system is not planning to investigate further -- or to interview school staff members who were present at the exam -- because nobody has come forward to claim that cheating occurred, at least in 2002 when the three teachers who spoke to Blum had already left the school.
George Smitherman, the Moten principal in charge when the irregularities occurred, told Blum the 2003 scores might have been lower because the interim principal who followed him did not emphasize the Stanford 9 test enough, and some disgruntled teachers "might have lost their motivation." Charlene Quander, who has been installed as Smitherman's permanent replacement, gave only the mildest rejoinder to this remarkable statement by saying teachers she spoke to thought the students had been affected by the change in administration, and that she was moving forward. The interim D.C. superintendent told Blum she was not sure whether more investigation was needed, or whether a change in procedures was needed, and would not be able to answer those questions until after reviewing what actions school officials had taken. She is stepping down soon, and it seems unlikely this unfortunate episode will be pursued.
And yet I think it should be. Blum has produced plenty of evidence of cheating, it seems to me. The three former Moten teachers told him of test score results that did not reflect their students' work in class, beginning with that first jump in scores in 2000. And one of them, Kelly Kneppner, let him use her name. She said the first 10 students on her class's alphabetical roster scored in the top percentiles on the Stanford 9 in 2000, yet her other 10 students scored poorly.
"I knew something was wrong," she said. When she tested her students in the fall of 1999, 61 percent were below basic in reading and 50 percent below basic in math. The next spring their test results showed NONE of them below basic in either subject, and the majority in the proficient or advanced categories. One child who tested at the 99th percentile in math could not subtract 3 from 5 two days before the examination.
If it is so easy to ignore such an egregious case in a big, important city, with no hearings, no organized parental protests, and no apparent efforts to prevent a recurrence, what do you think are the chances that similar hijinks in your local school district might get buried in a back file, and quickly forgotten?
I don't think this kind of thing happens often. It is hard to imagine any school administrator, no matter how craven, signing his name to such a statistical fantasy. But if such cases are routinely shelved so that we don't hear about them, how will we know for sure?
These news clips are provided by the Public Affairs Department of The California State University. They are intended for the internal use of The California State University system and should not be redistributed. Questions and submissions may be sent to firstname.lastname@example.org.