Cheating on CRCT signals test's high stakes

When the results came back, students from four of the five schools posted modest gains, on a par with others around Georgia. But for Deerwood students, scores surged. The Governor's Office of Student Achievement last week released preliminary conclusions from an investigation revealing someone erased incorrect answers and penciled in correct responses.

The cheating at Deerwood and three other public elementary schools in Georgia, documented in the state report, underscores the desperation of some teachers and administrators to avoid the embarrassment of failing to meet the requirements of the decade's overarching education reform effort, No Child Left Behind. In each case, scores increased enough to reverse the schools' first-ever failures to make "adequate yearly progress" toward student achievement goals.

More surprising than the cheating, perhaps, is the brazenness with which it was carried out, interviews and state records suggest.

At Atherton Elementary in DeKalb County, for instance, 32 fifth-graders retook the math portion of the CRCT. All 32 passed, 26 of them with scores in the top tier. In fact, after the summer retest, every single Atherton fifth-grader had passed the math exam. In other grades at Atherton, no more than two-thirds of students met or exceeded the test's standards.

State officials have not determined who is responsible for the alleged cheating at Deerwood, Atherton, Parklane in Fulton County and Burroughs-Molette in Glynn County. Any school employee found to have participated in a cheating scheme could be fired and could lose his or her state certification as a teacher or school administrator.

Atherton's principal, James Berry, resigned Thursday and DeKalb school officials transferred an assistant principal after preliminary results from the state's investigation were released. The other school districts say they are reviewing the state's findings. But how thoroughly the districts intend to pursue the matter remains uncertain.

At first, a spokeswoman for the Atlanta Public Schools said it would conduct its own investigation. By late Friday, however, the system had decided to defer to the state Professional Standards Commission, which polices educator credentials.

"We'll take appropriate disciplinary action if it is warranted," said Su Yeager, a spokeswoman for the Atlanta schools. "We will look to find out exactly what happened in that classroom."

The state identified 11 Deerwood tests as possibly being tampered with.

Asked whether the district is investigating Deerwood's principal, Lisa Smith, Yeager answered, "Absolutely not."

The state investigation began last December after The Atlanta Journal-Constitution disclosed improbably steep gains in CRCT retest scores at a handful of schools. Officials commissioned a sophisticated study of test papers, widely known as "bubble sheets" for the small ovals where students mark answers with No. 2 pencils. The analysis of the four schools' papers showed that both the number of erasures and the number of answers changed from wrong to right far exceeded any statistical likelihood.

Another indicator of cheating: A large proportion of students from the four schools had more erasures on their answer sheets than 98 to 99 percent of all students who took the test, documents show. All 32 Atherton fifth-graders fell into the top percentile of changes, as did four in five Parklane students. The state's research showed that no more than 1 in 20 students should make enough erasures to reach that level.

"Data like that cannot possibly exist in a valid testing environment," said Kathleen Mathers, executive director of the student achievement office. "It is statistically impossible."

The investigation's findings suggest that whoever was behind the cheating didn't fear getting caught, despite their obvious tampering with test papers, Mathers said.

"If you thought you had done it in complete privacy and no one saw you do it and you didn't know such a thing as erasure analysis existed, that puts things in a different context."

'Not adequate'

Deerwood opened in 2004, the first new public school in southwestern Atlanta in decades. It quickly became a fixture of the Deerwood Park neighborhood, its students' achievements a source of community pride.

Enough students qualified for free or reduced-price meals that the federal government designated Deerwood a Title I school, making it eligible for additional funding. After making "adequate yearly progress" — the performance target based on standardized test scores and other factors, such as attendance rates — for three consecutive years, Deerwood earned an additional accolade: "Distinguished."

But several fifth-grade students failed the CRCT's math exam in the spring of 2008, causing Deerwood to lose its elite status. For the first time, they failed to make adequate progress.

Deerwood could reverse its fortunes only if the low-scoring students did better in summer school.

"A Title I Distinguished school has pride in being a Title I Distinguished school," Mathers said. When a school loses that title, she said, "I would guess it's an embarrassing situation."

At Deerwood, as at other schools named in the cheating probe, testing procedures often seemed lax, investigators found.

A recently retired educator supervised Deerwood's summer retesting program last year, Mathers said. Early in the summer term, she asked for an assistant — another retired educator who agreed to work for free.

The volunteer helped administer the CRCT retests, the investigation determined. In addition, Mathers said, investigators were told a volunteer proctor may have pulled students out of the classroom where the retest was taking place and given them the exam elsewhere in the school.

Both situations violate state regulations for CRCT security, Mathers said; only paid school employees, accountable to their supervisors, are supposed to so much as touch testing materials.

During the week the CRCT was administered, investigators found, Deerwood stored testing materials in a room that wasn't secure. No school employee was assigned to make sure the door was locked or even closed.

Test security was even looser at Burroughs-Molette Elementary in Glynn County. Test papers were not collected promptly from classrooms, investigators found, and they were left with clerical workers until a courier came to pick them up. The supervisor of the summer retesting program had neither a school administrator's license nor training for the job.

Atherton and Parklane both posted improvements in test scores that defied logic, the state determined. Like Deerwood, Parklane taught summer classes for students from several schools — but only the Parklane students made great strides on the test. Like all other DeKalb schools, Atherton used a scripted curriculum called "Ladders to Success" – to a much greater degree of success than others.

"If 'Ladders to Success' were in and of itself a truly remarkable program, [the student achievement office] would expect to see similar gains across all summer school programs in DeKalb," the state report said. "That did not happen."

'High stakes'

The importance attached to standardized testing — the CRCT is almost universally described as "high stakes" — may explain why some teachers or principals would be tempted to manipulate the scores, educators say.

"There are so many pressures from so many directions," said Tim Callahan, spokesman for the Professional Association of Georgia Educators, the state's largest teachers' group. "This, at first blush, appears to be well organized, goal-oriented and rather on a large scale."

Principals of the schools where cheating appears to have occurred did not respond to messages. Deerwood's Lisa Smith referred questions to the school district's public relations staff.

In early 2005, a few months after Deerwood opened, Smith was quoted in an AJC article about the school's meaning to its neighborhood.

"It brings hope and opportunity to families," she said. "I like to think of the school as the pulse of the community, where parents, community representatives, business partners and all other stakeholders share in the accountability of student success."