Holy Family School

ATLANTA — Early on in Beverly L. Hall’s 11-year tenure as superintendent of Atlanta Public Schools, she figured that the academic gains she intended to make with the city’s mostly poor, black students would face skepticism.

“I knew the day would come when people would question, was the progress real?” she said in an interview last week.

So Dr. Hall took a risk, signing up for a trial program to track and compare urban school districts. Since then, Atlanta has made the highest gains in the program in reading and among the highest in math, making it a national model and Dr. Hall a star in the education field.

But that has not insulated her from a cheating scandal that initially threatened to engulf two-thirds of the district’s 84 schools. Even after an independent investigation recently found that the problem was much less widespread, critics have called for her resignation and attacked the investigation’s credibility.

The scandal has revived age-old questions about the ability of urban students to achieve — skepticism that Dr. Hall, who rose through the ranks of New York City school system and is now one of the longest-serving urban superintendents, has spent her career trying to dislodge.

The investigation centered on suspicions that answer forms on the state achievement test used to measure progress under the federal No Child Left Behind law had been tampered with by educators. The investigation found statistical indications of widespread cheating at 12 schools, isolated cheating at 13 schools and little to no evidence at the remaining 33 schools, but no smoking gun. It referred 109 educators for further investigation.

Throughout the crisis, Dr. Hall has responded with a cool professionalism rather than the outrage that some critics have demanded. Even as she has vowed to ferret out any dishonest educators and has removed the principals of the 12 schools, she has insisted that pervasive wrongdoing has yet to be proven.

“Hall’s own refusal to accept reality,” Jay Bookman, a columnist at The Atlanta Journal-Constitution, wrote in one of the more polite dissections of Dr. Hall, “is downright stunning.”

To Dr. Hall, there are plenty of reasons to resist the idea that the district’s progress is a mirage. During her tenure, the graduation rate has increased by 30 percentage points. In the last three years, the college scholarship money offered to Atlanta graduates has doubled. And in the urban district tracking program, where progress is measured by a gold-standard test called the National Assessment of Educational Progress, or NAEP, scores have continued to climb.

“I’m surprised that the allegations persist in the face of very convincing evidence from NAEP that the gains are real,” said Michael Casserly, the executive director of the Council of the Great City Schools, a coalition of 66 of the country’s largest school systems. The national assessment is less vulnerable to cheating because it is not self-administered by school districts.

The cheating suspicions were raised after the 2009 state achievement tests, when a state report measured erasures that changed wrong answers to correct ones, finding an unusually high number of erasures at 58 Atlanta schools. Erasure analysis is a blunt instrument that flags potential cheating but does not, on its own, prove that administrators or teachers tampered with answer forms.

After the report came out, the 2010 achievement tests were given under intense scrutiny, and the results were not good. Scores dropped districtwide, particularly at the flagged schools.

Meanwhile, a blue-ribbon commission appointed by a nonprofit education group was investigating the 58 schools. The commission, using Caveon Test Security, a firm that specializes in forensic data analysis, conducted a more nuanced erasure analysis than the state’s, taking into account factors like whether the erasures actually made a difference in whether the student passed.

John Fremer, the president of Caveon, said that if the company had conducted the initial screening, 33 of the schools would not even have been flagged. Investigators interviewed almost 300 people and reviewed 50,000 e-mails. No one confessed to cheating, and the commission found no direct evidence of it.

But local news organizations seemed unable to digest the investigation findings. WABE, the local NPR affiliate, incorrectly reported — twice — that the commission had referred all 58 schools for further investigation. On its Web site, The Atlanta Journal-Constitution first trumpeted “Cheating found at 58 public schools,” then did an about-face and accused the investigators of disregarding irregularities in hundreds of classrooms.