Beverly Hall could hardly contain her delight.
The Atlanta school superintendent had just received her district’s scores on the 2009 Criterion-Referenced Competency Test, Georgia’s most important examination for students in grades 1 through 8. The results were preliminary, unverified and subject to change. But Hall didn’t wait to share them.
“This year’s CRCT scores,” she wrote in an e-mail to a district supporter on May 21, 2009, “look better than ever.”
A year later, with the 2009 scores under suspicion, Hall waited anxiously for new results.
“Please let me know as soon as they come back,” she wrote to the district’s research director. But after seeing a precipitous drop in the 2010 scores, Hall and her aides tried to keep them under wraps as long as possible — from the public and even from members of the Board of Education.
E-mails, memos and other documents recently obtained by The Atlanta Journal-Constitution provide the most detailed look to date at the intensity with which Hall followed the yearly CRCT results. She parsed scores in detailed exchanges with district researchers. She praised subordinates whose numbers improved. She kept business leaders and other supporters apprised of successes.
Nothing in the e-mails and other material suggests that Hall, Atlanta’s superintendent since 1999, ordered anyone to tamper with test papers or to behave illegally or unethically to achieve certain outcomes. How the district achieved its impressive 2009 scores is at the center of two criminal investigations.
In a statement Friday, Hall said she was proud of the 2009 results. Each year’s scores, she said, reflected “steady and genuine” gains by Atlanta students.
“The 2009 CRCT results showed more of the progress we have made over a decade, and we were proud of that,” Hall said Friday. “Those results are now under a cloud of suspicion, but they were not at that time. And since I do not tolerate cheating, motivate educators to cheat or cheat students out of the opportunity for academic help, I certainly wouldn’t have known in 2009, when I applauded those results, that they were suspect or compromised.”
The e-mails from 2010, however, betray an element of anxiety about the inquiries into test scores, while also showing that as recently as last spring, Hall expected the controversy over test-tampering to quickly run its course.
On May 3, 2010, Hall e-mailed a former Atlanta school board member as she waited on airplane to leave Denver. The district’s internal investigation of the 2009 tests was under way, following articles in the AJC that reported unbelievably large increases in scores. But the previous day, the American Educational Research Association had honored Hall for her “data-driven” approach.
“I received a very prestigious award,” Hall wrote. “The rest of the country are [sic] not buying into the AJC representation of widespread cheating.”
She added, though: “I can’t wait to get it resolved.”
‘Champagne!!!!!’
Kathy Augustine, the district’s deputy superintendent for instruction, was the first to share the good news with Hall in 2009.
In an e-mail from her BlackBerry late that May 14, Augustine gave Hall a summary of what she called raw files from the CRCT: Scores had increased or held steady on 26 of 31 test sections. And through statistical rounding, she said, at least three of the five declines could end up looking like increases.
“I think,” Augustine wrote to Hall, following a smiley-face emoticon, “we need to drink champagne!!!!!”
The 2009 scores were coming in just as Hall’s prominence in education-reform circles was peaking. The previous year, she was named the top administrator of a U.S. urban school district. In February 2009, she won the highest honor for any public educator: national superintendent of the year.
Hall’s discussions with top aides about the 2009 scores reflect her continual attempt to link the story of the district’s turnaround, as measured by test results, with her personal narrative of the transformative leader of a long-troubled school system. She focused on good news and said little about suggestions of trouble.
A week after Augustine summarized the test scores, Hall seemed to be growing impatient. She wrote to Lester McKee, then the district’s research director, “When can I anticipate receiving the preliminary CRCT scores?”
Six minutes later, McKee replied: “The data is ready and will be ‘hand-delivered’ to you shortly. As you know, the data is preliminary and should be embargoed until the [final] state data is available.”
The same day, Hall was sending an e-mail describing the results as “better than ever,” interpreting them as across-the-board good news.
McKee pulled back that evening, slightly. Lower scores in some areas, he said, had to be acknowledged.
“Clearly, the decreases do not suggest the need for panic or major adjustments,” McKee wrote. “However, we celebrate small victories of 1 or 2 percentage points each year, so the ‘devil in the details’ cuts both ways.”
“Nonetheless,” he added, “the story of improvement stands and the exceeds [expectations] story line should be our resulting headline.”
Hall apparently did not respond.
‘A storm coming’
As the next school year opened, the story line began to diverge.
Articles in the AJC and a state investigation questioned gains on CRCTs administered at some Atlanta schools during the summer term of 2008. Hall was frustrated.
“Every time we show success,” she said in August 2009, “we have to prove it 300 percent.”
Greater doubts arose that October, when the AJC reported on statistically improbable gains on the 2009 CRCT at several Atlanta schools. State officials then ordered an analysis of changes on test answer sheets; the results, released in February 2010, showed suspicious numbers of wrong-to-right erasures in 58 Atlanta elementary and middle schools, far more than in any other Georgia district.
Under close observation by outside monitors to prevent cheating, students took the 2010 CRCT in late April. The results seemed to confirm the irregularities suspected in the previous year’s test: In the schools suspected of tampering, the percentage of students who passed in 2010 fell by an average of 11 points.
The decline left Hall trying to minimize the metric she had so long promoted. The AJC requested the scores last May under the state open records law. But in contrast to her willingness to share the results the previous year, Hall was reluctant to turn them over.
“The state will release the official data soon,” she wrote to the district’s press spokesman. “What we have is preliminary. We need ... to check if we should release these.”
The district gave the newspaper the preliminary scores in June, along with a statement from Hall touting a decade of “steady progress.”
In the meantime, however, school officials resisted giving the information to anyone. One frustrated school board member, Brenda Muhammad, fired off an e-mail to other board members, a copy of which was forwarded to Hall.
“Had I not inquired about when the CRCT scores would be received from the state, I would not have even known that they were back,” Muhammad wrote. “When were we supposed to be informed? ... The interest in the results of the current CRCT scores is a highly anticipated, sensitive and public matter. Surely, the board should be aware of what even the ‘preliminary’ findings are.”
LaChandra Butler Burks, then the board’s chairwoman and an ally of Hall’s, wrote back that the superintendent probably would not make a formal presentation of the scores to the board before July, after district officials verified the accuracy of the results. In 2009, that presentation took place in August.
Muhammad responded: “I feel a storm coming on about these results. If we have them, it would be in our best interest to release them sooner than later.”
Burks said last week she tried to help Muhammad get the scores. “I did not deny Ms. Muhammad any information,” Burks said. “I did not have any more information than Ms. Muhammad.”
Muhammad did not respond to two requests for an interview.
Muhammad’s last e-mail received a reply from Sharron Pitts, Hall’s chief of staff.
“Right now we have the raw data,” Pitts wrote last May 24. “Believe me, we are not trying to hide anything, only staying true to the process that has been created. I know some are trying to create a linkage with the 2009 CRCT erasures issue ... but they aren’t linked and staying with our process to ensure accuracy is the best way for us to drive that message home.”
Pitts responded in a similar manner to a parent who asked the district to release CRCT scores as soon as possible, especially for schools suspected of cheating the previous year.
“It is very important that this process of verification be followed so that we all can ensure the accuracy of the information distributed,” Pitts wrote, copying the message to the superintendent.
Hall replied with a single word: “Perfect!”
‘All the credit’
With the district’s investigation nearing its completion last spring, an event was scheduled in Atlanta to divert attention from the testing scandal.
Federal education officials came to Atlanta last May to announce results of another standardized test, the National Assessment of Educational Progress. Atlanta had improved at a greater rate than other urban districts, but its scores still lagged most of its peers, as well as Georgia and the nation at large.
Still, in an exchange of e-mails surrounding the event, Michael Casserly, executive director of the Council of the Great City Schools, told Hall the NAEP scores proved her success. Hall is chairwoman-elect of the Washington-based organization.
“It makes me crazy when the story of our progress gets muddled by the kinds of things you’ve had to deal with lately,” Casserly wrote. “You deserve all the credit you can garner.”
The evening after the announcement, Hall e-mailed Casserly to say the district was receiving “very positive feedback.”
Casserly answered: “It looks like you got very good press, which was part of the goal.”
“It has been a very good day and we needed it,” Hall wrote. “The press interest nationally in the story has been impressive.”
But Hall faced additional problems: another confirmation of apparent cheating on the 2009 CRCT.
A consultant hired to study the results for the district, Andrew Porter of the University of Pennsylvania, largely confirmed the AJC’s October 2009 analysis, which found score increases that, from a statistical standpoint, were nearly impossible. In many classrooms, the odds of such a large improvement approached 1 in infinity.
Hall received Porter’s first draft in February 2010, followed by another version in the spring. When the AJC requested the document under the state open records law, the district claimed it didn’t exist. On May 25, 2010, however, Hall exchanged several e-mails about the study with Casserly.
“What did you think of Andy Porter’s report?” Casserly asked.
“I think he did a good job and it will be factored into the overall investigation,” Hall replied. “I am not sure how the AJC will view it, but at this point I guess I just want it to be over so we can move on.”
“As long as you think it was fair and accurate,” Casserly wrote.
Hall responded: “It was.”
Casserly said last week he had not seen Porter’s study at the time. He said Porter, not Hall, may have told him the report was completed.
Otherwise, the district kept the document — even its existence — secret until Nov. 11, when news organizations obtained a copy from other sources. Ten days later, Hall announced she would retire when her contract ends in June.
About the Author
