"I want to stress that this does NOT mean that all of these schools cheated," Kathleen Mathers, the executive director of the Governor's Office of Student Achievement, wrote in an e-mail.
Her office conducted an investigation of fifth-grade math scores on the state's Criterion-Referenced Competency Tests. The investigation -- the office's first into cheating -- focused on only the six most extreme improvements, those improvements of 3.5 standard deviations or more.
Ms. Mathers, however, told the State Board of Education last week that about 20 schools would have been scrutinized if 2 standard deviations, the more common statistical approach, had been the threshold.
"Summer school is unusual in that considerably smaller numbers of kids are tested, which makes it easier to post bigger score gains than what we'd see with larger groups of students," Ms. Mathers said in her e-mail to The Chronicle . "I don't want readers to get the impression that the schools that were not investigated cheated simply because they posted impressive gains."
Improvement of just one deviation would be significant and a great success, said Gordon Eisenman, dean of the College of Education at Augusta State University.
"Two is quite a bit as well, but three is almost out of the realm of possibility, especially in that time frame," Dr. Eisenman said.
For one student to improve by three deviations is statistically similar to winning the lottery, he said. For an entire class to make such an improvement is like someone winning the lottery multiple times.
A school making a two-deviation improvement is so significant that it's worth looking into, he said.
"I think every educator in America would like to know what they are doing so it can be duplicated," Dr. Eisenman said.
The Governor's Office of Student Achievement contracted with the Georgia Center for Assessment at the University of Georgia to further investigate the six most suspicious schools by analyzing the number of erasure marks and how often wrong answers were changed to right answers.
Through that process, insufficient evidence was found to continue investigations into Fair Street Elementary School in Gainesville and Adamsville Elementary in Atlanta. But such strong evidence was found that the State Board of Education tossed out the scores of the other four schools last week. Those schools are Deerwood Academy of Atlanta Public Schools, Atherton Elementary of DeKalb County, Parklane Elementary of Fulton County and Burroughs-Molette Elementary of Glynn County.
Walter Jones of Morris News Service contributed to this story.
Reach Greg Gelpi at (706) 828-3851 or firstname.lastname@example.org.
SCHOOLS THAT MADE VAST IMPROVEMENTS
The following schools scored improvements of 2 standard deviations or more last year on the math portion of the fifth-grade Criterion-Referenced Competency Test. Listed by system, they are:
ATLANTA: Adamsville Elementary, Deerwood Academy, Slater Elementary
BRANTLEY COUNTY: Waynesville Elementary
DEKALB COUNTY: Atherton Elementary
DOUGHERTY COUNTY: Lamar Reese School of the Arts
FLOYD COUNTY: Alto Park Elementary
FULTON COUNTY: Parklane Elementary
GAINESVILLE: Fair Street Elementary
GLYNN COUNTY: Burroughs Molette Elementary
JEFFERSON COUNTY: Louisville Academy
LOWNDES COUNTY: Clyattville Elementary, Hahira Elementary
OCONEE COUNTY: Oconee County Elementary, Colham Ferry Elementary
SAVANNAH-CHATHAM COUNTY: Gould Elementary, Ellis Elementary, Georgetown Elementary
Source: Governor's Office of Student Achievement