How Cheating on High-Stakes Testing Works

Time to read
3 minutes
Read so far

How Cheating on High-Stakes Testing Works

November 03, 2008 - 16:47

The New York Times ran a story over the weekend about the sad case of MiShawna Moore, the former principal of a public school in South Carolina. Moore had been widely praised as a miracle worker for her leadership in turning around students at a low-performing, largely minority school in Charleston but is now facing possible criminal charges for cheating on standardized tests.

As test scores rocketed at her school, Sanders-Clyde Elementary, the city held her up as a model. The United Way and the Rotary Club honored her, The Charleston Post and Courier called her a “miracle worker,” and the state singled out her school to compete for a national award. In Washington, the Department of Education gave the school $25,000 for its achievements.

Somehow, Ms. Moore had transformed one of Charleston’s worst schools into one of its best, a rare breakthrough in a city where the state has deemed more than half the schools unsatisfactory. It seemed almost too good to be true.

It may have been. The state has recently started a criminal investigation into test scores at Ms. Moore’s school, seeking to determine whether a high number of erasure marks on the tests indicates fraud.

In the brilliant, best-selling book Freakonomics by Stephen J. Dubner and Steven D. Levitt, Dubner tells the story of how Levitt devised a computer algorithm to catch cheaters on standardized tests and then applied that algorithm to tests given in the Chicago Public Schools. In the academic paper which forms the basis for the chapter in Freakonomics on cheating on high-stakes testing, Rotten Apples: An Investigation of The Prevalence and Predictors of Teacher Cheating, published in MIT's Quarterly Journal of Economics, Levitt, an economics professor at the University of Chicago and Brian A. Jacob of the Kennedy School of Government at Harvard University concluded that "serious cases of teacher or administrator cheating on standardized tests occur in a minimum of 4–5 percent of elementary school classrooms annually".

Levitt and Jacob's study found it did not take much in the way of incentives to push teachers and administrators over the edge.

The observed frequency of cheating appears to respond strongly to relatively minor changes in incentives. Our results highlight the fact that high-powered incentive systems, especially those with bright line rules, may induce unexpected behavioral distortions such as cheating.

Under the No Child Left Behind Act of 2002, states have 12 years to bring children up to academic proficiency or lose federal funding.

Boosting test scores does not require significant effort. In the Chicago Public Schools study, Levitt found that teachers cheated by filling in correct answers on just six questions. They were caught because they filled in the same six answers on every test (bad teachers and administrators are also lazy teachers and administrators).

Critics of high-stakes testing say that examples of cheating around the country show that such testing does not work. In 2005, The Christian Science Monitor noted that new regulations have had the worst impact on minority schools many of which are considered "low performing;" under pressure to get their scores up, these schools were the first to dump traditional curriculum and do test prep almost exclusively.

In the South Carolina case, the Times reports that District officials are now realizing they should have asked tougher questions at the seemingly wonderful and unexpected results.

The school’s turnaround after Ms. Moore took over in 2003 seemed miraculous. Under South Carolina guidelines, most of the students that year were not ready for the next grade; the state deemed the school’s progress unsatisfactory. Then things started to change. Many students made spectacular gains — leaps that in retrospect seem unlikely.

“You don’t go from nonreader to proficient reader over the course of a year,” said Janet Rose, a Charleston school official.

By 2007, 96 percent of third graders taking a South Carolina test at Sanders-Clyde met the state standard in English, compared with an average of 78.3 percent at other city schools.

Doubts turned to anger when students from Moore's school struggled when they went to other schools.

Ms. Kusmider was dumbfounded to find her son’s friend, a student at the school, having great difficulty reading. “I said, ‘What’s going on? You’re under MiShawna Moore,’ ” she said. “I was very angry.”

Another parent, Tanika Bausley, recalled, “It was hard for me to believe the scores that my daughter had, knowing the struggles she was having,” adding that her child had a “borderline learning disability.”

Moore quit and moved to another state after auditors administered tests to her students in a supervised setting. The test results that followed only further fueled speculation and spurred the criminal investigation by South Carolina Law Enforcement Division into allegations of fraud.

This year, after the tests were closely monitored, the scores plummeted. Suddenly, 44.4 percent of third graders taking the state science test met the state standard, compared with 84.6 percent in 2007.

On the Freakonomics blog, Levitt notes that catching cheaters on tests using data is not new and tells the story of how a cheating scandal in Chicago involving sanitation workers back in the 1960's.


Isaac Young Teacher Disputes Criticism, Defends Fellow Teachers

Martin Daly, Teacher's Union Chief, Lashes Out at Talk of the Sound

Isaac Young Board Meeting, Failure Every Which Way But Loose

Isaac Young Math Department Chairman Struggles with Basic Calculations