Selective Discipline Raising Scores?
Florida schools may be trying to manipulate state test results by giving longer suspensions to low-achieving students during the weeks immediately before and during testing, a study argues.
The study by David N. Figlio, an economist at the University of Florida in Gainesville, was published last month by the National Bureau of Economic Research, based in Cambridge, Mass.
Mr. Figlio used an unusual data set constructed from the administrative records of a subset of unidentified Florida districts. The data set provides demographic and test-score information on every student, as well as information on all in-school and out-of-school suspensions during school years 1996-97 through 1999-2000, the first years following the introduction of the Florida Comprehensive Assessment Test, or FCAT.
Using the data, Mr. Figlio was able to compare the disciplinary actions taken against two students suspended for the same incident and explore whether those whose prior-year test scores predicted low grades on the FCAT were likely to be suspended for more days than their higher-achieving peers. In all, he examined 41,803 incidents.
He found that, across the school year, suspensions lasted longer for low achievers than for high achievers. On average, a suspension lasted 2.35 days for students predicted to score in the lowest proficiency group in reading and mathematics on the FCAT, with 23 percent receiving suspensions of one week or longer. Other students averaged 1.91-day suspensions, with 18 percent kept out for one week or longer. During the testing window, schools generally reduced their suspension penalties for higher achievers in high-stakes grades and raised them for lower achievers, even after controlling for students’ race, family income, and sex.
Although the effect on school ratings may be modest, the economist concludes, it could make a difference for schools on the margin. He cautions that school test scores should be taken with a grain of salt, and that efforts should be made to close such loopholes.
Vol. 24, Issue 30, Page 6