The Primary Sources section of the July/August 2005 issue of The Atlantic looks at a study of how Florida suspended more low-scoring students than high-scoring students during testing periods--even when the students committed similar school crimes.
From The Atlantic:
Since 2001, when the No Child Left Behind Act tied federal school funding to performance on annual tests for students in grades three through eight, critics have charged that the law encourages schools to boost their test scores artificially. A new study of one potential score-padding maneuver–suspending probable low scorers to prevent them from taking the test–provides grist for this argument. Researchers examined more than 40,000 disciplinary cases in Florida schools from the 1996-1997 school year (when Florida instituted its own mandatory testing) to the 1999-2000 school year. They found that when two students were suspended for involvement in the same incident, the student with the higher test score tended to have a shorter suspension. This isn't in itself surprising: high achievers are often cut some slack. But the gap was significantly wider during the period when the tests were administered, and it was wider only between students in grades being tested that year.
The full study –"Testing, Crime and Punishment," David N. Figlio, National Bureau of Economic Research, is here.
You can read all about potential score-padding maneuvers in my Reason, June 2005, How Schools Cheat story.