You say you want to know what the passing score is on Georgia's high-stakes reading exam for third-graders? Well, nyah-nyah, they're not going to tell you:
For weeks, the state Department of Education has refused to say how many questions Georgia's third-graders needed to answer correctly in order to pass a high-stakes reading test required for promotion to the fourth grade. Without knowing where the state set the standard, many observers say it is unclear what Georgia's surprisingly good performance on the test means.
Georgia officials were overjoyed that 91 percent of the state's 115,000 third-graders passed the proficiency exam. The state had expected as many as 26,000 students to fail the test, which was given last spring...
State Superintendent of Schools Kathy Cox says the state did not manipulate the test to improve performance. The cut score — the number of correct answers a student must have in order to pass — was the same this year as in 2002, the last time the third-grade reading test was given, she said.
Yes, but while that tells us the standard wasn't lowered from 2002, that still doesn't tell us what it is. Keeping the cutscore in place doesn't guarantee that everyone actually improved, because the items could have gotten easier.
Standardized tests are not graded the same as classroom tests, which typically have a scale of 100 with a score of 70 being the minimum required to pass. Tests like the state's curriculum exam have determined cut scores based on how much a group of educators decides students need to know in order to prove they have mastered the subject.
On Gwinnett County's standardized Gateway exam, used for promotion in grades four and seven, a passing score is less than 50 percent of correct answers. The cut score is low because the questions are difficult, said Linda Mitchell, head of testing in Gwinnett.
It's quite possible that there's no reason for the hush-hush other than to satisfy the lawyers. I'm not going to jump on the conspiracy bandwagon. But knowing the placement of the cutscore does allow for more argument about how useful the test is, so naturally some have jumped to the conclusion that the DOE wants to forstall those conversations.
Say it's revealed that the cutscore is 40% of the items. This could mean the items were very difficult, or it could mean the cutscore is so low that almost anyone can pass. Since everyone's seen the test, the public has a shot at gauging the difficulty of the test, and can use the cutscore knowledge to get an estimate of what "proficient" really means. Texas reveals their cutscore, and it isn't any sort of infringement on copyrights or test security to do so.
Posted by kswygert at June 16, 2004 04:33 PM