Confident Scoring of Standardized Tests
Standardized test scores are often criticized for trying to measure a student’s academic achievement based on just a few hours of testing. This is a valid question, but there actually IS a way to tell how much confidence you can put in the accuracy of the scores.
Our testing service uses a score-report style that includes “confidence bands.” This is a way of showing the actual score and how sure the statistics are of its accuracy. Here’s how it works:
Find It
You’ll find it in the upper-right-hand chart on the score report. It will look like a floating black bar.
Read it
The actual score may or may not match the student’s actual achievement. The score may be inflated by good guessing (though this rarely helps significantly). Or the score may be artificially low due to distractions, fatigue, or simply from not finishing the test.
Because of these varying influences, the publishers know the student’s true achievement may be higher or lower than what he actually scored. So they use the black band to show this range. On one day a student might score higher and another day maybe lower. But on any day the student will likely always score somewhere in that range.
The Iowa Test of Basic Skills (ITBS®) confidence bands have a white diamond in them. (The diamond is not always centered.) The white diamond shows the actual National Percentile Rank (NPR) that the student scored on the day he tested. On the Stanford Achievement Test report, simply look for the N-PR to the left of the chart—it should be listed with the other scores.
Use It
The confidence band width varies from subject to subject and student to student. The smaller the band, the surer you can be that the student’s actual score matches his actual achievement in that subject. The longer the band, the less certain the student’s score matches his achievement.
If you have a long band, you’ll want to dig deeper on your score report. Check out the lower half of the report where the Content Clusters or Skills are listed. Here you’ll find out how many questions were on each subtest, how many your student attempted, and how many (out of the total available) he answered correctly.
Look for the accuracy and completion rates. If a student attempted only 5 out of 10 questions, but had 4 of them correct, his accuracy rate was good even though his score may be low. In this case, the confidence band is likely wide because there’s a good chance he could have scored higher if he had been able to finish the test.