Self-grading more effective and accurate than peer grading

iStock_000021006935XSmallMany teachers ask students to grade their work or their peers’ work not only to save themselves time, but also because they believe self-grading and peer-grading provide students with another learning opportunity. However, a study published in the January 2006 issue of Educational Assessment concludes that self-grading can be a more effective teaching tool than peer grading. Self-grading also can be a better substitute for teacher assessment than peer grading, the researchers found.

The researchers looked at the correlation of grades by comparing self- and peer-grading with the test grades that a seventh-grade science teacher assigned to 101 students in four classes. They also measured the impact on learning by analyzing students’ performance on an unannounced second administration of the test a week after self- or peer-grading.

Tests improved dramatically

“Students who graded their peers’ tests did not gain significantly more than a control group of students who did not correct any papers but simply took the same test again,” the researchers say. “Those students who corrected their own tests improved dramatically.” “When grading others, students awarded lower grades to the best performing students than their teacher did,” they report. While lower performing students tended to inflate their own low scores, the researchers speculate that “students may be more conscientious in grading their own papers than those of a randomly selected, nameless peer, because they have more at stake in making sure their own paper gets graded accurately.”

Students help develop rubric

The study by Philip M. Sadler of the Harvard-Smithsonian Center for Astrophysics and Eddie Good of Ephraim Curtis Middle School, Subdbury, MA, was conducted in four seventh-grade science classes all taught by the same teacher. Mean scores for prior tests and quizzes in the four classes were very close. The test included nine fill-in-the-blank items, seven classification tasks, 13 questions to be matched with answers and five constructed response items. Students could refer to their notes while taking the test. Students in each class helped develop a rubric for grading. Identification numbers were substituted for students’ names so neither the students nor the teacher knew whose tests they were grading.

“The study strove to identify ideal conditions for student-grading,” the researchers say. “Students were well-trained in grading and had a rubric that codified their own and the teacher’s judgment. Students could ask questions and work together during the teacher supervised grading process; they were not rushed. For those who were peer-grading, names were hidden. Also, the teacher graded tests blind to student names, removing a potential source of bias. The accuracy that results from these preparations and precautions should be viewed as an ideal situation; it would probably be difficult to improve conditions further.”

The researchers recommend that teachers replicate these conditions in the classroom as much as possible when they employ self- and peer-grading. They suggest training, blind grading, incentives for accuracy and checks on accuracy by comparing scores with teacher grades. Self-grading could be used to introduce students to grading and then teachers could have students participate in creating rubrics. The teachers would be responsible for ensuring that the rubrics are being used correctly.

“The very high correlations between grades assigned by students and those assigned by the teacher in the study support the idea that this kind of substitution can be attained even by seventh grades,” the researchers conclude. “The Impact of Self- and Peer Grading on Student Learning” Educational Assessment Volume 11 Number 1 January 2006 Pps. 1-31

Published in ERN April 2006 Volume 19 Number 4

Leave a Reply

  • (will not be published)