Friday, November 20, 2015

Is Your Grading Biased? Beware the Halo Effect

Humans are prone to taking mental short cuts, it's part of the physiology of our brains. We are exposed to so much information throughout our lives and from day-to-day that it is necessary for our brain to utilize mental short-cuts (i.e., heuristics) to help sort information and to determine what should grab one's attention. Heuristics are mental guidelines you create from your experience to help you reach an answer more quickly than by algorithm searching through every possible answer. The only problem is that this quick thinking can also bias cognition and decision-making, and grading is no exception.

The Halo effect is one form of heuristic bias that impacts everyone's decision making. It happens when we find good qualities in an individual and, through experience, incorporate those good qualities as part of our perception of the person. This biases the way we perceive their actions, whether good or bad. For instance, attractive individuals are often perceived to be more intelligent, talented, and generally good more often than individuals of average attractiveness.

The Halo effect can produce biases in grading just like any other area of judgment. In a previous post we discussed briefly how faculty will frequently grade students based on their personal characteristics and past performance rather than solely on their performance on the work in question. Faculty may begin reading an assignment, stop and think, 'so-and-so usually does great work' and allow that judgment to alter the way they grade the assignment.

This bias in judgment and decision-making process is exactly what researchers found at work in a study examining the Halo effect in subjective student assessments. The researchers randomly assigned faculty to observe a student perform either very poorly or very well on an oral presentation, and then graded the student on a written assignment. When grading the exact same student on the exact same writing assignment, faculty members gave substantially higher grades when the students had done well in a presentation before being graded and lower grades when the students had done poorly before being graded. Other than giving either a poor or good oral performance, the student was the same and the quality of the written assignment was the same. And yet the grades were different. Prior experience with students will bias the way faculty members grade students on future assessments, but by being aware of this "Halo effect", faculty can more effectively protect against bias. Making student assessments anonymous is another effective was to reduce biased grading.

No comments:

Post a Comment