Including Peer-Assessment in Official Assessments

Peer-assessment, the evaluation of student work by their peers, was commonly used in my elementary and high school classes. At the university level, peer-assessment was a regular component of group projects and presentations in several of my courses. At each of these academic levels, the scores given by a student’s peers were included in the official assessment of that student. Peer-assessment can supplement teacher-guided instruction, and is known to have positive effects on student performance. It is increasingly used in academic settings, such as medical schools, that use problem-based learning methods of instruction. Beyond its direct benefits, peer-assessment seems to be well suited for engaging students, and may increase opportunities for reflection on one’s own knowledge. Much like peer-teaching, peer-assessment makes students personally responsible for information, a key component of active learning techniques. While I agree that peer-assessment can be a powerful tool for increasing student responsibility and interaction, recent discussions with secondary school educators have raised doubts in my mind about its appropriateness in official assessments. The validity of an official assessment is likely compromised when peer-assessments are included. The instructional objectives for a unit are rarely provided to students before completion of the unit. Objectives at the lower-cognitive levels may be more apparent (and there is evidence that students are capable of making distinctions between performance levels at lower-cognitive levels), but I would suspect that many students would fail to recognize higher-cognitive level objectives, even in instances when they demonstrated achievement in those areas. How, then, can we trust the scoring of students on their peers’ work? Without a clear understanding of the objectives for the assessment, students will apply their own interpretations of their objectives in the scoring and grading. The accuracy of peer-assessment appears to vary somewhat with student performance and external circumstances. In a 2006 study by Geoffrey Kennedy at the University of Sydney, students varied widely in their willingness to distribute a fixed amount of points within their group, regardless of the level of student participation. Some students were extremely punitive with their scoring, while others were reluctant to damage the grades of fellow students. In a study of Australian medical students, lower-performing students scored their peers higher than did faculty, while higher-performing students judged their peers’ work similarly to faculty. A 2005 study in the UK, which compared peer-assessment between subgroups of student assessors for oral presentations, found that men were more likely to score male speakers higher, while speaker gender did not impact female scoring. Although some methods have proven effective at improving accuracy of peer-assessment, these results suggest that the inclusion of peer-assessments can sometimes reduce the accuracy of an official assessment in a manner that is difficult to predict or control. My sense is that the level at which a student will judge their peers’ work is closely tied to that student’s mental state. While this is also true of the educator, the bias introduced in that case would more likely impact students across the board. This would not be true with student assessments of their peers. Should an official assessment be impacted by a student’s relationship status or daily eating patterns? Ideally, it should not. We know that a student’s mental state affects their academic performance; there is no reason to compound that phenomenon by allowing the mental state of their peers to feed into the assessment of that performance as well. The circumstances under which peer-assessments take place may also influence student scoring. Peer-assessment is often used with group projects or presentations. Students receive a group score, and then are scored by other group members on their participation. I have used this method before, thinking it would make group members more accountable to each other. However, the aforementioned study by Kennedy explored several issues with peer-assessment in this circumstance. Interestingly, some students thought that the inclusion of peer-assessment actually changed student behavior. The pressure to get a higher score led some students to co-opt group discussions, form small alliances, and control the development of their group assignment. This phenomenon may undermine one goal of group participation, which is to foster cooperation and peer-instruction. While the situation in the study was slightly unnatural, I think the results are illustrative of the types of unintended consequences that may arise from official peer-assessment. Based on these concerns, I feel that peer-assessment serves to reduce the efficacy of official assessments at evaluating student knowledge. Some obvious steps, such as anonymous feedback and pre-assessment training, can help alleviate some of the problems discussed above. Given the potential benefits of peer-assessment, both in formative evaluations of student progress and enhancement of student learning (both of which are well supported in the literature), I strongly support their inclusion in the classroom. However, their inclusion in official assessments should be weighed against the time and energy investments necessary to guard against potential problems.