Exploring the value of peer assessment

Abstract

We have collected peer-assessment (PA) and self-assessment (SA) data from two resident sections of a software construction course. This course is a core requirement in a graduate program in software engineering at a large research university. While the body of research gives strong evidence that there are many benefits to implementing peer and self-assessment, concerns remain. Two concerns are that students will inflate their evaluation of themselves and that they may collude to give each other high ratings ("cronyism"). These concerns motivated this exploratory study of student bias in peer and self-assessment in a graduate engineering program. Our results confirm previous research that students tend to rate themselves higher than their peers, but we found no evidence of cronyism.