Grade Language Heterogeneity in Simulation Models of Peer Review

Simulation models have proven to be valuable tools for studying peer review processes. However, the effects of some of these models’ assumptions have not been tested, nor have these models been examined in comparative contexts. In this paper, we address two of these assumptions which go in tandem: (1) on the granularity of the evaluation scale, and (2) on the homogeneity of the grade language (i.e. whether reviewers interpret evaluation grades in the same fashion). We test the consequences of these assumptions by extending a well-known agent-based model of author and reviewer behaviour with discrete evaluation scales and reviewers’ interpretation of the grade language. In this way, we compare a peer review model with a homogeneous grade language, as assumed in most models of peer review, with a more psychologically realistic model where reviewers interpret the grades of the evaluation scale heterogeneously. We find that grade language heterogeneity can indeed affect the predictions of a model of peer review.