How do referees integrate evaluation criteria into their overall judgment? Evidence from grant peer review

TL;DR

This study investigates how peer reviewers use evaluation criteria to form overall judgments, comparing complex and simple integration styles, and finds evidence supporting a more complex, nuanced approach.

Contribution

It introduces a revised uniform style model showing referees use many criteria and complex rules, challenging previous assumptions of simple heuristics.

Findings

Logistic regression predicted judgments with high accuracy.

Referees tend to use many criteria and complex integration rules.

The uniform style needs revision to account for complex judgment processes.

Abstract

Little is known whether peer reviewers use the same evaluation criteria and how they integrate the criteria into their overall judgment. This study therefore proposed two assessment styles based on theoretical perspectives and normative positions. According to the case-by-case style, referees use many and different criteria, weight criteria on a case-by-case basis, and integrate criteria in a complex, non-mechanical way into their overall judgment. According to the uniform style, referees use a small fraction of the available criteria, apply the same criteria, weight the criteria in the same way, and integrate the criteria based on simple rules (i.e., fast-and-frugal heuristics). These two styles were examined using a unique dataset from a career funding scheme that contained a comparatively large number of evaluation criteria. A heuristic (fast-and-frugal trees) and a complex procedure (logistic regression) were employed to describe how referees integrate the criteria into their overall judgment. The logistic regression predicted the referees' overall assessment with high accuracy and slightly more accurately than the fast-and-frugal trees. Overall, the results of this study support the uniform style but also indicate that the uniform style needs to be revised as follows: referees use many criteria and integrate the criteria using complex rules. However, and most importantly, the revised style could describe most - but not all - of the referees' judgments. Future studies should therefore examine how referees' judgments can be characterized in those cases where the uniform style failed. Moreover, the evaluation process of referees should be studied in more empirical and theoretical detail.