When peer reviewers go rogue - Estimated prevalence of citation manipulation by reviewers based on the citation patterns of 69,000 reviewers

There is anecdotal information available that some reviewers attempt to increase their citation counts by using the peer review process, adding references to reviewed publications. There have been studies around citation coercion from the perspective of journal editing and boosting of journal indicators. This study builds further on that work, with a different angle: by measuring excessive citation manipulation at the level of reviewers. In order to assess the extent of this behaviour, access to a large pool of peer-review records is required: connections between authors and reviewers, connections between reviewers and reviewed work and so forth. This study explores this area in two phases. In phase one we detect the overall patterns of citations from reviewed material to reviewers, and assign a value to the proportion of citations originating from reviewed work. The second phase further explores the citation patterns, by taking the outliers from phase one and identifying the citations that have been added during the review process. We find in the results that highly suspicious cases of this behaviour can be successfully detected and that the scale of suspicion of clear misconduct behaviour is relatively limited (0.79%).