Bootstrapping to Evaluate Accuracy of Citation-Based Journal Indicators

Introduction Bibliometric indicators ranking aggregate units have a long tradition, including criticisms of methodology, interpretation and application. Despite the criticism, there is a demand for these indicators, and recent developments have led to improvements of methodology and interpretation. An essential element of these interpretations is to provide estimates of the accuracy, robustness, stability and confidence of bibliometric indicators, thereby providing the reader with data required to interpret results. This has, for example, been demonstrated for the set of indicators in the Leiden ranking (Waltman et al., 2012), the Journal Impact Factor (Chen, Jen, & Wu, 2014) and other journal indicators (Andersen, Christensen, & Schneider, 2012) as well as author metrics (Lehmann, Jackson, & Lautrup, 2008). The present study applies the same type of bootstrapping technique to estimate stability, as is used in the Leiden ranking (Waltman et al., 2012), on an array of citation-based journal indicators. The purpose of this analysis is to compare recent methodological advances, as well as traditional approaches. The study is based on clinical medicine journals in the Web of Science (WoS).