How to improve the outcome of performance evaluations in terms of percentiles for citation frequencies of my papers

Using empirical data I demonstrate that the result of performance evaluations by percentiles can be drastically influenced by the proper choice of the journal in which a manuscript is published. 1. Introduction In order to evaluate the impact of publications in terms of citation counts recently percentile-based bibliometric indicators have gained more and more attention (Bornmann, Leydesdorff, & Mutz, 2013). They are based on the idea of determining the position of a publication within a citation distribution of its field and its publication year (Leydesdorff & Bornmann, 2011). This position is usually expressed in terms of a quantile of the citation distribution. Commonly, percentiles are used for that purpose. These can then be utilized to attribute the evaluated papers to certain percentile rank classes. A simple and frequently used indicator is the proportion of publications which belong to the top 10% most frequently cited publications (Tijssen, Visser & Van Leeuwen, 2002, Bornmann, De Moya Anegón & Leydesdorff, 2012) in a given field for a given publication year. This would mean two percentile rank classes, namely the top 10% and the bottom 90%. Of course, different thresholds might be useful for different evaluation purposes (Lewison, Thornicroft, Szmukler & Tansella, 2007). Also more complicated indicators combining several thresholds and several percentile rank classes have been proposed (Leydesdorff, Bornmann, Mutz & Opthof 2011, Leydesdorff & Bornmann, 2012). The much appraised advantage of percentile-based indicators is that they allow a straightforward comparison of the evaluation results for publications in different fields and a suitable comparison of old and new publications, because the citation frequency of each publication of an author (or an institute, or a country, …) is compared with the complete citation distribution of publications in the corresponding field and publication year. It is well known that these reference sets can differ strongly between different fields. Therefore it appears fair to measure the impact of a publication in this way, i.e. in relation to its field. By the same token this means, however, that one can influence the position of one's publication in its citation distribution by choosing the journal and thereby the field and thus the reference set in a more or less clever way. It is the purpose of the present paper to demonstrate this effect for an empirical example, namely five of my own publications for which a choice of a different journal would have drastically improved the position of …