On the fairness of using relative indicators for comparing citation performance in different disciplines
Claudio Castellano, Filippo Radicchi
Abstract
Claudio Castellano, Filippo Radicchi
Abstract
Relative indicators are commonly used to remove biases due to different citation practices in various scientific fields. Here we extend our recent investigation on the viability of the use of relative indicators for comparing article impact in different disciplines. We consider citation distributions for papers published in 14 of the 172 disciplines categorized by the Journal Citation Reports. The distribution of the number of citations received by publications in a certain discipline divided by the average number for the discipline is a universal function. Based on it, we compute the relative number of citations needed to be among the q percent most-cited publications in a discipline. The effect of finite samples is also discussed. The average number of citations is shown to be strongly correlated with the impact factor, but fluctuations are quite large. A similar universal distribution is found (with exceptions) when citation distributions restricted to papers published in a single journal are considered.
OpenAlex reports 33 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Relative indicators are commonly used to remove biases due to different citation practices in various scientific fields. Here we extend our recent investigation on the viability of the use of relative indicators for comparing article impact in different disciplines. We consider citation distributions for papers published in 14 of the 172 disciplines categorized by the Journal Citation Reports. The distribution of the number of citations received by publications in a certain discipline divided by the average number for the discipline is a universal function. Based on it, we compute the relative number of citations needed to be among the q percent most-cited publications in a discipline. The effect of finite samples is also discussed. The average number of citations is shown to be strongly correlated with the impact factor, but fluctuations are quite large. A similar universal distribution is found (with exceptions) when citation distributions restricted to papers published in a single journal are considered.
Key concepts: Citation, Impact factor, Distribution (mathematics), Function (biology), Statistics, Econometrics, Computer science, Mathematics