Skip to ContentSkip to Navigation
University of Groningen Library
University of Groningen Library Research Intelligence Services

Ask Our Experts

We receive daily questions from individual researchers and/or support staff working with or interested in research analytics. "Ask Our Experts" is a section in our quarterly newsletter aiming to capture some of the most frequent or intriguing questions we have received and our attempt to provide an answer and guidance. Hopefully, these become useful tips to our research and ENRA community (Expertise Network Research Analytics) at the UG.

Why can FWCI be misleading when analyzing publication sets?

The Field-Weighted Citation Impact is the ratio of the total citations actually received by an entity (researcher, research group, department, university, etc), and the total citations that would be expected based on the average of the subject field (see Scopus knowledgebase ). As such, it provides a "level playing field" for comparison as publications are compared to the average in their particular field, type and publication year.

However, when analyzing a larger publication set, simply calculating the average FWCI for this set can be very misleading especially when making a comparison between two such sets produced by two similar research groups. In cases like this, we recommend that you instead look at and plot the frequency distribution of the FWCIs of the publication set. For a clear example and more details, see "Profiles, not Metrics " paper published by the Web of Science Institute for Scientific Information (Jan, 2019).

The pitfall with Scival’s field-weighted citation percentiles

When conducting bibliometric analysis, we often stress the importance of field-weighted indicators as these ensure fair comparison across document type, publication year and research area.

However, a blog post posted on the Bibliomagician platform recently exposed a subtle difference in Scival when computing field-weighted citation percentiles compared to the traditional bibliometric approach.

In a nutshell, instead of using total citation counts for each publication in the dataset under analysis, Scival uses citation ratios. Moreover, Scival merges subject areas before computing the top percentiles. As noted by the blog post authors, Scival’s method doesn’t lead to very different results for large datasets. However, it certainly affects the analysis for small datasets especially when that set includes publications from more subject areas, which have very different citation distributions.

We strongly encourage anyone using Scival to carefully read the whole blog post and send us any additional questions that may arise.

Last modified:28 September 2022 12.01 p.m.