What is the RCR? How is the RCR score calculated?

The Relative Citation Ratio (RCR) is a citation-based measure of the scientific influence of a publication. It is calculated by the National Institutes of Health (NIH) in the United States as a field- and time-normalized citation rate, benchmarked to the median NIH-funded publication in the same area of research and year.

A publication's area of research is defined by the corpus of publications co-cited with it (the "co-citation network") — it is therefore dynamically defined. In other words, the RCR indicates how a publication has been cited relative to other publications in its co-citation network, and this is assumed to be reflective of the publication's area of research.

The RCR requires a PubMed record. Dimensions displays an RCR where a publication has been matched to a PubMed record (PMID) present in the NIH's iCite database, which covers publications published from 1980 onwards.

The RCR is available for research articles as well as other document types with a PubMed record, such as reviews and editorials. Preprints are excluded unless they are sourced from PubMed.

Because the RCR depends on PubMed coverage, it is concentrated in the biomedical and life sciences literature. Publications outside PubMed — including much of the engineering, social sciences and humanities literature, books and conference proceedings — do not have an RCR. An empty RCR indicates that a publication falls outside the metric's coverage or is too recent; it is not an indication of low citation impact. For publications without an RCR, the Field Citation Ratio (FCR) offers a comparable field-normalized measure across the whole Dimensions catalog.

Citation statistics need time to become meaningful, so Dimensions does not show an RCR for publications from the current or the previous calendar year. Publications from earlier years have an RCR where one is available from the NIH. One further publication year becomes available each January 1st.

The NIH flags values for recent publications as provisional, indicating that they are based on a short citation window and are expected to change substantially as citations accumulate. Values for the most recent year available in Dimensions are typically provisional and should be treated with more caution than values for older publications.

An RCR is not a fixed value at any stage. Because it compares a publication's citations to those of NIH-funded publications in the same area of research, it is recalculated with each NIH release and will change as the publication and its comparison set accrue citations. Values shown in Dimensions come from the most recently imported iCite snapshot.

RCR values are normalized so that 1.0 represents the median NIH-funded publication in the same area of research and year. A publication with an RCR of 1.0 has received the number of citations that would be expected based on this NIH norm, while a publication with an RCR of 2.0 has received twice as many citations as expected.

It is important to note that NIH-funded publications set the benchmark, not the comparison population. An RCR below 1.0 therefore does not mean a publication is below average for its field — it means it is cited less than a typical NIH-funded publication in that field. This is worth keeping in mind when analysing portfolios that are not NIH-funded.

The RCR is made available in the NIH Office of Portfolio Analysis's iCite database, with snapshots posted on NIH's figshare platform. Once available, the latest snapshots are imported into Dimensions and displayed for all applicable publications.

Because this is an externally calculated indicator based on the NIH's own citation data, values in Dimensions may differ from those obtained by calculating an RCR from Dimensions' citation counts. In a small number of cases a publication shown with no citations in Dimensions will still have an RCR, because the NIH has identified citations that Dimensions has not. Values may also differ from those shown directly in iCite, since the NIH updates its database on its own schedule and Dimensions reflects the most recently imported snapshot.

If you would like to see a detailed description of how the RCR is calculated, you can find this on the iCite support page or in the original publication. Note that the original paper benchmarked against NIH R01-funded publications, whereas the RCR as currently published by the NIH is benchmarked against all NIH-funded publications.

RCR Mean is the arithmetic mean Relative Citation Ratio (RCR), which indicates the relative citation performance of a set of publications compared to other publications in their areas of research. The values per year are the years in which the publications were published.

Because the RCR is only available for publications with a PubMed record, an RCR Mean is calculated over the subset of a result set that has an RCR, not over the full result set. For result sets with limited PubMed coverage, the mean may be based on a small proportion of the publications and should be interpreted with care.

If a publication of "average" influence is supposed to have an RCR score of 1.0, what does this mean for a score of 2 versus a score of 5, for example?

An RCR of 2 indicates the publication is roughly twice as influential (through citations) as the median NIH-funded publication in its area of research and year, and a score of 5 roughly five times as influential. However, given the complexity of establishing a true midpoint for citation distributions, these are best treated as general indicators of comparative influence rather than exact figures.

How is the RCR different from the FCR?

The RCR is only calculated for publications found within PubMed, and uses the co-citation network to define the comparison group. The FCR is calculated for all publications with an assigned Field of Research (FoR) code, and compares against other publications with the same FoR code and publication year rather than a co-citation network. The RCR is benchmarked to NIH-funded publications, while the FCR is normalized to 1.0 within the selection itself.

Why does a publication have no RCR?

The most common reasons are that the publication has no PubMed record, or that it is from the current or previous calendar year — see When does a publication get an RCR? above. Publications with no citations in the NIH's citation data also do not receive an RCR, and publications published before 1980 are outside iCite's coverage.

Why has a publication's RCR changed?

The RCR is recalculated by the NIH with each release, so values change as a publication and the publications it is compared against accrue citations. A change does not indicate an error.

Why does the RCR in Dimensions differ from the value shown in iCite?

Dimensions displays the value from the most recently imported iCite snapshot, so if the NIH has published a newer snapshot the two can differ. In addition, iCite shows RCR values for very recent publications which Dimensions does not display.