Field-weighted citation impact (FWCI) is a Scopus-derived metric that divides a publication’s actual citation count by the citation count expected for similar documents in the same subject field, publication year and document type — a result of 1.0 marks the global average, above 1.0 marks above-average impact. Before an institution builds review, promotion or tenure (RPT) criteria around it, the underlying normalisation assumptions need scrutiny.
Field-weighted citation impact is defined by Elsevier as the ratio of citations actually received by an output to the citations that would be expected based on the average for the global scientific output of the same subject field, publication year and document type. It is calculated using Scopus data and surfaced through SciVal and Pure.
- What is field-weighted citation impact?
- How is FWCI calculated?
- FWCI vs CiteScore and the Journal Impact Factor
- Where FWCI breaks down: five assumptions to scrutinise
- Should FWCI drive review, promotion and tenure decisions?
- Frequently asked questions
What is field-weighted citation impact?
Field-weighted citation impact is a normalised, article-level citation metric built into Elsevier’s SciVal and Scopus platforms. It expresses how a specific output, author, or institution has been cited relative to a global benchmark of comparable publications, rather than in raw citation counts that inevitably favour older papers and citation-heavy fields such as biomedicine.
An FWCI of 1.48 means a document has been cited 48% more than expected for its field, year and type. An FWCI of 0.6 means it has been cited 40% less than expected. Because the benchmark is fixed at 1.0 by construction, roughly half of all outputs in any given field will sit below that line — a distributional fact that is frequently lost in institutional reporting.
How is FWCI calculated?
The field-weighted citation impact formula is simple on its face: FWCI = actual citations received ÷ expected citations for similar documents. The “expected” figure is the average citation count for all Scopus-indexed documents sharing the same Scopus subject classification (ASJC code), publication year, and document type (article, review, conference paper, and so on).
- A microbiology article published in 2023 that has received 20 citations, against a field average of 10 for similar 2023 microbiology articles, scores an FWCI of 2.0.
- A humanities article with 3 citations against a field average of 2 scores an FWCI of 1.5 — a superficially similar score built on a far smaller, more volatile citation base.
- SciVal aggregates FWCI across an author’s or institution’s full output set by summing actual citations and expected citations separately, then dividing the totals — not by averaging individual FWCI scores.
This matters: a single highly cited outlier can lift a whole portfolio’s FWCI, which is why SciVal documentation recommends reading FWCI alongside output volume and citation distribution, not as a standalone score.
FWCI vs CiteScore and the Journal Impact Factor
FWCI is often confused with journal-level metrics because all three numbers look similar — a decimal hovering near 1 to 10. They measure different things at different units of analysis, which is the first source of misapplication in policy documents.
| Metric | Unit of analysis | Field-normalised? | Source and window |
|---|---|---|---|
| Field-weighted citation impact (FWCI) | Article, author, or institution | Yes — field, year, document type | Scopus data via SciVal; typically a rolling multi-year citation count |
| CiteScore | Journal | No | Elsevier/Scopus; launched December 2016; citations in a year to the prior 3 years of documents |
| Journal Impact Factor (JIF) | Journal | No | Clarivate Journal Citation Reports; historically a 2-year citation window |
Neither CiteScore nor the JIF adjusts for subject field, so comparing a mathematics journal’s CiteScore to an oncology journal’s compares citation cultures, not quality. FWCI’s field normalisation is what DORA-aligned reformers have asked journal metrics to do and mostly do not — which is also why FWCI is sometimes waved through review committees as the “responsible” metric without further scrutiny.
Where FWCI breaks down: five assumptions to scrutinise
FWCI’s field normalisation is a genuine improvement over raw citation counts and journal-level proxies, but it inherits several assumptions that DORA-aligned institutions should test before writing it into criteria.
- Mean-based benchmarking, not percentile-based. FWCI compares an output to the field average, but citation distributions are heavily right-skewed: a small number of highly cited papers pull the mean upward, so most papers structurally score below 1.0 even when performing typically for their field. This is precisely why the Centre for Science and Technology Studies (CWTS) at Leiden University uses percentile-based indicators, such as the share of a unit’s output in the global top 10% most-cited, in its Leiden Ranking methodology rather than a mean-normalised ratio.
- Subject classification is assigned to journals, not articles. Scopus’s ASJC subject codes are largely applied at the source-title level. An interdisciplinary article published in a broad-scope journal inherits that journal’s field classification, which can misrepresent the “expected” citation benchmark for a genuinely cross-disciplinary piece of work.
- Small-sample volatility. For low-citation fields (much of the humanities, parts of engineering and mathematics) or for single articles, a difference of one or two citations can swing FWCI dramatically, because the expected-citation denominator is itself small. A score of 2.0 built on 20 citations is far more stable than one built on 2.
- Self-citation is not excluded by default. Author, institutional, and journal self-citation inflate the numerator unless a self-citation exclusion is explicitly applied — a configurable option in SciVal, but one that is easy to omit when scores are pulled into a spreadsheet for a committee.
- A single number cannot represent research quality, originality, or societal value. FWCI measures citation uptake within a fixed window; it says nothing about methodological rigour, reproducibility, data sharing, or the qualitative judgement DORA asks assessors to exercise in its place.
Should FWCI drive review, promotion and tenure decisions?
The San Francisco Declaration on Research Assessment (DORA), issued in 2012, recommends that institutions not use journal-based metrics as a surrogate for the quality of individual articles, individual researchers’ contributions, or as inputs to hiring, promotion, and funding decisions. FWCI’s article-level, field-normalised design addresses DORA’s specific objection to journal-level proxies such as the JIF — but it does not exempt FWCI from DORA’s broader principle that quantitative indicators should supplement, not replace, expert reading of the work itself.
Institutions building RPT criteria around FWCI should require committees to read the underlying subject classification applied to a candidate’s outputs, check whether self-citations are excluded, and treat single-digit-citation scores as statistically unstable rather than definitive. A candidate’s FWCI trend across a full portfolio, read alongside narrative evidence of contribution, is a materially more defensible signal than a single score cited in isolation.
As UK Research and Innovation and equivalent funders continue to align assessment frameworks with responsible-metrics principles, institutions that document how they weight FWCI against qualitative peer judgement — rather than adopting it as a pass/fail threshold — will be better positioned to defend their research administration processes to auditors, funders, and appeals panels alike.
Frequently asked questions
What is the average FWCI?
The global average FWCI is always 1.0 by mathematical construction, because the benchmark for “expected citations” is itself the average of comparable outputs. A score above 1.0 indicates above-average citation performance for that field, year, and document type; a score below 1.0 indicates below-average performance.
How do I get my field-weighted citation impact?
FWCI is retrieved through a SciVal subscription, where institutional users can search an author, publication set, or institution and view the FWCI directly on the metrics dashboard. Some institutions also surface FWCI through Pure, which synchronises the metric from Scopus on a scheduled basis where the integration is enabled.
What is field-weighted citation impact ranking?
FWCI is not itself a ranking system — it is a ratio, not a percentile or league-table position. Institutions sometimes rank authors, departments, or outputs by their FWCI scores internally, but this practice inherits all the mean-based and small-sample limitations described above and should be treated cautiously.
Is field-weighted citation impact the same as CiteScore?
No. FWCI operates at the article, author, or institution level and is field-normalised; CiteScore is a journal-level average citation rate with no field normalisation. A journal’s CiteScore says nothing about how any single article within it actually performed relative to its field.
FWCI remains one of the more defensible citation metrics precisely because it was built to correct the field-blindness of journal-level indicators. Its value depends entirely on institutions applying it the way its own documentation recommends: alongside output volume, subject classification checks, and self-citation controls — not as a solitary number standing in for expert judgement in a promotion file.