CiteScore, SNIP and SJR are three journal-level metrics, derived from the Scopus database, that describe a journal’s citation profile in complementary ways. CiteScore offers a straightforward average of citations per document, SNIP corrects for differences in citation behaviour between fields, and SJR weights citations by the prestige of the citing source — together giving a fuller picture than any single number.
All three respond to a recurring problem: raw citation counts are easy to misread. They differ by discipline, are inflated by self-citation and concentrate around a few highly cited papers. The metrics below each address part of that problem, but none should be treated as a measure of an individual article or author — a principle central to DORA and responsible assessment.
How the three metrics differ
| Metric | Source | Core idea | What it adds |
|---|---|---|---|
| CiteScore | Scopus | Citations per document over a multi-year window | Transparent, broad coverage, easy to interpret |
| SNIP | Scopus / CWTS | Source-normalised impact per paper | Corrects for differing citation density between fields |
| SJR | Scopus | Prestige-weighted citations (eigenvector logic) | Values citations from influential journals more highly |
CiteScore
CiteScore divides the citations a journal receives over a defined multi-year window by the documents it published in that window. Its appeal is transparency and breadth: it is calculated consistently across the whole Scopus catalogue and is simple to interpret. Because the underlying calculation is openly described and the numerator and denominator are defined the same way, the figure is reproducible and comparatively hard to dispute. Its limitation is that, like any raw average, it does not account for the fact that some fields simply cite more than others, nor for the heavy skew in how citations distribute within a single journal.
SNIP — source-normalised impact per paper
SNIP, developed at CWTS in Leiden, addresses exactly that field-difference problem. It normalises a journal’s citations against the citation potential of its subject area — how often papers in that field tend to cite at all. The effect is to make a journal in a sparsely cited field comparable to one in a densely cited field, so cross-disciplinary comparison becomes more meaningful.
SJR — Scimago Journal Rank
SJR takes a different angle, weighting citations by the prestige of the citing journal using an eigenvector approach related to the logic behind the Eigenfactor and altmetrics. A citation from a highly regarded, frequently cited journal contributes more than one from an obscure source. SJR therefore captures something closer to influence within a citation network than a simple count can.
How these metrics relate to the Impact Factor
The traditional Journal Impact Factor, drawn from a different database, is essentially a two-year average of citations per citable item — conceptually closest to CiteScore, though calculated over a shorter and differently defined window. CiteScore, SNIP and SJR were each designed to address a weakness of that older measure. CiteScore widens coverage and uses a longer window; SNIP adds the field normalisation the Impact Factor lacks; and SJR replaces flat counting with prestige weighting. Understanding these lineages helps explain why a journal can rank differently on each metric: they are not measuring the same thing, and a journal strong on raw volume may look more modest once citations are normalised by field or weighted by source. The full critique of the older measure is set out in our piece on the journal impact factor, and the broader question of how to read these indicators sensibly runs through our coverage of responsible research assessment. Treating any of them as interchangeable, or as a single league table, is the mistake responsible-assessment frameworks were written to prevent.
Why field normalisation matters
The single most consequential difference between these metrics is whether they correct for discipline. Citation cultures vary enormously: a molecular biology journal and a pure mathematics journal can have wildly different baseline citation rates simply because of how often, and how quickly, their fields cite. A raw average such as CiteScore will make the biology journal look more impactful even if both are equally central to their fields. SNIP exists to neutralise that distortion by benchmarking against each field’s citation potential, which is why it is the most defensible choice for comparing journals across disciplines. SJR addresses a different axis — not how many citations, but how prestigious their sources are.
Choosing the right metric for the question
No single metric is “best”; each answers a different question. The table below maps common questions to the most informative indicator.
| Your question | Most informative metric | Why |
|---|---|---|
| How much is this journal cited overall? | CiteScore | Transparent raw average |
| How does it compare across fields? | SNIP | Corrects for citation density |
| How prestigious are its citing sources? | SJR | Weights citations by source standing |
| Is this paper or author any good? | None | Read the work; use peer judgement |
Common misuses to avoid
The most frequent error is transferring a journal-level number onto an individual paper, researcher or grant application. Because citations within a journal are concentrated in a minority of articles, a high journal metric says almost nothing about a randomly chosen paper it contains. A second misuse is comparing across fields with an un-normalised metric, and a third is treating any of these indicators as a measure of quality rather than of citation behaviour. A fourth, subtler error is using one year’s figure as if it were stable, when journal metrics fluctuate from year to year for reasons unrelated to quality. Each is explicitly cautioned against by responsible-assessment frameworks, which recommend reporting a range of indicators with their definitions rather than a single decontextualised number.
Using metrics responsibly
These indicators describe journals, not the work inside them. The distribution of citations within any journal is highly skewed, so a journal’s metric tells you little about a specific paper. Responsible-assessment principles — set out in our coverage of responsible assessment and grounded in the critique of the journal impact factor — recommend judging research on its own merits, using metrics only as context and never as a proxy for quality. The wider toolkit, including network-weighted and attention indicators, is covered in our guide to standardised terminology and our overview of the Eigenfactor and altmetrics. Authors documenting their contributions can consult our guidance for authors.
Frequently asked questions
What does CiteScore measure?
CiteScore measures the average number of citations received per document a journal published over a defined multi-year window, calculated consistently across the Scopus database.
Why is SNIP useful for cross-field comparison?
SNIP normalises citations against the citation potential of a journal’s field, so journals in disciplines that cite sparingly are not unfairly penalised against those in heavily citing fields.
How is SJR different from CiteScore?
CiteScore treats all citations equally, whereas SJR weights them by the prestige of the citing journal, capturing influence within the citation network rather than a flat count.
Can these metrics evaluate an individual paper or researcher?
No. They are journal-level indicators. Citation distributions within journals are highly skewed, so under DORA they should not be used to judge individual articles or people.
