Impact Factor vs CiteScore vs h-index Explained

Three numbers turn up on journal websites, CVs and promotion forms. They come from different databases, count different things, and cannot stand in for one another.

Explainer · 10 min read · Directive Publications Editorial Team

Three numbers, three different questions

A journal website can show an impact factor, a CiteScore and an h-index side by side, as if they were three readings of one quantity. They are not. The Journal Impact Factor (JIF), from Clarivate's Journal Citation Reports (JCR), and CiteScore, from Elsevier's Scopus, are both citation rates, measured over different windows in different literatures. The h-index is not a rate at all: it asks how much of an output has cleared a citation threshold.

So they are not interchangeable inputs to one judgement. Where to send a clinical manuscript is a journal question; whether to appoint a candidate is not, and no journal average will answer it.

How the Journal Impact Factor is calculated

Clarivate defines the JIF as all citations in the JCR year to items published in the previous two years, divided by the number of scholarly items (articles, reviews and proceedings papers) published in those two years. Say a journal publishes 250 scholarly items across 2023 and 2024, and during 2025 that content is cited 700 times: 700 divided by 250 gives a JIF of 2.8, appearing in the JCR released in 2026.

Four details change how you should read the result.

• The fraction is not symmetrical. The numerator counts any citation to the journal, whatever item was cited, while the denominator counts only scholarly items. Citations to editorials or letters therefore add to the top without adding to the bottom.
• It is shown to one decimal place. Since the 2023 release, JCR has displayed the JIF with one decimal rather than three, to push users toward its other indicators when comparing journals.
• Retractions are handled. Since the 2025 release, citations to and from retracted content are excluded from the numerator, while retracted articles still count in the denominator.
• Coverage is set by the database. Only journals in the Web of Science Core Collection can receive a JIF. Since the 2023 release that includes journals in the Emerging Sources Citation Index and the Arts and Humanities Citation Index, which had been listed in JCR without a JIF since the 2021 release.

That first asymmetry bites hardest in medicine, where many clinical journals carry a heavy body of letters, editorials and correspondence: those can be cited without enlarging the denominator, so content mix shifts a JIF independently of the research published.

A raw JIF also means little outside its subject category, and inside medicine the categories sit far apart: oncology and cell biology titles collect citations at a different rate from nursing, rehabilitation or primary care titles. That is why JCR ranks journals within each category and reports a JIF quartile (Q1 is the top quarter by rank) and a JIF percentile. It also reports a 5-year JIF, which suits clinical fields where uptake is slow, and the Journal Citation Indicator, normalised for category, document type and publication year so the average for any category is 1.

How CiteScore is calculated

CiteScore uses a four-year window in Scopus. Elsevier's own example: CiteScore 2024 counts citations received in 2021 to 2024 by articles, reviews, conference papers, book chapters and data papers published in those same years, then divides by the number of those documents.

• The same document types sit on both sides. The five types define both numerator and denominator, so citations to editorials and letters do not raise the score.
• Annual values are fixed; the Tracker is not. CiteScore is reported once a year and does not change afterwards, while CiteScore Tracker applies the same method to the current, incomplete year and updates monthly, so a Tracker figure is provisional.
• It is not field-normalised. Scopus advises against comparing raw CiteScore values across subject fields, pointing instead to the CiteScore Percentile, which ranks a title within its own field, or to SNIP and SJR, both built for cross-field comparison: SNIP weights citations by the citation density of the subject field, and SJR weights each citation by the prestige of its source.
• It is free to check. Elsevier makes CiteScore freely available on Scopus through the Sources page, no subscription needed.

Impact factor vs CiteScore: why the numbers differ

A journal can score quite differently on the two without anything suspicious going on: it may be indexed in Scopus but not Web of Science or the reverse, and each database counts citations only from the literature it indexes. The gap is widest for clinical journals with heavy correspondence sections, since editorials and letters lift the JIF but not CiteScore.

Decision rule: when a journal quotes one metric, look it up at the source for the same year; when you compare journals, use one metric for all of them and compare within a category rather than across raw values. Two candidates for a trial report, one advertising a JIF of 4.1 and the other a CiteScore of 5.6, cannot be set against each other at all. Pull both journals' CiteScore for the same year from the Scopus Sources page and read each one's percentile in its own field: a title at the 82nd percentile of a clinical category and one at the 40th of a broad life sciences category tell you what the advertised numbers hid.

The h-index: designed for researchers

Physicist J. E. Hirsch proposed the index in PNAS in 2005 to characterise a researcher's output. A researcher has index h when h of their papers have at least h citations each. Take five papers cited 25, 9, 6, 2 and 1 times: three have at least three citations, but there are not four with at least four, so the h-index is 3. The paper cited 25 times counts the same as the one cited 6.

That design has predictable effects:

• It grows with the number of papers and the time they have had to be cited, favouring long careers and large outputs over strong individual work.
• It depends on the database. Scopus shows an author h-index, and other citation services calculate their own from different coverage, so one person can have several.
• It cannot cross disciplines, which in a medical school bites at department level: a clinician with a few practice-changing trials can score below a colleague with a long list of modest papers.

Journal versions exist. Google Scholar Metrics reports an h5-index, the h-index of a publication's articles from the last five complete calendar years, and SCImago Journal & Country Rank shows an H index based on Scopus data. The same volume effect applies, so compare journals on it only within one field and next to a per-item average such as the JIF or CiteScore.

How to spot a fake impact factor

"Impact factor" is a phrase anyone can print on a website; the Journal Impact Factor is the one Clarivate calculates in JCR. Clarivate warns that journals carrying a JIF are a more attractive target for fraud, including hijacked journals that impersonate a legitimate title using its name and ISSN. Before relying on a claimed metric, check:

1. The value is credited to Clarivate's Journal Citation Reports, not to another organisation using the words "impact factor"
2. The journal appears in the Master Journal List under the same title, ISSN and publisher as the site in front of you (searching is open; profile pages need a free account)
3. The year is possible: each release carries the previous year's citation data, so a JIF for the current calendar year does not exist yet
4. A recent value shows one decimal place, as JCR displays it; three decimals is a reason to look closer
5. A claimed CiteScore matches the Scopus Sources entry for the same ISSN and year, and says Tracker if it is provisional

A single failed check does not prove a journal is predatory; it may be quoting an old value. Ask the journal to explain, and read the result alongside the other warning signs in our guide to predatory journal red flags and the triage for research offices in How Institutions Can Spot Predatory Publishing Risks.

How institutions should use journal metrics

Clarivate states that the journal-level indicators in JCR are designed to support journal evaluation, not to assess individual researchers or articles. The San Francisco Declaration on Research Assessment (DORA) recommends against using journal-based metrics such as the JIF as a surrogate for article quality, to assess a scientist's contributions, or in hiring, promotion and funding decisions; signatories of the Agreement on Reforming Research Assessment (CoARA) make a parallel commitment covering the JIF and the h-index. The reason is partly statistical: citation distributions within journals are highly skewed, so a journal average describes the journal, not any one paper in it.

In a hospital or medical school this bites in familiar places: promotion templates asking for "impact factors of journals published in", annual reports totalling a department's impact factors, and shortlists filtered on a threshold. Each rewards venue and volume over the work, and the last discards exactly the specialty and implementation research a clinical institution needs.

Rules a research office can apply:

1. Keep journal metrics for journal questions. Use them to describe or shortlist venues, not to score a CV or a grant application. Where a metric is standing in for something else, collect that directly instead.
2. Name the source and edition. "JIF quartile in the journal's JCR category, 2026 release" can be checked. "A high-impact journal" cannot.
3. Use category position, not a fixed threshold. A rule such as "publish in journals above 3" penalises every specialty whose category tops out below it, which in a medical school usually means nursing, rehabilitation, public health and the surgical subspecialties.
4. Find out why a JIF is missing. It may only mean the journal is not in the Web of Science Core Collection, which says nothing about the soundness of its articles. A covered journal can also be suppressed from a release when an excessive concentration of citations, such as self-citation or citation stacking, distorts its JIF.
5. Record six fields whenever a value is used. The metric's exact name, the provider, the year or release, the journal's ISSN, the category rank or percentile, and the date of the lookup. A metric your office cannot describe that precisely is not yet evidence.

When authors ask which journal to target, point them to fit and legitimacy first, as set out in How to Choose the Right Journal Without Getting Scammed, and treat metrics as one input after those.

Directive Publications sets out its editorial, ethics and access policies on its For Authors page, a good place to start when you want to judge a journal on more than a single number.