BiomedicalPublications

Impact Factor vs h-index vs CiteScore: what they actually measure

Three numbers get quoted as though they were the same kind of evidence. They are not. Each counts something different, each breaks in a different way, and none of them tells you what a single article is worth.

The short version

All three are properties of a journal or an author. None is a property of your paper.

Impact Factor: an average that hides its own shape

The Journal Impact Factor divides citations received this year by items published in the previous two years. It is calculated by Clarivate from the Web of Science database and published in Journal Citation Reports. It is proprietary: you cannot recompute it, and journals not selected for Web of Science do not have one at all.

The structural problem is that citation distributions are extremely skewed. A minority of articles collect most of the citations, so the mean sits far above what a typical article in that journal receives. Quoting a journal's Impact Factor as a proxy for an individual article is quoting an average that most of the articles do not achieve.

When it misleads: comparing across fields, where citation cultures differ enormously; judging any single paper; and comparing a review-heavy journal against a research-only one, since reviews attract disproportionate citations. The two-year window also penalises fields where work is cited slowly.

We do not publish Impact Factors. They are a licensed commercial product, and reproducing them would mean either paying for redistribution rights or misrepresenting a different number as though it were one. The metrics we show come from OpenAlex, which is open data.

h-index: durable, but it only goes up

A journal has an h-index of 40 if 40 of its items have each been cited at least 40 times. It was proposed for authors and later applied to journals. Its appeal is that it resists single outliers: one wildly cited paper cannot lift it, because you need h papers at that level.

Its weakness is the mirror image. The h-index is cumulative and monotonic — it can only rise. An old journal that has declined badly keeps the h-index it accumulated decades ago, and a genuinely excellent new journal cannot have a high one, because it has not existed long enough to have h papers cited h times.

Across the 11,447 journals in our directory the median h-index is 26. That number is mostly a statement about age and volume.

When it misleads: comparing journals of different ages or sizes; treating it as current, when it is a lifetime total; and comparing across fields, since it scales with how many people are publishing and citing.

CiteScore: broader, more transparent, still an average

CiteScore is Elsevier's metric, computed from Scopus over a four-year window rather than two. The longer window smooths year-to-year noise and treats slower-citing fields more fairly, and the methodology is published, so it is more transparent than the Impact Factor. Scopus also covers more journals than Web of Science, so more titles have one.

But it is computed by the largest commercial publisher in the field, over that publisher's own database, and it is still a mean over a skewed distribution — so the central objection to the Impact Factor applies unchanged.

When it misleads: same failure as the Impact Factor for judging individual articles; and the denominator counts document types differently, so a journal's editorial mix moves the number independently of quality.

What none of them measure

Citations measure attention, not correctness. Work is cited for being wrong, for being fashionable, for being a convenient methods reference, and for being written by someone whose field cites heavily. Retracted papers accumulate citations for years after retraction.

None of the three tells you whether a journal reviewed the paper carefully, whether it will still exist in ten years, or whether it charges you $12,690 for the privilege. Those are separate questions, and they are the ones we try to answer.

This is also the position taken by the San Francisco Declaration on Research Assessment, which asks assessors not to use journal-level metrics as a proxy for the quality of individual articles.

Why we show h-index and mean citedness anyway

Two reasons. They are open data, recomputable by anyone from OpenAlex, so we are not asking you to trust a number you cannot check. And they are useful for a narrow purpose: establishing that a journal has a real, sustained publication record rather than being three months old with a stock-photo editorial board.

Used that way — as evidence of existence and activity, not as a quality verdict — they earn their place. On every journal page we label them explicitly as OpenAlex citation statistics and state that they are not Journal Impact Factors.

What to use instead

For deciding where to send a paper, the questions that actually change the outcome are: is it in scope, is it indexed where your field and your assessors look, what does it cost, how long will it take, and is it legitimate.