H-index

Extended definition

The h-index was proposed by physicist Jorge Hirsch in 2005 (PNAS) as a single metric to quantify a researcher’s scientific output by combining productivity and impact. The definition is direct: a researcher has h-index equal to $h$ if they published $h$ articles each with at least $h$ citations. A researcher with $h = 25$ has 25 articles with 25 or more citations; a researcher with $h = 5$ has 5 articles with 5 or more citations. Calculation depends on the citation database used — Google Scholar (most inclusive, often the highest), Scopus (peer-reviewed, indexed by Elsevier), Web of Science (more conservative, indexed by Clarivate). As the number of citable publications grows over a career, the h-index only increases over time, and is structurally correlated with academic age and research area.

When it applies

The h-index is appropriate for longitudinal comparisons of the same researcher across successive periods, or for comparisons among peers in the same field and career stage. It has legitimate use in rapid synthesis of productivity combined with impact, and in some institutional decisions (program admission, prize eligibility) where no single metric is used as final criterion. Modern databases (ORCID integrated with Scopus/WoS) calculate it automatically, updating as new citations come in.

When it does not apply

It is not appropriate for cross-field comparisons — citation cultures vary drastically. A particle physicist with $h = 30$ and a historian with $h = 12$ may have equivalent trajectories in rigor and relative impact. It does not capture quality of recent work, since new articles have little time to accumulate citations. It does not distinguish co-authorship contribution — the first author of an extremely cited paper and the ninth author of the same paper receive the same credit in the calculation. It does not replace direct portfolio reading. DORA (2012) and CoARA (2022) explicitly recommend that h-index not be used as a single metric in hiring, promotion, or funding decisions.

Applications by field

— Exact and biomedical sciences: common use in curriculum synthesis; reference values known by subfield. — Engineering: partial use; alternative metrics (conference citations, patents) compete in real weight. — Applied social sciences: variable use; longer citation window makes h-index structurally lower than in STEM. — Humanities: rarely used; tradition of single monographs and slow citation makes the metric uninformative.

Common pitfalls

The first pitfall is comparing h-index across fields — differences are cultural, not meritocratic. The second is treating h-index as a measure of individual article quality: the index is robust to outliers but insensitive to the top of the distribution. A researcher with 1 revolutionary paper cited 10,000 times and 4 papers cited 5 times has the same h-index ($h=4$) as someone with 4 papers of 50 citations each — qualitatively very different situations. The third is manipulation through self-citation or citation rings, less frequent than with impact factor but documented. The fourth is using Google Scholar h-index as equivalent to Scopus or WoS — differences can be factors of 1.5 to 2x. The fifth is evaluating early-career researchers by h-index — the time needed for citations to accumulate makes the number structurally low, and variants like h5 or the m-quotient ($h$ divided by career years) are fairer alternatives.