Extended definition
SNIP and Eigenfactor are two journal metrics that correct, by opposite routes, the central defect of the impact factor: treating every citation as equal. SNIP (Source Normalized Impact per Paper) was proposed by Moed (2010) and adopted by Scopus. It normalizes by field: it divides citations per paper by the citation potential of the field, defined from the average length of the reference lists of the works that cite the journal. Because fields that cite heavily (life sciences) generate larger raw impact than fields that cite sparingly (mathematics), SNIP places journals from different fields on a comparable scale without relying on fixed field boundaries. Eigenfactor, created by Bergstrom and West in 2007 and incorporated into the Journal Citation Reports, follows the inverse logic: it weights each citation by the prestige of the citing journal, in an iterative scheme similar to Google’s PageRank. A citation from a heavily cited journal is worth more than one from a peripheral journal. Eigenfactor uses a five-year window and removes journal self-citations; its per-article version is the Article Influence Score.
When it applies
SNIP applies when the comparison crosses fields with different citation cultures: assessing an engineering journal alongside a biochemistry one requires the normalization it provides. It is the right metric when the goal is to neutralize field bias before comparing. Eigenfactor applies when what matters is prestige-weighted influence, not a simple count: it captures the journal’s position in the citation network and resists self-citation manipulation better, since it removes it from the calculation. Rizkallah and Sin (2010), comparing impact factor, Eigenfactor, and Article Influence Score across 35 medical journals over eight years, showed that the three metrics produce distinct orderings, which supports their joint use: each answers a different question about the same journal.
When it does not apply
Neither applies as a measure of the quality of an article or a researcher: they are journal metrics, and using them to judge individual work repeats the misuse that DORA condemns in the impact factor. Eigenfactor does not apply when a per-article measure is needed: by construction it is size-dependent, and Rizkallah and Sin (2010) showed that journals publishing many articles exhibit a larger Eigenfactor simply from volume, which distorts comparison between journals of different sizes. SNIP does not apply naively to a time series: Waltman and colleagues (2013) revised the indicator precisely because the first version had unstable statistical properties, and comparing SNIP across different years or definitions without care invites error. And neither applies where citation is not the relevant signal, as in applied fields whose impact is mainly technical or clinical.
Applications by field
- Multidisciplinary evaluation: SNIP is the choice when journals from fields with different citation potentials must be compared on a single scale.
- Bibliometrics and science policy: Eigenfactor measures prestige-weighted influence in the citation network, useful for mapping the structure of a field.
- Medical and life sciences: combined use of impact factor, Eigenfactor, and Article Influence Score for a less misleading reading than any single indicator.
- Libraries and collection management: both metrics, openly available, support subscription and deselection decisions on a cost-versus-influence basis.
Common pitfalls
The first pitfall is conflating the two axes: SNIP normalizes by field, Eigenfactor weights by prestige, and treating them as synonyms loses exactly what each one corrects. The second is comparing Eigenfactor across journals of very different sizes without recalling that it grows with article volume; for per-article comparison, the correct indicator is the Article Influence Score. The third is assuming that SNIP’s normalization removes all field bias: it attenuates, it does not zero it, and it still depends on database coverage. The fourth is reading either as a measure of individual merit, an error DORA made explicit. The fifth is ignoring the source database: SNIP comes from Scopus, Eigenfactor from the Journal Citation Reports, and the same journal can hold different positions depending on the base behind the calculation.