Journal Impact Factors (JIF) is a system conceived by Eugene Garfield in 1955 as a tool for evaluating journals independent of their size. Originally the purpose of the tool was to assess which periodicals were most relevant for indexing in the ISI Science Citation Index (SCI). JIF are used increasingly as an indicator of the academic importance of journals – for example when choosing a journal for submission of your manuscript.
JIF are calculated for an individual journal by ascertaining the sum of the number of citations in a given year of articles/reviews/letters/notes published in the journal in the two preceding years and dividing this sum by the number of articles/reviews in those years. For example:
JIF are published annually in the Journal Citation Reports, which are part of Thomson Reuters' Web of Knowledge. Thus, the data used to calculate JIF are defined by Thomson Reuters' Web of Science, and the citation count only includes references from articles etc. indexed in the database.
Self-citations and JIF
Some journals receive a JIF even though the journal itself does not contribute citations in the field in question. In other words, the JIF of some journals are calculated, including self-citations (when a journal's articles cite its own articles), while the JIF of other journals are calculated excluding self-citations.
Read more on coverage of Thomson Reuters Web of Science.
JIF has been under constant scrutiny and there are a multitude of articles that present the pros and cons and the applicability of JIF – both with regard to the value of the indicators used to assess the importance of a journal, and also with regard to the increasing tendency to use indicators to assess the impact of researchers and their institutions (see a short assessment in Garfield (1999)). Some critics point out, for example, how easy it is to manipulate the indicators, while others focus more on the drawbacks of the actual calculation basis.
Differences in reference practices make it difficult to compare JIF
Some research fields use many references in their publications while others use few. For example, history articles contain on average 48 references, while library and information science contain about 15 (Glänzel and Schoepflin, 1999, p. 41). Furthermore, literature is used for varying lengths of time within different domains. This means that within certain subject areas, it can be a long time before publications receive their first citations, while literature in other research areas quickly becomes obsolete. Glänzel and Schoepflin (1999) show, for example, that medicine uses literature that is on average 7-8 years old, whereas history uses literature that is on average 39-40 years old.
The number of references per journal article, together with the speed at which new publications are recorded as references, influences the average number of expected citations per publication, in that publications within certain subject areas will receive many citations more quickly than publications within other fields.
JIF and database coverage
As there is a large discrepancy in how well Web of Science covers the different research fields, and since publishing practice in individual research fields varies considerably (publication types, use of references), there are major differences in the validity of JIF for assessing a journal's academic importance. As a rule, JIF can be used more easily in less dry fields and JIF ought not to be compared across different research areas.
As a student and academic at a Danish university you may have access to Journal Citation Reports and Web of Knowledge via the university library if it has a licence. Contact your library for more information.
Eugene Garfield (1999). Journal impact factor: A brief review (editorial). In: Canadian Medical Association Journal (CMAJ), Vol. 161, Iss. 8, pp. 979-980. http://www.cmaj.ca/content/161/8/979
Wolfgang Glänzel and Urs Schoepflin (1999). A bibliometric study of reference literature in the sciences and social sciences. In: Information Processing & Management, Vol. 35, Iss. 1, pp. 31-44. http://dx.doi.org/10.1016/S0306-4573(98)00028-4