Citation indexing began in the 1950s, and has long been dominated by the Institute for Scientific Information (ISI, which was acquired by and is now renamed Thomson Reuters), the creator and publisher of the three citation indexes available today: Science Citation Index (SCI), Social Sciences Citation Index (SSCI), and Arts & Humanities Citation Index (AHCI). This page focuses on the use of SCI in the "hard" sciences, which have the longest track record in citation studies. SCI covers more than 6,000 journals across all science and engineering disciplines.
The ISI citation indexes, back to 1975, are available online to SIOC users via the Web of Science system.
Citation -- An entry in a bibliography or footnote that refers to an earlier work of some kind. Also frequently called a "reference".
Cited Reference -- The same as a citation. An entry in a bibliography or footnote describing an earlier work.
Citing Reference -- The work in which a cited reference appears. This is what you're really searching for when you use a citation index to look forward in the literature.
For example, if you have an excellent paper on a particular topic that was published in 1992, you can use Science Citation Index (via Web of Science) to find papers published after 1992 that cited that paper. Citation implies a direct subject relationship between the papers. So, by searching for later papers citing your known paper, you can find more documents on the same or similar topic without using any keywords or subject terms.
For example, Journal A has an impact factor of 4.327, and Journal B has an impact factor of 1.045. Is Journal A "better" than Journal B? You could conceivably make that argument, if you first accept the notion that quality equates with citedness, AND if journals A and B are both in the same field, such as Analytical Chemistry or Spectroscopy. But if A is in Biochemistry, and B is in Clinical Pharmacy, no such judgment can be made. Citation behavior varies considerably from field to field. Thus, impact factors are only meaningful in context with other journals in the same field. An editor might say, "My journal has an impact factor of 3.254 this year." That means little by itself. He would rather say, "My journal ranked 2nd out of 38 journals in its subject category this year."
Impact factor can also vary based on the number and types of articles a journal publishes. Review articles tend to be more heavily cited than full papers or communications, so journals and annuals that publish mostly reviews will often have high impact factors. Journals that publish only a few articles in a given year may also have disproportionately high impact factors. Similarly, one very highly cited paper can skew a journal's impact factor significantly.
While Impact Factors are useful within certain limits, they are somewhat arbitrary and simplistic, and are subject to manipulation by editors and publishers. Lack of transparency and reproducibility has been cited as an ongoing problem with ISI's data (see readings below). Nor does a journal's IF relate in any way to the impact or quality of an individual author or paper. As a result, the use of journal impact factors in personnel and funding decisions is strongly discouraged.
Impact factors for journals covered by ISI are published annually in an electronic compilation called Journal Citation Reports. All ISI Source Journals are ranked within one or more relevant subject categories, such as CHEMISTRY, ORGANIC or SPECTROSCOPY. You can also compile customized lists. JCR also contains data on historical trends, immediacy index, cited half-life, etc.
A different journal ranking system based on citation data from Elsevier's Scopus database launched in 2007: SJR SCImago. Its methodology has yet to be thoroughly evaluated. Another project called Eigenfactor uses network theory to group journals and their citations and evaluate their relative importance.