Citation analysis refers to a group of related bibliometric techniques that examine scientific and scholarly literature using citation data.
Although acquiring citation data can be done by hand, modern citation analysis usually acquires data through citation indexes such as Web of Science, Scopus, Google Scholar, and more recently, Microsoft Academic Search. Citation analysis at its most basic can involve simply generating a list of the most cited authors, documents, or journals over a certain period of time. A variety of metrics have developed using citations to measure the impact of a particular author or work in a field. Some of these uses of citation data have generated considerable criticism (Schoonbaert and Roelants, 1996; Van Raan, 2005). As pointed out by Harzing (2010), caution must be used in the interpretation of citation data, and one cannot assume that because a work has been infrequently cited that it is unimportant or of low quality.
Citation analysis can also include the networks created by citation patterns. In direct citation analysis, the pattern of authors, documents, and journals that directly cite one another are examined. This method is particularly good for looking at the chain of influences in a field over a number of years. Here is a very simple direct citation chart created by Microsoft Academic Search, showing the top 16 works that cite Mileti’s Disasters by Design (1999):
In EDM a recent example of direct citation analysis can be found in Comfort, Waugh, and Cigler’s (2012) examination of public administration and EDM.
In 1973, Henry Small, a colleague of Eugene Garfield (Garfield and Small are considered the fathers of modern citation analysis) introduced a new citation relationship: co-citation (Small, 1973). Co-citation looks for pairs of cited items in the reference lists of source articles. Unlike direct citation analysis, co-citation does not require that the co-cited pairs (whether they be authors, documents, or journals) cite each other. Here is a simplified diagram to help explain co-citation:
Because all of the pairs are equally cited by all three source articles, the references A, B, C, and D are likely similar to one another in some conceptual respect, such as subject matter, methodology, school of thought, etc. The higher the co-citation frequency of a pair becomes, the closer the likely similarity. It does not matter that the articles do not cite each other: they share an intellectual link. Small recognized the implications of co-citation for mapping scientific disciplines and their change over time. Small’s original form of co-citation analysis used cited articles as the co-cited pairs, and it became known as Document Co-citation Analysis (DCA). In 1981, White and Griffith showed that Author Co-citation Analysis (ACA) also revealed the intellectual structure of disciplines. Later, Journal Co-citation Analysis (JCA) emerged. Use of co-citation principles has also expanded into the analysis of co-occurring terms and keywords in source documents.
From early on in citation analysis, the mapping of direct citation and co-citation networks was an integral part of the technique. Over time this would evolve into what is now known as knowledge domain visualization (KDViz).
NEXT: KDViz AND CITESPACE II