Academic Career Confidential: “How Am I Doing?”- The Perils of Personal Citation Counting
Based on the aggregate of all a journal’s articles within a given period of time, the impact factor was not designed to measure the impact of a single article or author. Still, evaluating individuals based on the impact factors of the journals in which they have published is nevertheless a common practice at institutions in the United States and abroad. Some of its pitfalls:
• One study found a skewed distribution in citation patterns: the most frequently cited 50 percent of articles accounted for 90 percent of all citations.1 Articles in the other half of the pool thus receive an unmerited boost.
• Authors in fields that are multidisciplinary or that contribute to the work of other fields have a potential advantage over those in smaller, more self-contained fields. The former have the option of publishing in journals that are more likely to be widely read and cited, while the latter do not. In a large academic research enterprise, this circumstance could exaggerate differences among groups of colleagues.
• The impact factor is recalculated every year, so both the impact factor itself and a journal’s relative ranking can fluctuate over time.
Rather than going to Journal Citation Reports to look up overall impact factors, researchers can check how often their individual articles are being cited using Web of Science, another Web of Knowledge component. (Ask an HSLS reference librarian to show you how to do a cited reference search.) But this measurement also has its flaws. As with impact factor data, inaccurate citations are common. And there is no correction for vanity practices. While a certain amount of self-citation is inevitable as authors explain the history of their work, self-citation can also be used less benignly to elevate one’s citation count.
--Pat Weiss
References:
1. Seglen P. The skewness of science. J Am Soc Information Sci 1992;43:628-638. Quoted in: Seglen PO. Why the impact factor of journals should not be used for evaluating research. BMJ 1997;314(7079):498-502.
