Too often, journal impact factors serve as proxies for the influence of individual research papers. Journal impact factors express the average number of times a journal’s papers have been cited in the recent past, but they often overestimate the influence of a paper. But even if one obtains the actual number of citations, the question remains: “What does that number indicate?”
|Economics and business||10.19||9.22||8.51||7.43||7.12||5.80||4.57||3.08||1.66||0.67||0.13||5.02|
|Plant and animal science||12.59||12.04||11.48||10.41||9.25||7.86||6.52||4.62||2.79||1.20||0.22||6.90|
The data in the table above clarify matters in two ways. Different fields of investigation exhibit widely different average rates of citation, and publication year affects the score, because older papers have had more time to attract citations. The field with the highest average citation rate is molecular biology and genetics, while that with the lowest average is mathematics, according to data from Thomson Reuters’ Essential Science Indicators database, 1998-2008. Of course, field definitions matter, and in this case groups of journals serve to define subject categories; in addition, individual articles in multidisciplinary journals are assigned to specific fields. One might think that the number of papers published or the population of researchers in a field are the predominant factors that influence the average rate of citation, but it is mostly the average number of references presented in papers of the field that determines the average citation rate. Mathematics papers typically list few references, whereas those in molecular biology display extensive citations.
Time matters, too. Naturally, more recent papers have lower averages. The only reasonable use of journal impact factors as proxy measures for individual papers is when papers have not had time to collect citations; in these cases, the journal gives some indication about the probable impact of a paper because journals with high impact factors in their field typically have more stringent review standards than those with lower impact factors. Still, do-it-yourself citation analysts are urged to focus on the number of citations a paper has collected and to weigh that number by the average citation rate for papers in the same field and the same year of publication. Finally, the type of paper is relevant. The statistics above are limited to articles in journals indexed by Thomson Reuters. Other types of articles such as editorials, correction notices and meeting abstracts exhibit lower citation averages.
See Henk F. Moed, Citation Analysis in Research Evaluation, Springer, 2005, and Joint Committee on Quantitative Assessment of Research, Citation Statistics: A Report from the International Mathematical Union (IMU) in co-operation with the International Council of Industrial and Applied Mathematics (ICIAM) and the Institute of Mathematical Statistics (IMS), June 2008 (www.mathunion.org/fileadmin/IMU/Report/CitationStatistics.pdf)