Portal de Periódicos da CAPES

Trends in the Impact Factor of Scientific Journals

2006; Elsevier BV; Volume: 81; Issue: 10 Linguagem: Inglês

10.4065/81.10.1401

1942-5546

Matthew E. Falagas, George M. Zouglakis, Paraskevi A. Papastamataki,

Meta-analysis and systematic reviews

To the Editor: The impact factor has been widely used as an index of the quality of publications in scientific journals.1Garfield E Citation indexes for science: a new dimension in documentation through association of ideas.Science. 1955; 122: 108-111Crossref PubMed Scopus (1642) Google Scholar, 2Luukkonen T Bibliometrics and evaluation of research performance.Ann Med. 1990; 22: 145-150Crossref PubMed Scopus (106) Google Scholar Although several investigators have criticized various weaknesses of the impact factor,3Seglen PO Why the impact factor of journals should not be used for evaluating research.BMJ. 1997; 314: 498-502Crossref PubMed Google Scholar including the 2-year window that is used in the calculations,4Dong P Loh M Mondry A The “impact factor” revisited?.Biomed Digit Libr. 2005; 2: 7Crossref PubMed Scopus (173) Google Scholar it has not been replaced by any other means of rating the quality of journals. In addition, no data are available regarding trends in the impact factor of scientific journals as a whole that may help in relevant comparisons of journals that are partially based on the use of this index. Thus, we sought to analyze trends in the impact factor during 6 recent years (1999-2004) and compare those with relevant data from 15 to 20 years ago (1984, 1986, and 1988). We used information from the Science Edition of the Journal Citation Reports (JCR) database of the Institute for Scientific Information (ISI) to identify the impact factor of journals at various points of the distribution of the impact factor (maximum, median, minimum, and 99th, 95th, 90th, 75th, and 25th percentiles) (Table 1). A gradual increase in the impact factor of scientific journals occurred during the study period, specifically at various points of the distribution during the last 6 years of our analysis (1999-2004). In addition, the median impact factor of journals increased as much during the last 6-year period as it had during the previous 15 years (1984-1998). The percentage of journals with an impact factor higher than 1.0 was 46.6% and 49.0% in 2003 and 2004, respectively (this value increased to 51.5% in 2005). The median value (50th percentile) of the 1999 impact factors (0.76) was the impact factor of the 47th percentile of the 2004 distribution. Of note, the most prominent increase occurred with the 99th percentile of the impact factor of the journals between the study periods.TABLE 1Main Points of the Distribution of Journal Impact Factors*Journals indexed in the Science Edition of the Journal Citation Reports of the Institute for Scientific Information. From 1999 Through 2004 and 15 to 20 Years Ago†Data available only for the years 1984, 1986, and 1988.Percentiles of impact factorYearNo. of journals indexedMaximum99th95th90th75thMedian25thMinimum2004596852.4313.114.823.371.910.970.4802003590752.2812.204.603.251.820.920.4702002587654.4612.114.593.021.680.850.4302001575246.2311.984.502.901.610.810.4002000568650.3411.094.172.831.560.780.3801999555047.5610.264.152.781.510.760.3601988439848.317.873.172.171.240.520.2001986431631.627.403.182.191.070.540.2001984417729.408.023.192.181.130.510.210* Journals indexed in the Science Edition of the Journal Citation Reports of the Institute for Scientific Information.† Data available only for the years 1984, 1986, and 1988. Open table in a new tab These findings cannot be explained by the increase in the number of journals indexed by the relevant ISI database from which the calculations of the impact factor are derived. In fact, the number of journals indexed during the last 6-year study period increased by 418 journals (from 5550 in 1999 to 5968 in 2004) while it increased by 1373 journals between the beginning of the earlier study period and the beginning of the later study period (from 4177 in 1984 to 5550 in 1999). A contributing factor in the higher rate of increase in the impact factor of scientific journals during the last 6 years (1999-2004) compared to the previous study period (1984-1998) may be an increase in the mean number of references per published article during the later period. In addition, modifications in the relative frequency of publication of review articles vs original research articles (ie, review articlesare cited more frequently5Amin M Mabe M Impact factors: use and abuse.Perspect Publishing. October 2000; 1: 1-6Google Scholar) as well as the calculations of the impact factor may be other factors that contributed to the observed changes between the study periods. The availability of full text of articles on the Internet is another factor that may have influenced the trends in journal impact factors. We focused on trends in the impact factor of a subset of scientific journals, namely, those indexed in the Science Edition of JCR, a major part of which includes biomedical journals. Our analysis did not include journals from the other 2 JCR databases, Social Sciences and Art and Humanities. In addition, we did not perform comparative analyses of the trends of the distribution of the impact factor of journals in the various categories of the Science Edition of JCR. Despite these limitations, we believe that our limited analysis provides some quantitative data that may be useful to the scientific community in understanding the direction and the magnitude of the changes in the impact factors of a large proportion of scientific journals. Specifically, our analysis shows that the impact factors of a particular journal during different years cannot be compared directly without considering the overall temporal trends in the distribution of impact factors of all scientific journals.