The Disaster of the Impact Factor

Abstract

Journal impact factor (IF) is a value calculated annually based on the number of times articles published in a journal are cited in two, or more, of the preceding years. At the time of its inception in 1955 (Garfield 1955), the inventor of the impact factor did not imagine that 1 day his tool would become a controversial and abusive measure, as he confessed 44 years later (Garfield 1999). The impact factor became a major detrimental factor of quality, creating huge pressures on authors, editors, stakeholders and funders. More tragically, in some countries the number of publications in journals with ‘‘high impact factors’’ condition the allocation of government funding for entire institutions (Plos Medicine Editorial 2006). Based on the assumption that IF reflects scientific quality, the impact factor produces a widespread impression of prestige and reputation, though no experimental data support this hypothesis (Brembs et al. 2013). The impact factor was originally conceived as a bibliometric assessment tool for publishers and librarians to provide helpful information for subscription and library collection purposes, but over the years, it has been used abusively to assess the quality of not only journals, universities, and institutions, but also individuals and countries to promote or to ‘‘denigrate’’. Its deficiencies and perverse effects are tangibly harmful that make one wonders why such a biased tool continues to exist in science. Apart from technical and business-oriented goals (Abbasi 2007, Seglen 1997), the journal impact factors suffer from two important flaws, often neglected by researchers, that need to be highlighted due to their intrinsic contradiction with fundamental scholarly ethics. First, if an article on the subject of impact factors (based on the arithmetic mean used in the calculation of impact factor) is submitted to one of the so-called ‘‘high impact factor’’ journals, the submitted article would be immediately rejected. The editors and/or reviewers would argue that the statistical approach is biased due to