PhysMath Central Blog
Not even wrong: a disassembly of impact factors and indices
An editorial in Europhysics News (http://dx.doi.org/10.1051/epn/2009704) caught my eye recently as it is one of the few dissections of impact factors I have read which applies some scientific and logical thinking to the process of awarding impact factors to journals or individuals.
The authors, Franck Laloë and Remy Mosseri, tackle some the assumptions made in the bibliometry game and come out with a few startling suggestions for how to maximize your own h-index.
Central to their point is the real purpose of references within scientific articles. They are not, they argue, some sort of prize list among all published articles in the relevant scientific domain. Rather, their real purpose is to give the reader information that is needed to understand the article. In context a citation could be made to; another article which describes much of the background (e.g. review articles), similar articles (to simplify writing and avoid tensions with colleagues), 'wrong' articles (in case of corrections) and articles describing instrumentation and methods.
Looking at impact factors derived from the SCI, the following problems were identified by the authors:
1 - Books are not counted towards impact factors, despite the fact that they are often the best way of making a big impact within a given field.
2 - Whether you are a lone author or a coauthor with several other people, the citation is counted as the same for each person. Why not, they ask, divide citations by the number of authors of an individual paper?
3 - Short-termism: examine the longevity of ideas by only looking at citations from 3 years or more.
4 - Data input is not homogenous over time and there are inevitable human errors in inputting/correcting it
They argue that the real reason for the success of impact factors is based on facility - saving time for in-depth evaluations of journals and, especially, individuals. They then give a tongue in cheek 'recipe for success' in which your research is geared towards improving your h-index:
1 - Work in same group of 5-6 colleagues who publisher their articles in common
2 - Don't take too much scientific risk, choose to work in larger scientific domains
3 - Don't waste time publishing books
4 - When you write articles, remember that public relations are important.
I leave the final word to the authors:
Finally, it seems that the faith in these indices is not very different from something that escapes rationality. One could compare this with astrology and numerology, which pretend to be scientific but have never gone through the process of scientific selection.
Posted by Chris Leonard at 12:10 Comments (0)