Lies, damned lies, and impact factors

In a commentary published four years ago in the Journal of the American Medical Association, Eugene Garfield outlined the history of the journal impact factor.

He and Irving Sher created the impact factor in the early 1960s to help determine which journals should be included in the then new Science Citation Index . Relying solely on the number of papers published in a journal, they feared, risked ignoring thin, highly selective journals, such as the Proceedings of the National Academy of Sciences.

In case you didn’t know, a journal’s impact factor for a given year is the average number of citations received by papers published in the journal during the two preceding years. Letters to the editor, editorials, book reviews, and other non-papers are excluded from the impact factor calculation.

Review papers that don’t necessarily contain new scientific knowledge yet provide useful overviews garner lots of citations. Five of the top 10 perennially highest-impact-factor journals, including the top four, are review journals.

In physics the top journal by impact factor, and the 10th highest in all of science, is Reviews of Modern Physics, which published its first issue in 1929. When I checked this afternoon, a 2004 paper by Igor Zutic, Jaroslav Fabian, and Sankar Das Sarma entitled “Spintronics: Fundamentals and Applications ” had racked up 1669 citations!

Now suppose you’re a journal editor or publisher. In these tough financial times, cash-strapped libraries use impact factors to determine which subscriptions to keep and which to cancel. How would you raise your journal’s impact factor?

Publishing fewer and better papers is one method. Or you could run more review articles. But, as a paper posted recently on arXiv describes, there’s another option: You can manipulate the impact factor by publishing your own papers that cite your own journal.

The paper’s authors are two mathematicians from the University of Minnesota, Douglas Arnold and Kristine Fowler. “Nefarious Numbers” is the title they chose for the paper. Its abstract reads as follows:

When I ask physicists how they pick which journals to publish in, “audience” is the most common answer: They want their peers to read their work. I’m not sure whether the manipulation of impact factors hurts research, but it would be a pity if libraries canceled their subscriptions to journals whose impact factors are honestly earned.