Friday, May 29, 2015

Impact factors

As chief executive editor of a journal (Geoscientific Model Development), I've learned how the "impact factor" is calculated. Basically it counts the number of citations a paper gets in the 2 years after it is published. When I first learned this I was really surprised that the timescale is so short. I think 2 years isn't really long enough for someone else to pick up your ideas and run with them, so publish something really new without many co-authors and you are likely to not score well on this metric. Nevertheless, I think that papers in GMD, which is all about publishing model descriptions, really ought to get cited - otherwise it seems no one is using the models!

In Nature this week, there are some articles for us all to take comfort from. They are about "Sleeping Beauty Papers". 

They say, 
"Scientific papers typically accrue citations steadily, peak and then decline. Those that at first lie dormant, before being discovered and enjoying a late surge, are dubbed sleeping beauties. In many cases, the awakening comes when the published research finds applications in a different field, such as when statistical methods acquire a use in biology. Some papers were ahead of their time, and described techniques that could not be exploited properly until the creation and curation of large modern databases." 

They quote famous examples, with zillions of citations. Maybe they are just weird outliers, and all they prove is that this will never happen to us - our zero citation papers are zero citation papers for ever? But we have a more normal example, right in our very family. My very own Uncle-John-In-Law published a paper in 1978:

MONTE-CARLO METHODS FOR INDEX COMPUTATION (MOD P)
By:POLLARD, JM (POLLARD, JM)
MATHEMATICS OF COMPUTATION
Volume: 32  Issue: 143  Pages: 918-924
DOI: 10.2307/2006496
Published: 1978

Here's the citation chart:




Published in 1978, there are only 4 citations in the first decade (none in the first two years, which means it contributed two zeros to the journal's impact factor), and 253 thereafter, peaking in 2010!

It is maths, so not so much the "application in a different field", as "application in any field". In this case, it is something to do with interwebs security.


2 comments:

WT said...

Although perhaps not a sleeping beauty comapared to the papers listed in the Nature article, Stommel's 1961 paper "Thermohaline Convection with Two Stable Regimes of Flow" is a good climate example.

Barely cited for 20 or so years after publication, it now has over 1000 (Google Scholar) citations.

jrkrideau said...


And in some cases, at least in psychology, it can take one to two years just to do a study, so if you publish in year 0, it may be year 2 before you can reasonably expect a lot of people to react to your work.