The way in which physicists and astronomers team up to write technical papers has changed over the years, and not only is it interesting to look at this behavior for its own sake, but by analyzing the data it may be possible to better understand what role, if any, does the number of authors have on the scientific impact of a paper. Likewise, such an analysis can allow physics and astronomy journals to make decisions about their publishing policies.
I was curious about the trends in the number of authors per refereed astronomy paper, so I set out to write an R script that would read in data from the NASA Astrophysics Data System, an online database of both refereed and non-refereed academic papers in astronomy and physics. The script counts the monthly number of refereed astronomy and physics papers between January 1967 and September 2013, as well as the number of authors in each of those papers, and plots the results. It was a good exercise in manipulation of regular expressions. It is possible that the script can be made more efficient by using different R structures than the ones I used, but I wanted to look at the results first.
Here’s a plot of the monthly mean and maximum number of authors:
The upward trend in both sets of data is readily apparent. The maximum is much more variable, though. The record maximum number of authors for a single paper is 3062, for May 2011, which belongs to a study of proton-proton collisions performed at the Large Hadron Collider. The variability of the mean number has increased in the last three years or so.
Let’s decompose both series into their constituent parts, namely trend, seasonal and random:
In both cases, there is very little seasonality, and the random component is more stable for the series of mean number of authors.
As it turns out, there was a nice study by Schulman et al. (1997) in which it is reported that the annual mean number of authors per astronomy paper, published in different journals, increased between the years 1975 and 1995 by a factor of between 1.5 and 2. We see something similar here. In their article, those authors showed that the fraction of papers written by a single author decreased steadily in the same period by a factor of about 3. To check that out, I looked at the monthly data of fraction of papers written by N authors, with N=1, 2, 3, 4, 5, and also N>5. This is what I found:
Single-author papers have decreased by a factor of almost 4 since 1967 (black line). The period covered by Schulman et al. shows a similar trend here, even though I’m also including physics articles. Papers written by two authors (red line) grew in quantity until the mid-1980s, and then slowly decayed to their present fraction of 20%.
Perhaps the most noteworthy behavior is that of papers written by more than five authors (orange line). From making up only 1% of the total papers in 1967, they are now the most numerous, at 28%. There was an abrupt rise in the trend of these N>5 publications in 1990, which coincides with the launch of the Hubble Space Telescope, and it hasn’t changed significantly ever since. Performing a Holt-Winters fit, which is a method that uses exponentially weighted moving averages to update estimates of the seasonally adjusted mean, the slope and seasonals, we can make a forecast of the fraction of papers that will be written by more than five authors in the next 24 months. R has functions called HoltWinters and forecast.HoltWinters that do this. The result can be seen in this figure:
The blue line represents the forecast of relative number of papers until December 2014. The dark and light shaded areas are the 85% and 95% confidence regions, respectively.
Schulman et al. (1997) have pointed out possible explanations for the decrease in the number of single-author papers and increasing numbers of multi-author articles. One is the “growth of multiwavelength astrophysics… which requires astronomers to be proficient in multiple wavebands or to collaborate with experts in other wavebands”. Another possible cause is that the increasing competition for jobs and grants forces applicants to look for collaborations that will result in a longer publication list for their CV. Also, the development and availability of rapid communication through the Internet has boosted the interactions among colleagues.
It would be interesting to look at economic indicators side by side with the data shown above, to try to elucidate if there’s a connection between number of authors and economic trends.
Reference: Schulman, E. , French, J. C., Powell, A. L., Eichhorn, G., Kurtz, M. J. and Murray, S. S. 1997, Publications of the Astronomical Society of the Pacific, 109, 1278