ABSTRACT
I looked at statistically large samples of references in the Astrophysical Journal and Astronomy & Astrophysics at approximately decade intervals from 1952 to 2009. They show an increase of references to journals from 76% to 90.0% at the expense of references to monographs (down to 3.4%), conference papers (down to 2.6%), in-house papers (down to 1.2%), theses (down to 0.3%), and private communications (0.0%). References to preprints (1.3%) and review papers (1.2%) have also increased. The general conclusion is that references that are available online are increasing in usage while the remaining references are decreasing.
Export citation and abstract BibTeX RIS
1. INTRODUCTION
Publication practices in astronomy have been changing during recent decades. Now authors produce manuscripts that are computer readable, which avoids typesetting, and virtually all journals are available online. These have substantially increased the speed of distribution. For instance, in 1995 The Astrophysical Journal Letters took 73 ± 5 days to reach readers after acceptance while now they take 30 ± 4 days. Similarly, the places where our papers are published have changed. While observatory publications and theses used to be frequent, it was realized that those place the entire cost of publication and distribution upon the observatories or universities whereas for journals the costs are shared with the readers. Private communications have essentially disappeared because the readers cannot verify their contents. Conference papers in astronomy usually do not contain supporting data or detailed analyses, so they are less favored than journal papers as references. While we do not doubt the importance of review papers and monographs for education and research, it is generally fairer to reference the original research papers. Thus the sources that we cite have changed. This study is a statistical analysis of those changes.
Another change has been in the development of search engines, such as the ADS (Astrophysics Data System), which allow researchers to find the papers they need, independent of where they are published. However, those have not affected our referencing results as one would expect. Abt (2009) found that the frequency of self-referencing of journals has not changed in the past 21 years and the frequency of references to journals outside the eight major astronomical ones decreased from 42% to 23%. Thus, surprisingly, the references used by authors have not expanded with the availability of search engines.
2. THE METHOD
Previously (Abt 1995) I scanned issues of the Astrophysical Journal (ApJ) in the January issues of each decadal year from 1952 to 1992 and for Astronomy & Astrophysics (A&A) from 1972 to 1992 (because that journal did not start until 1969). These are typical of the standard astronomical journals at the 9% level (Abt 2009). The 1995 study was updated to include 2006 in Abt (2007), and we now include 2009 here. It became clear that references to journal papers, preprints, and reviews were increasing in frequency while most other sources were declining. The later study was not widely available so the current study includes data from 2009 and summarizes all the results.
The method is to look at all the references in issues of those two journals and to assign them to one of the eight sources listed in Table 1. I initially scanned all of the January issues of the two journals, yielding 129 references in 1952, 264 in 1962, 1889 in 1972, 4613 in 1982, 6909 in 1992, and 14,877 in 2006. In 2009 that was impractical and I counted only the 7278 references in the first January issue of each journal. Although each journal tends to favor references to its own journal (Abt 2009), we are not separating references to individual journals here, so that is not a factor. The differences between the assignments in ApJ and A&A to the sources listed in Table 1 are within the counting errors.
|
The data for 2006 (one month) and 2009 (one issue) are listed in Table 2. We see that the references to journals increased from 2006 to 2009 and all the others decreased, but not always significantly.
|
The distributions of the fractions for each source are given in Figures 1–8. In those the standard error bars were computed from the square roots of the numbers of counts. In each case, the scatter from the fitted curves is 2–4 times those errors, so the scatter is not represented simply by those curves or lines. Other factors are occurring, such as the rapid increase in the number of conferences held and monographs published. Note that the vertical range is 100% for the journals in Figures 3 and 10% for most of the other graphs.
Fig. 1.— Fitted curve is a quadratic, a poor fit.
Fig. 3.— Data for references to journal papers suggest some curvature, but the upward slope is obvious in any case.
Fig. 8.— References to theses increased until about 1980 and have now decreased. The curve is a quadratic.
3. CONCLUSIONS
Figure 1 shows that the frequency of usage of conference papers increased rapidly in 1950–1980, probably due to the rapid increase in the number of conferences during those decades. Recently they have decreased in importance as references, perhaps because of the more rapid publication of journal papers and the availability of many of those papers as preprints.
Figure 2 shows a rapid decrease in references to in-house publications, undoubtedly due to the gradual demise of observatory publications. However, other in-house publications (catalogs, instruction manuals, computer programs, etc.), from, e.g., Goddard Space Flight Center and the Center for Astrophysics, are slowly increasing in numbers. Hence the tail after about 1990.
Fig. 2.— Curve is a fourth-order polynomial that well fits the demise in observatory publications and perhaps a slowly rising tail into the future.
Figure 3 shows a steady increase in references to journal papers that has now reached 90% of the total references. This is probably because virtually all journals are now online and available to researchers, even when they work away from any libraries.
The data in Figure 4 for monographs surprisingly shows a downward slope, even though more books in astronomy are published each year. We do not doubt that many of those monographs are useful and important for education and conducting research, but the books cited as references tend to be ones that are especially useful or broad summaries. I collected the titles of all the books that were cited in the 2006 January issues of ApJ and A&A; Table 3 lists the eight books that were cited five or more times. This table was first printed in Abt (2007).
Fig. 4.— Linear fit to the data on references to monographs is not a good fit, but it is obvious that there is a gradual decline despite increasing numbers of books published annually in astronomy during recent decades.
|
The data for preprints shown in Figure 5 could be fitted to either a quadratic that peaks around 1980 or a linear relation. Considering the popularity of preprints servers, I would expect a slow increase with time unless authors volunteer or are pressured to generally state to which journals their preprints have been submitted; then they are included with those journals.
Fig. 5.— Data for preprints could be fitted with either a quadratic that peaks around 1980 or a linear relation (shown). A linear relation is shown because of my guess that preprints will continue to be important in a field of science that is developing rapidly.
The data for private communications, shown in Figure 6, are definitely disappearing, because it has been realized that readers have no easy way to verify the claims made.
Fig. 6.— Private communications are disappearing at an approximately linear rate.
Reviews are very important in a field that is so extensive so it is difficult for astronomers to scan all of the literature; therefore the steady rise shown in Figure 7 is not surprising. This is particularly true when astronomers enter a new field of research. However, editors and referees encourage authors to give credit to the original research papers, rather than to the reviewer.
Fig. 7.— Citations to reviews or summaries of papers published in a given field show a gradual increase with time.
As shown in Figure 8, theses showed a gradual increase during 1950–1980 while the numbers of Ph.D.'s granted rose. However, universities are now shifting toward a policy of requiring acceptance by standard journals of papers on the thesis research instead of verbose documents. This is a step forward because it trains young researchers to write concise papers. This change in policy is probably the reason for the decrease shown since 1980, even though the number of Ph.D.'s granted annually still increases. Even if a thesis is written, other researchers are more likely to read and cite the subsequent journal papers.
In summary, references to journal papers, reviews, and probably preprints are increasing with time and all other sources are fading. We note that the reference types that increased are available online while most of those not online are decreasing. Placing monographs online would probably increase their usage too. However, sources such as conferences and theses have problems, such as the lack of data and verbosity, so placing them online will probably not increase the citations to them. Conference papers are interesting to read, but they lack the data and analyses needed for research. In-house publications involve financial problems of production and distribution, so they are unlikely to become popular. Private communications cannot be verified by the readers, and verbose theses are outdated.
Although we do not publish papers to collect citations, we should be concerned that our hard work is seen and used by others, such as posting them on arXiv before publication. The statistics presented here simply tell us the kinds of references that are cited by others. It is not obvious to me that the nature of astronomical papers has been changing, other than to be ever more concise. Readers are welcome to draw further implications from these statistics.