|
|
Science Watch is truly spoiled for choice, so many are the fascinating statistics about science that are only a few keystrokes away. Take the accompanying table, for example. Science Watch asked which fields of science, defined here by sets of journals, show the most concentration in influence (and possibly excellence), as measured by the percentage of papers in the field required to account for 50% of all citations to the papers in the field. The papers surveyed were articles, notes, and reviews indexed by ISI between 1981-97, and the citations to these papers span the same interval. Naturally, older papers would have had more time to collect citations than younger papers in this 17-year period, but that is true for all the fields listed. In library and information science circles, the skewed nature of citation distributions—whether of papers, journals, scientists, or research institutions—is well known. The pattern is that relatively few members of a population collect a great number of citations. This is recognized in other spheres as well, especially in matters of human choice, and the phrase "80/20" is often used as a shorthand description of this phenomenon. The reasons offered for the extreme skewness of citation distributions are many, but one obvious conclusion is that some things really are better than others, and that citations, as an indicator of many independent and generally expert judgments about utility and merit, reflect this reality. But why would this skewness be significantly different from one field to another, as the data indicate? Computer science, economics and business, and molecular biology and genetics, ranking first, second, and third, respectively, require only about 1 in 20 papers to account for half the field’s total citations. On the other hand, chemistry, plant and animal sciences, and astronomy and astrophysics, which rank 19th, 20th, and 21st, require 1 in 10 papers—that is, twice as many—to achieve the same citation take for their fields. This suggests that outstanding contributions are less evenly distributed in the former than in the latter. But the data can be viewed, as well, in a different light: that there is more citable material (or less of low utility) in the latter than in the former. To check the idea that some fields tend to publish more of lower utility than other fields, Science Watch offers data in a second column of the table, which provides the percentage of uncited papers in each field, again for the period 1981-97. Of course, papers published at the end of the period would be more uncited than those published in the beginning of the period, so the uncitedness measure is a blended rate of papers from different years. But this does not prevent comparisons across fields. Very broadly speaking, there seems to be an inverse relationship between measures of concentration of influence and uncitedness, but not invariably. Interesting exceptions on either end include molecular biology and genetics with a low rate of uncitedness (14.66%) and a high concentration of influence (6.68% of the paper population), and agricultural sciences, with a relatively high rate of uncitedness (33.54%) and a relatively low concentration of influence (9.04% of the paper population).
An even more revealing analysis would focus on the concentration of citations among individual scientists, not papers. This is not a practical undertaking currently for several technical reasons, such as the inability to distinguish different individuals whose names appear in identical form (surname and initials) in the ISI database, as well as the problem of eliminating the effects of coauthorship. Since many citation studies show that highly cited authors tend to produce multiple highly cited papers, this type of analysis would doubtless show an even more powerful concentration of influence, which would, Science Watch suggests, reflect the outsized talent of a tiny elite in science. In this context, Science Watch can cite its own study of several years ago that revealed a group of only a few hundred Howard Hughes Medical Institute investigators contributed a quarter of the world’s top cited papers in biomedicine (see Science Watch, 5[5]:7, May 1995). The table on the following page will allow readers to determine whether a
paper, published in a certain field during the 1981-97 period, qualifies as a
one-in-a-hundred, a one-in-a-thousand, or a one-in-ten-thousand contribution, in terms of
citations. Age is an advantage in meeting these thresholds: papers published in the early
1980s papers will have accumulated citations over a longer period than those of the
mid-1990s. But again, the interest is comparing the citation frequency thresholds across
fields. The difference between education and molecular biology and genetics is an order of
magnitude! These numbers should serve to caution citation analysts from using absolute
citation counts in comparing papers (or researchers) in different fields. The golden rule
is to compare like with like, and to match time and any other vital parameters. But the
table above will allow some proud authors to show that one or more of their papers stands
at or even beyond the mark of a citation classic. See |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Science
Watch®, January/February 1999, Vol. 10, No. 1 Citing URL: http://www.sciencewatch.com/jan-feb99/sw_jan-feb99_page1.htm |
Search | Jan/Feb 1999 Index | Archives | Contact | Home
|
|
|
|
|
Science
Watch® is an editorial component of Essential
Science Indicators |
|
|
|
(c) 2008 The
Thomson Corporation. |
"
.
