Analysis of UTIG Publications and Citations

IV.A. Definition of Citation and Basic Procedure

IV.B. Citations Statistics; Excluding Self-Citations

IV.C. Problems

IV. Obtaining Citations from the Science Citation Index

IV.A. Definition of Citation and Basic Procedure

In this study, a citation is a reference on one of the annual SCI CD-ROM disks to a specific article which appears in the CL. The SCI CD-ROM disks are only available at present for the years 1993 to 1997; thus, we do not include citations from the SCI bound volumes for 1992 and any earlier years. For this study citations for a specified year are for the year of the SCI CD-ROM disk on which they appear; e. g., if the 1993 disk lists an article cited in a journal with a 1992 publication date, this will be counted in 1993.

However, in this study we have excluded citations to certain articles in the UTIG CL. In particular, because we are specifically concerned with the performance of the geophysics laboratory which is now UTIG, we have not obtained citation statistics for the 178 articles authored by the biological scientists at MSI Port Aransas (author category B1st). Similarly, although we have obtained citation statistics for 30 articles in the CL which are authored by DGS faculty without coauthorship by UTIG staff (author category F1st), in the summary graphs and tables these are excluded unless noted otherwise.

The programs that accompany the SCI CD-ROMs search for citations using character strings that represent the author, year, journal, volume, and first page of the cited article. For example, suppose we are interested in finding citations to the Gary V. Latham et al. 1975 paper which begins on page 299 of a bound collection entitled Cosmic Chemistry, Moon and Planets (#44). The string [LATHAM-G-1975-*] would find all citations to G. Latham papers in 1975; while [LATHAM-G*-1975-*-P299] would expand the search to include both G. and G. V. Latham, but limit it to articles beginning on page 299. To alleviate problems with mispellings, the string [LAT*M-G*-1975-*-P299] would find all the above citations as well as citations from the Russian literature which often spell Latham's name as Latem. One must use the wild-card character '*' with care; e. g., in the 1997 SCI CD-ROM the last string finds citations both to the targeted paper in the CL and also to a paper by a G. P. Latham which appeared on page 299 of the Journal of Applied Psychology. In any case, one can initiate these searches either by typing in the search string by hand, or from a list of up to 50 search strings in a file on an appropriately formatted floppy disk.

For this study we prepared files of appropriate search strings for articles in the CL in groups of 40 contributions. Then, we executed these files for each annual SCI CD-ROM. While so doing, for each individual search string we inspected the authors, coauthors, and titles of all citing articles to determine the self-citation counts (see below) and to insure that wild cards hadn't caused us to find citations to articles not in the CL.

IV.B. Citations Statistics; Excluding Self-Citations

For each year we counted citations to each article in the CL in three ways; a count of total citations, and two counts of citations with self-citations excluded. Since the principal reason for counting citations is to measure the influence that a scientific paper, investigator, or group exerts on academic research, the justification for excluding self-citations is that they could be construed to represent self-promotion more than influence. However, while total citations is simply the number of citations on the SCI CD-ROM found as described above, there is more than one possible way to define a self citation.

For example, to illustrate the ambiguity in definition of self citation, suppose that in 1990, Winken, Blinken, and Nod coauthor a paper in Journal of Soporific Research. Suppose further that in 1994 W, B & N's paper is cited in five other papers:

1) Sandman, Journal of Sleep and Dreams

2) Lincoln and Blinken, Woolgathering Letters

3) Blinken and Lincoln, Journal of Irreproducible Results

4) Nod and Winken, Sheepcounters' Monthly

5) Winken, Somnalescence Illustrated

Clearly, in 1992 the Winken, Blinken & Nod paper receives:

A: 5 citations total

B: 4 citations in papers which the first cited author is not the first citing author (paper 5 doesn't qualify - Winken is 1st cited and 1st citing author)

C: 3 citations in papers where the citing coauthors don't include the first cited author (papers 4 & 5 don't qualify - Winken is 1st cited author and a citing coauthor)

D: 2 citations in which the first citing author is not among the cited coauthors (papers 3, 4 & 5 don't qualify - the first citing authors, Blinken, Nod, and Winken, respectively, are cited coauthors)

E: 1 citation in which no citing coauthor is among the cited coauthors (papers 2, 3, 4 & 5 don't qualify - a cited coauthorÐBlinken, Nod, or Winken - is among the citing coauthors).

While there are clearly 5 total citations, what is the appropriate number of citations with self-citations excluded? In this study we obtained counting statistics as in A, B and E (bold letters above), but not as in C and D.

• The justification for counting as in B is that this number is what has generally been counted by others in the past. This is partly because it does exclude the most blatant form of self-promotion, but more importantly because it was the most easily obtainable statistic considering the format of the bound volumes of SCI. That is, in the bound volumes if you look up 'Winken' you find only cited articles where he is the first author; under each such article the citing articles are referenced only by their first authorsÐit isn't specified that Winken was a coauthor on 4) or that Blinken and Nod were coauthors on the cited article.

• The justification for counting as in E is that it excludes the a more general form of self-promotion, where small groups of coauthors regularly cite their joint work.

• The justification for not obtaining the counts as in C and D is simply that it is a lot of trouble, even using the SCI CD-ROMs; moreover, we felt that the statistics A, B and E together presented a sufficiently comprehensive picture.

IV.C. Problems

The search for citations as described above undoubted leads to some errors. First, we undoubtedly miss some citations. These include citations to articles that are in press, citations where the author's name is misspelled, and citations to journal articles which are cited with an incorrect page or volume number. To minimize some of these problems:

• We routinely used wild card characters to avoid having middle-name initials interfere with author name; e. g., we searched on DALZIEL-I* instead of DALZIEL-IW or DALZIEL-IWD;

• We sometimes used wild card characters to avoid misspellings of certain names which are routinely misspelled; e. g., for Gail Christesen we searched on CHRIST*N instead of CHRISTESEN to so as to find references where her name was spelled Christensen or Christeson.

Second, there is always the danger of finding extra citations, especially when there is liberal use of wild-card characters in the search string. For example, a search in the 1994 SCI on the string FROHLICH-C-1987-J-GEOP*-R-V92-* uncovers six citations to papers by C. Frohlich in volume 92 of the Journal of Geophysical Research. Of these, three were to papers by a co-writer of this report, Cliff Frohlich. The remaining three were authored by one Claus Frohlich of the World Radiation Center in Switzerland; this C. Frohlich turns out to be a curly-haired, small-but-robust German man who looks remarkably like one would expect Santa Claus himself to have looked in mid-life, and who is a world-recognized authority on solar luminosity (Figure IV.C.1).

To avoid blunders of this kind in our search for UTIG citations we included page numbers, when available, in all search strings, and also checked the titles of all citing papers we found. The checking of titles is important, especially since a considerable fraction (about one quarter) of contributions are in special collections, etc., rather than journals, and thus there is no volume number or standard SCI format for the title.

Finally, we miss some citations because SCI doesn't search for citations in all journals. Some of the excluded journals, e. g., Seismological Research Letters and Antarctic Journal of the U. S., Review, are important journals in scientific fields which historically have been strongly represented at UTIG.