The present UTIG contribution list possesses 1326 entries (see Table A.I.1, Appendix I; and Appendix V). Of these, 272 are attributable to MSI biologists (category B1st) and 30 are attributable to UT faculty not on the UTIG staff (category F1st). The remaining 1024 are attributable to UTIG staff (categories G1st and Gnth) or students (category S1st); these last three categories will form the basic data for the remaining analysis in this section.
We present the time-history of these 1024 contributions and their 1993-1997 citations with a six-part graph (Figure V.A.1 and Table A.I.3); the three left-hand graphs count publications in three different ways - the top graph presents all 1024 publications, the middle graph presents only those 727 (or 71%) which were first-authored by UTIG staff or students, and the bottom graph presents time-history for the 'normalized' publication count.
What are 'normalized' counts? These are counts whereÐwhen papers had more than one author - we have assigned each author a fraction of an authorship, using the scheme in Table V.1. Note that the sum of counts for all the authors is always 1.0 (since 1 = 1/2+1/4+1/8...+1/2n+1/2n).
Table V.1 Counting for 'Normalized' Contributions and Citations
Count given to kth author
| No. Authors on Paper | 1st | 2nd | 3rd | 4th | 5th | 6th | |
| 1 | 1 | ||||||
| 2 | 1/2 | 1/2 | |||||
| 3 | 1/2 | 1/4 | 1/4 | ||||
| 4 | 1/2 | 1/4 | 1/8 | 1/8 | |||
| 5 | 1/2 | 1/4 | 1/8 | 1/16 | 1/16 | ||
| 6 | 1/2 | 1/4 | 1/8 | 1/16 | 1/32 | 1/32 | |
| etc. |
The utility of normalized counts is that they allow us to estimate what proportion of the effort represented by the CL can be attributed to a particular individual, such as Gary Latham, or to a group of individuals such as UTIG students, Senior Research Scientists, or co-authors not employed by University. For Gary Latham, simply counting his 59 coauthorships overrepresents effort, since many of these contributions were publications with five or more authors on which Latham was a fourth or lower-order author. However, counting only Latham's 11 first authorships doesn't recognize Latham's contributions to many projects. Several different ways to assign fractional credit have been proposed (Van Hooydonk, 1997); however, we have chosen the method in Table V.1 since it seems to gives appropriately more credit to 1st and second and less credit to last and near-last authors.
The graphs also show clearly that there was a distinct change in the rate of UTIG contributions that occurred in about 1984; The rate was approximately 23/yr prior to 1984 and has been 51/yr subsequently. Presumably this is related to the move of the UTIG laboratory from Galveston to Austin in 1982. Only part of this increase can be attributed to students; students were authors or coauthors on 5 papers/yr prior to 1984, and 14 papers/yr subsequently.
The three right-hand graphs (Figure V.A.1) summarize the time-history of citations to these 1024 contributions for the 1993-1997 period. Here the top graph presents all citations, the middle graph excludes citations where a first author cites a paper on which he/she is first author; and the bottom graph presents a normalized count with self-citations included. While these graphs clearly indicate that UTIG contributions have been steadily receiving about 750 citations/yr, it is probably not meaningful to convert this into a simple per/scientist or per/contribution rate. The difficulty is that citations have a characteristic time-history of their own, with most articles receiving about half of their citations within 1-5 years following their year of publication (Figure V.A.2). Thus, the 1993-1997 statistics do not adequately count citation activity for contributions that were published prior to 1990 or so, nor do they reflect probable future citation activity for contributions published after about 1995.
Figure V.A.2. Scientific articles receive most of their citations within a few years following their publication. The plotted data are for citations reported in the Science Citation Index between 1973 and 1996 to articles authored or coauthored by UTIG staff member Cliff Frohlich; citation statistics have not been compiled for other UTIG staff members for years prior to 1993.
What proportion of the UTIG staff and student results are attributable to students?
Table A.I.1 and Figure V.A.3 indicate that students are authors or coauthors on 261 of the 1024 UTIG contributions; thus they are involved in some way in about 25% of all the research performed at UTIG. However, they are first authors on 164 of these publications, with normalized publications comprising 120 of the 697 total normalized UTIG publications (Table A.I.2). This suggests that about one sixth of all research performed at UTIG is directly attributable to student effort. Since 1983 the proportion is even higher, or about one-fifth. Citation statistics provide nearly identical results; citations to student-authored or -coauthored papers are about one-quarter of the UTIG staff and student total, while normalized citations are one sixth of the total.
What proportion of the total are attributable to the efforts of UTIG's current senior staff?
Currently there are nine individuals at UTIG holding the title of Senior Research Scientist; as of 1 January 1998 these individuals have a mean duration of employment at UTIG of 18.7 years (range - 12.7 to 25.3 years). Altogether they are authors or coauthors on 549 contributions, first authors of 215 contributions, and normalized authors of 250 contributions (Table A.I.2). Thus, since the lab was founded in 1972 they are participants in about 54% of all the research represented by the CL, and their contributions represent about 36% of the total research effort.
What proportion of the CL is attributable to the efforts of coauthors not employed by UTIG?
Of the 1054 non-biology contributions, both the normalized contributions and first authorships indicate that about one third are attributable to UT faculty or non-UTIG authors (Table A.I.2).
Appendix II (NOTE: Appendix II has not been included with the web version of this report) presents graphical results for most Ph. D.-holding scientists currently employed at UTIG, as well as for several who have been employed at UTIG in the past (see Appendix II). For each scientist the six graphs present the same time-history information concerning publications and citations that has previously been presented for groups of individuals in Figures V.A.1 and V.A.3. For individuals note that for each graph type the vertical axes have been adjusted to the same rate scale (e. g., a maximum of 4.5/yr for all contributions; 100/yr for all citations), making it easier to compare productivity rates for individuals who have been at UTIG for different lengths of time. For various individuals, Appendix I also includes tables of the total number (Table A.I.4 and Table A.I.6) and rate/year (Table A.I.5 and Table A.I.7) of UTIG contributions and citations.
For citation rates, we emphasize that it is inappropriate to compare citation rates for scientists who have been at UTIG for considerably different lengths of time. First, as demonstrated by Figure V.A.2, publications receive the majority of their citations beginning about three years after their date of publication; second, scientists who have been at UTIG only a few years are unlikely to have very many publication on the CL. For both reasons, individuals who have been at UTIG for ten years or more are likely to have higher citation rates.
The time-history graphs for individual scientists suggest that individuals exhibit characteristic patterns in both the rate and type of publications; these patterns are undoubtedly indicative of differences in personal style as well as differences in the specific subfields of research. For example, consider the publication pattern for three UTIG scientists, each who author or coauthor several papers each year:
The individual characteristic publication patterns which are evident visually in the figures in Appendix II could be quantified with a relatively few easily calculable statistics; this might possibly be useful if one were charged with evaluating the relative productivity and styles of individual scientists, overcoming some of the shortcomings that arise from simple counts of publications. For example, consider the following six statistics:
1) Tservice - years of employment at UTIG
2) RAll - average rate of UTIG contributions/year, all types
3) fMain - fraction of all publications in mainstream journals ('other' publications excluded)
4) fArch-Proc - fraction of all archival and proceedings publications ('other' publications excluded)
5) f1st/all - fraction of all authorships that are 1st authorships ('other' publications excluded).
6) fNorm - ratio of normalized authorships to all authorships
These statistics clearly indicate the differences in the characteristic patterns of individuals such as Buffler, Dalziel and Goff (see Table A.I.8). With them it is also possible to calculate many other quantities that might be of interest, such as the total number of mainstream publications (TserviceRAllfMain), or the fraction of publications in 'other' journals (1 - fMain - fArch-Proc).
If individual characteristic patterns of publication are established early enough in a career, one could use statistics such as RAll to predict future productivity. This would be quite useful for making decisions about hiring and promotion. For scientists employed ten or more years at UTIG, we compare publication rates for their first four full years of employment to rates for all subsequent years (Figure V.B.1). For senior scientists now at UTIG there is a moderate correlation; in most cases individuals have contributed at comparable rates during the initial and subsequent phases of their careers; where there are exceptions the difference can be attributed to changes in their responsibilities. However, for scientists no longer at UTIG there is little correlation. In some cases these individuals contributed at a reasonable rate in their first four years; then subsequently their productivity decreased markedly. These data suggest that current publication rates are not a reliable predictor of future productivity.
Figure V.B.1. For scientists employed at UTIG for 10 or more years, annual rates of contibutions for first four years of employment vs. rates for subsequent years.
Octagons are scientists currently employed at UTIG; triangles are scientists who were formerly employed at UTIG. Top graph is for first authorships; bottom graph is normalized authorships. In both graphs only mainstream, archival, and proceedings publications are included.
Where are UTIG contributions published?
Of the 1024 contributions from UTIG staff and students (G1st, S1st, and Gnth; see Table A.I.1), 43% are in mainstream journals, 37% are archival publications or technical reports, 12% are proceedings, maps, or cross sections, and 8% are 'other' or 'comment' publications (Table A.I.10). The journals having the most contributions are Journal of Geophysical Research (100 contributions) and Geology (53 contributions); three others mainstream journals all have about 25 contributions - AAPG Bulletin, Geophysics, and Geophysical Research Letters. Altogether, contributions in these five journals constitute 23% of the total for UTIG staff and students, and 53% of all the mainstream contributions.
Of the 502 archival publications from UTIG staff and students, Initial Reports, DSDP has 36 contributions, Proceedings, ODP, Initial Reports and Scientific Results have 38, and Proceedings, Lunar and Planetary Science Conference has 22; these constitute 9% of all the UTIG staff and student contributions, and 25% of all the archival publications. The 'other' journal with the most contributions is EOS, with 15.
Not surprisingly, articles in mainstream journals were far more likely to receive citations than those in archival or 'other' journals (Table A.I.10). Indeed, the 438 mainstream articles in the CL received 2923 citations between 1993 and 1997, for an annual rate of 6.67 citations/article. In contrast, the 381 archival publications received 1.85 citations/article, 127 proceedings publications received 0.40 citations/article, while the 78 'Other' publications received 0.96 citations/article/year. Even this is misleading, since 59% of all the 'other' citations were to two articles (#940 and #1047). Only one archival publication (#902 by Lawver et al.) received more than 30 citations; no proceedings publication received even as many as 10.
|
|
|
|
|
|
|
|
|
|