|Home
|Astronomy
>Research
>Publications
>Talks
>Journals
>Institutes
>Databases
>Manuals
>Pictures
>Jobs
|Search
|Manuals
|Computers
|Travel
|News
|Personal

Roelof's Citation Statistics

There are many ways to measure the productivity and impact of one's research. Because we are held accountable for our work, I have calculated some statistics of my research at the end of 2007 to compare my work to the field. There are many pitfalls in these kind of comparisons, which are described in great detail in the papers I have used to compare my impact on the field.

In the first comparison I use the method described by Pijpers (2006, Astronomy & Geophysics, 47, 6.17), where he compares the total normalized citation rate per paper (i.e., the number of citations of a paper divided by the number of authors) to the total number of papers for different authors in the field of space research. Using normalized citations in stead of regular citations corrects for the effect that working in large teams creates much more (self-)citations from all team members to each others papers. Using the total number of refereed papers corrects to some extend for the seniority of an author. In the figure below I have indicated my position in this diagram, clearly indicating that in this metric my work scores several sigma above the mean of the field.

.

A much more rigorous treatment of the different forms of citation statistics has been given by Piet van der Kruit in his analysis of the performance of the permanent staff of Dutch astronomy. This has the disadvantage of only comparing my work to Dutch astronomy, but given the international respect Dutch astronomers have, this may not be such a bad comparison at all. Most of van der Kruit's numbers date from 2004, while mine date from end 2007. While most numbers scale in obvious ways, the typical number of citations that papers receive in ADS do not. The average number of citations per paper in MNRAS has gone up by a factor ~1.3 since 2004! Whether this is due to greater completeness of ADS or the tremendous increase of the number of references per paper is not entirely clear; probably both.

Below I reproduce the three tables in Table 11 from van der Kruit's analysis, with my numbers listed at top. The explanation of the meaning of all columns can be found in van der Kruit's paper, but most of them are self-explanatory. The first table shows the regular citation rates. This shows that my total productivity is not too far of the median with 2.2 papers per year, but that my citation rate per paper puts me in the top 10% of Dutch astronomers, and my citation rate per year in the top quartile. divided
    All publications   Refereed publications
    n cites cite/pap   n cites cites/pap 1st year paper pap/year cites/year
R.S. de Jong   100 2106 21.1   30 1959 65.3 1993 2.2 139.9
mean   150 2351 14.8   79 2176 26.2 1978 3.5 91.6
dispersion   108 2173 7.2   58 2040 12.7 14 2.4 70.7
ten-percentile   37 472 7  19 423 11.8 1966 1.0 20.6
first quartile   66 1095 9.6   38 1267 17.3 1972 1.7 39.7
median   130 1684 13.8   60 2051 24.7 1980 2.8 76
third quartile   213 3253 18.7   102 3203 33.1 1989 4.8 126.4
ninety-percentile   313 5646 24   175 5195 43.8 1995 6.8 178.9

The second table shows the same numbers, but now normalized by author count, i.e. the numbers are divided by the number of authors on each paper. My numbers are even better here, reflecting that I have written many single author papers or papers with small teams. The normalized number of papers produced per year is again close to the median, but the impact per refereed paper and the normalized citation rate per year would put me high in the top 10% of the Dutch permanent astronomer staff.
    All publications normalized by #authors   Refereed publications normalized by #authors
    n cites cite/pap   n cites cites/pap <#auth> pap/year cites/year
R.S. de Jong   49 1223 25   11 1099 99.9 2.7 0.8 78.5
mean   54 770 12.9   26 684 24.9 3.4 1.5 27.3
dispersion   40 757 7.3   21 675 14.3 1.3 1.1 22.3
ten-percentile   13 96 5.1   6 87 9.2 2.0 0.4 5.8
first quartile   24 215 7.2   11 180 15.3 2.4 0.6 11.4
median   46 508 11.6   22 420 22.0 3.3 1.0 19.7
third quartile   71 1549 17.0   33 1025 30.8 4.1 1.4 36.0
ninety-percentile   114 1932 25.2   52 1636 49.1 5.3 2.0 56.9


In the last table the first three columns calculate the average impact ratio per paper. The first column is the same as in the top table. The next column shows the expected average citation rate based on the average citation rate in MNRAS since publication of one's first paper. MNRAS is used as it is close to average for the field of astronomy as a whole. I have corrected this to the 2007 number for my work. The impact ratio is just the ratio of the actual citation rate to the expected one. This shows again that my papers have very high impact factors. The rest of the columns refer to one's most highly cited papers. The "all" column refers to any paper, the other columns refer only to one's most cited 1st author paper. My most cited 1st author paper (de Jong 1996, A&A 313, 45) puts me in the top quartile of Dutch astronomy, even though it is only 11 ears old, and it is in the top 10% in cites per year. My most cited paper (Bell & de Jong 2001, ApJ 550, 212) would make it easily on van der Kruit's "Highly Cited" list of papers with an impact ratio of 12.5, even though it is only 6 years old.
  Refereed journals   Most cited publication
  cites/pap expected impact ratio   all
(cites)
1st author (cites) cites/year impact ratio # authors
R.S. de Jong 65.3 25.6 2.55   339 253 23.0 7.5 1
mean 26.2 21.3 1.20   214 145 10.9 4.9 7.9
dispersion 12.7 3.2 0.57   152 133 10.6 4.9 7.9
ten-percentile 11.8 15.4 0.58   59 34 2.7 1.6 1
first quartile 17.3 20.3 0.86   77 55 4.2 2.8 2
median 24.7 23.9 1.11   186 101 6.2 4.4 3
third quartile 33.1 24.5 1.45   352 199 12.1 7.4 5
ninety-percentile 43.8 24.6 2.08   426 386 22.2 14.3 9

Back to Roelof's home page ()

Tue Dec 18 15:09:26 2007