Publication Metrics For Chandra Science

Previous: CXC 2003 Science Press    Contents   Next: Chandra Trains The Next



Publication Metrics For Chandra Science

Chandra is now in its prime. Operations are fairly smooth and well-understood. Astronomers, especially those who work in X-rays, are as informed as they may ever be of Chandra 's capabilities. To evaluate the relative scientific effectiveness of Chandra observations, we have undertaken to compile and study data on related publications. Across institutions and even across disciplines, professional publications and citations represent the most commonly used and accepted method of tracking scientific productivity. Another advantage is that, thanks to the Astrophysics Data Service (ADS), publication data are highly complete, and relatively easy to track.

The CXC has been accumulating data on Chandra-related publications. In December 2002, the Chandra User's Committee suggested that the ChandraDirector's Office expand a study of publication metrics, and recently (Spring 2003), the Space Telescope Institute published a Newsletter article on Hubble Science Metrics. To allow for metrics as complete and flexible as is practical, the Chandra Data Archive (CDA) Group at the CXC has designed and populated a large bibliographic database, described in the previous article by Arnold Rots. The ADS is queried daily for publications related to Chandra or to X-ray astronomy. These publications are scanned and reviewed by hand. Publications are categorized by several different parameters and entered into the Bibliography Database. When any publication analyzes or refers to Chandra data, Chandra target ObsIDs are sought in the publication, or if necessary directly from the authors. (Authors please list relevant Chandra ObsIDs whenever feasible!) ObsID links allow for cross-correlation with the full CXC Observation Catalog, a database that links together the X-ray data, observation status and description (date, exposure time, instrument, etc.). ObsIDs are in turn linked to a Proposal Database containing the scientific category of the proposal, proposal type, etc. The Bibliography Database is kept within a couple of weeks of ADS updates, and has been backfilled to pre-launch.

The Chandra Bibliography Database thus allows for a large number of publication statistics to be compiled. We are just beginning to study and understand some of these, and we are immediately aware that while a histogram may look simple, there are potentially many embedded assumptions and complications. For instance, publication statistics may include articles, books, theses, reviews, proceedings, circulars, etc. These may be refereed or not. Categories of Chandra publications include those that
1. present specific Chandra observations
2. refer to published results from Chandra
3. describe Chandra instrumentation, software, or operations
4. predict Chandra results
and a fifth category for publications that cannot be classified. Any of these may also be flagged as presenting analysis or modelling of previously published or reduced data, or multiwavelength follow-up. The moral of this story is that a lot may be folded into what looks like a simple statistic. With these caveats, we provide here plots of some of the simplest preliminary publication metrics. Library science afficionados can look forward to more detailed information soon on CXC web pages.



FIGURE 14: The total number of Chandra-related papers per year. This counts only papers in refereed journals, but we include Chandra data from all types of proposals (GO, GTO, DDT, TOO and CAL).


Figure 14 shows the total number of Chandra-related papers per year. We count here only papers in refereed journals, but we include Chandra data from all types of proposals (GO, GTO, DDT, TOO and CAL). Here we tally Chandra-related papers as all those in category 1, together with those category 2 papers that have at least one flag. Chandra was launched mid-1999, followed by 2 months of Orbital Activation, checkout, and calibration, then 2 more months of GTO observations. Publications from any new observatory take a while to gear up as operations become routine, as users gain experience, as its reputation grows, and as the volume of data increases. These effects are easily seen in Figure 14. We expect the number of papers per year to level off or slightly increase as the size and use of the data archive grows.



FIGURE 15: The number of citations to Chandra-related papers (categories 1 and 2) published in the years shown.


Figure 15 shows the number of citations to Chandra-related papers published in the years shown. The ramp-up of productivity in Figure 14 is convolved here with the delay between the date a publication appears in the literature, and when citations to it are published in subsequent papers. In the general astronomical literature, a lag of 2 years between a publication and the peak in citations to it is typical. Clearly, the impact of Chandra measured in this way is only just beginning to be felt.

Paul J. Green and Mihoko Yukita




Previous: CXC 2003 Science Press    Contents   Next: Chandra Trains The Next




cxchelp@head.cfa.harvard.edu