Tracking Chandra Science Productivity

1
Tracking Chandra Science Productivity

• Publication Metrics

Paul J. Green (CDO)
special thanks to John Bright, Arnold Rots, and
Sherry Winkelman (Archive Group) and Mihoko
Yukita (CDO)
2
Which Metrics?

• PhD dissertations
• papers
• citations
• pages
• subgroups
• all publications
• refereed papers
• highly cited papers
• certain journals

3
Chandra Bibliography Database
Archive Group A. Rots, S. Winkelman, J. Bright

• Queries the ADS weekly
• (Title or Abstract) contains (AXAF
  OR Chandra OR X-ray)
• Requires additional human scanning, culling, and
  categorization.
• Database now current and backfilled.
• Allows public searches that link data ??
  literature
• CXC internal also links to PropDB, allows many
  statistics to be derived

4
Tabulated Bibliographic Categories

• Presents specific Chandra observations
• Refers to published Chandra results
• Predicts Chandra results (could be either
  astrophysical theory or data extrapolation)
• Describes instrumentation, software or operations
  
• Cannot be classified
• and any of these FLAGS may accompany the above
  categories
• A) Complementary observations
• B) Simulations or Follow-up Analysis
• C) Astrophysical theory that explains Chandra
  results
• D) Instrument flags (ACIS, HRC, HETG, LETG,
  HRMA, PCAD, EPHIN)
• E) Operations
• F) Software

5
Definition of "Chandra Paper"

• Presents specific Chandra observations explaining
  theory, or followup
• Category 1 (all, any, none)
• Category 2 (flag A, B, or C required)
• in other words 1 (2A, 2B, or 2C)

6
Currently Available Categoriesand Variables

• Science Category
• Proposal Category
• (VLP, LP, GTO, GO, TOO, DDT, CAL)
• Exposure Time
• N papers
• N citations
• These are mutually exclusive definitions in the
  Chandra databases, e.g., a TOO from a General
  Observer is not also counted as a GO.

7
SCIENCE CATEGORIES

• 1. Solar System
• 2. Stars and WD
• 3. WD Binaries and CVs
• BH and NS Binaries
• 4. SN, SNR and Isolated NS
• 5. Normal Galaxies Diffuse Emission
• Normal Galaxies X-ray Populations
• 6. Active Galaxies and Quasars
• 7. Clusters of Galaxies
• 8. Extragalactic Diffuse Emission and Surveys
  Galactic Diffuse Emission and Surveys

8
Proposal Categories

• Mutually exclusive in the Chandra databases
• GO
• GTO
• TOO
• DDT
• LP (first defined for Cycle2)
• VLP (first defined for Cycle5)
• N.B. V/LP status is optional even over the
  nominal 300/1000ksec limits. For complete
  stats, best to tally by exposure time.

• e.g., a TOO from a General Observer is not also
  counted as a GO.

9
Totals by Proposal Cycle
PAPERS
CITATIONS
10
Totals by Proposal Cycle

• Includes
• Refereed Chandra papers only
• Statistics through May 26 2004
• Includes all proposal types
• GO, GTO, TOO, DDT (no CAL)
• Statistics best for Cycle1
• Strong ramp-down, reflected strongly in ksec-1
  plots as well

11
Citations per ksec by Cycle

• How would you know that statistics are best for
  Cycle1?
• check linked plot data
• check totals plots

12
Example of Linked Plot Data

• Values for Citations_by-cycle.txt
• Cycle Citations N_Proposals
• Cycle1 10401 312
• Cycle2 3985 278
• Cycle3 474 304
• Cycle4 9 258
• Cycle5 0 90

• Statistics best for Cycle1

13
Totals by Exposure Time
S(Papers referring to any Chandra Target in an
Approved Program) in bins by Total Program
Approved Exposure Time
S(Citations to Papers referring to any Chandra
Target in an Approved Program) in bins by Total
Program Approved Exposure Time
14
Citations/ksec by Exposure Time with Cycles
SNi1(Citations to Targets in each
Program)/(Total ksec in Program)/N
Appears as if short observations have greatest
impact/ksec.
15
Large Projects Have Longer Citation Lag
16
Conclusions

• Short exposures appear to provide a larger
  science return per ksec.
• Relative productivity (papers) and impact
  (citations) of larger projects appears to have a
  longer latency.
• Some results have large impact but do not produce
  large Npapers or Ncitations