1532-2890/asset/olbannerleft.gif?v=1&s=d833098325c9f1060bcbee51adf276c155608167)
1532-2890/asset/olbannercenter.gif?v=1&s=661179918edb4fa732edfd3408eb050a6ce87809)
1532-2890/asset/olbannerright.gif?v=1&s=1ef8a363944134c502cbffa1937878a71b4cc635)
Research
Yes, the GIGP really does work—and is workable!
Article first published online: 4 JAN 1999
DOI: 10.1002/(SICI)1097-4571(199303)44:2<61::AID-ASI1>3.0.CO;2-J
Copyright © 1993 John Wiley & Sons, Inc.
Issue
1532-2890/asset/cover.gif?v=1&s=c9d33fcfc2cef6291d3cfd7d7be1de8ca023f1c7 "Journal of the American Society for Information Science")
Journal of the American Society for Information Science
Volume 44, Issue 2, pages 61–69, March 1993
Additional Information
How to Cite
Burrell, Q. L. and Fenton, M. R. (1993), Yes, the GIGP really does work—and is workable!. Journal of the American Society for Information Science, 44: 61–69. doi: 10.1002/(SICI)1097-4571(199303)44:2<61::AID-ASI1>3.0.CO;2-J
Publication History
- Issue published online: 4 JAN 1999
- Article first published online: 4 JAN 1999
- Manuscript Revised: 7 AUG 1992
- Manuscript Accepted: 7 AUG 1992
- Manuscript Received: 10 JUN 1992
Funded by
- Science and Engineering Research Council
- Abstract
- References
- Cited By
Abstract
The fact that many informetric data sets exclude the zero-category—corresponding to the nonproducers being unobserved—has led to difficulties in the implementation of Sichel's generalized inverse Gaussian-Poisson (GIGP) process for informetric modeling, despite its theoretical attraction. These computational problems have been surmounted by the development of a program giving maximum likelihood estimates of the parameters of the zero-truncated GIGP. This allows a unified and theoretically sound approach to the fitting of the GIGP and is illustrated using several of the classic informetric data sets. The method also highlights situations in which the model motivating the GIGP is inappropriate. © 1993 John Wiley & Sons, Inc.