Book contents
- Frontmatter
- Contents
- Acknowledgments
- 1 Introduction
- 2 Theoretical origins
- 3 Theoretical revision: the multiple hierarchy model
- 4 Identifying local hierarchies and measuring key variables
- 5 Empirical investigations
- 6 Further investigations I: great power interference?
- 7 Further investigations II: an African (interstate) Peace?
- 8 Conclusions, implications and directions for continued research
- Appendix: Replication with Correlates of War capabilities data
- References
- Index
- CAMBRIDGE STUDIES IN INTERNATIONAL RELATIONS
Appendix: Replication with Correlates of War capabilities data
Published online by Cambridge University Press: 22 September 2009
- Frontmatter
- Contents
- Acknowledgments
- 1 Introduction
- 2 Theoretical origins
- 3 Theoretical revision: the multiple hierarchy model
- 4 Identifying local hierarchies and measuring key variables
- 5 Empirical investigations
- 6 Further investigations I: great power interference?
- 7 Further investigations II: an African (interstate) Peace?
- 8 Conclusions, implications and directions for continued research
- Appendix: Replication with Correlates of War capabilities data
- References
- Index
- CAMBRIDGE STUDIES IN INTERNATIONAL RELATIONS
Summary
As mentioned in chapter 4, the Correlates of War composite capabilities index is a very widely used indicator of national power. I rely primarily upon GDP as a measure of national power resources, because of greater agreement between GDP and the conceptualization of national power within the power transition research tradition, but present this appendix with replications of all of my analyses using COW power. In general, all of the results presented in chapters 5 and 7 are reproduced quite closely in the tables below.
The first result in chapter 5 is the likelihood ratio test investigating the statistical appropriateness of pooling the five regions into one unified analysis. When power is measured with the COW index, the difference in log likelihoods between the pooled model and the sum of the five individual regional models is just below 21. With the same twelve degrees of freedom used in chapter 5, this value is almost statistically significant at the p < 0.1 level. Consequently, one might want to be more cautious in pooling with the COW power variable than with GDP as the measure of power. However, given that the difference in log likelihoods here is essentially insignificant, and that the GDP power model clearly can and should be pooled, I pool.
The next statistical model regressed the onset of war against parity and dissatisfaction controlling only for possible consequences of duration dependence by inclusion of time-interval dummy variables in line with Beck, Katz, and Tucker's (1998) procedure.
- Type
- Chapter
- Information
- Regions of War and Peace , pp. 207 - 215Publisher: Cambridge University PressPrint publication year: 2002
- 1
- Cited by