Evolution or revolution?

Journal of Strategic Studies

Volumn 28, Issue 3, 2005, Pages 437-448

  Horowitz, Michael ^a   Rosen, Stephen ^a  

^a HARVARD UNIVERSITY   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

STRATEGIC APPROACH;

EID: 22444439573     PISSN: 01402390     EISSN: None     Source Type: Journal    
DOI: 10.1080/01402390500137317     Document Type: Review

References (4)

1. This is not an argument for a binary (0, 1) coding. There are a variety of ways one could measure the modern system based on types of military organisations, doctrine, training etc. If it is not possible to know, before a military campaign, whether a country actually is employing the modern system or a particular level of competency with regards to the modern system, the theory cannot make predictions about the outcomes of battles because all of its predictive power is post-hoc, meaning it risks tautology.
2. Though beyond our scope due to space constraints, this also matters due to the assumed independence of his observations. The independence of observations issue is a methodological issue with substantive consequences. It is untrue that each observation in the dataset, especially since so many observations involve the same country or countries, mostly in the same handful of conflicts, is independent. Battles in 1945 between the US and Germany were clearly influenced by previous battles between the two sides and by battles they had with other actors. The failure to correct for this biases the results. For a more in-depth explanation of the 'time' issue, see Nathaniel Beck, Jonathan N. Katz and Richard Tucker, 'Taking Time Seriously: Time-Series-Cross-Section Analysis with a Binary Dependent Variable', American Journal of Political Science 42 (Oct. 1998), pp.1260-88.
3. For a more complete explanation and/or replication information, please contact the authors. These results were generated using the postgr3 command in STATA 8.0 and Clarify. For more on Clarify, see Gary King, Michael, and Jason Wittenberg, 'Making the Most of Statistical Analyses: Improving Interpretation and Presentation', American Journal of Political Science 44/2 (2000) pp.347-61.
4. Also, while we lack the space to fully describe this issue, a replication of Biddle's results in Table 8.2 and Table 8.5 with robust standard errors, appropriate due to the variance in the dataset, makes the F test of relationships in the model insignificant. This means all of the relationships described by Table 8.2 and Table 8.5 could be spurious and deserve more investigation.