![Sample our Humanities journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5939/TOC-Subject-Sample-2021-Humanities.jpg"})
Abstract
The authors study the 30 insurgencies occurring between 1978 and 2008 using four methods crossing the qualitative/quantitative divide. The four approaches are narrative, bivariate comparison, comparative qualitative analysis, and K-medoids clustering. The quantification of qualitative data allows the authors to compare more cases than they could “hold in their heads” under a traditional small-n qualitative approach, improving the quality of the overall narrative and helping to ensure that the quantitative analyses respected the nuance of the detailed case histories. Structured data-mining reduces the dimensionality of possible explanatory factors relative to the available observations to expose patterns in the data in ways more common in large-n studies. The four analytic approaches produced similar and mutually supporting findings, leading to robust conclusions.
Notes
1. See, for example, the wide range of articles, opinions, and (most of all) discussions that have taken place on the Small Wars Journal blog (http://smallwarsjournal.com/).
2. Parsing specific types of conflict helps to differentiate between a conflict's organizational social structure, goals, and tactics. See Mayer Zald and Michael Berger (1978).
3. Of course, any likely application of these findings will seek to generalize or extrapolate beyond the 30-year population to some future case. Extrapolating of inferences from one population to another (or to a larger super-population) steps outside the bounds of traditional assumptions for statistical inference (Hahn and Meeker Citation1993). The assumption of applicability of inferences outside their population is routinely made but is not routinely quantifiable. In this case, comparability depends on one's answer to the question, “How like or unlike the insurgencies of the past 30 years will any future insurgency be?” We recognize that different answers to that question are eminently plausible.
4. The United States was involved in several cases in an advisory role or as a contributor of materiel support to one side or the other. In several cases, the United States contributed significant military forces (e.g., Kosovo, Afghanistan), but in those cases, the United States and its NATO allies were technically on the side of the insurgents. The United States also supported the insurgents in both Nicaragua cases.
5. Nearly two decades after the cease-fire that officially ended the fighting, the conflict still remains “frozen.” See Will Englund (Citation2011).
6. Throughout, when we refer to data for “the case” without qualification, we mean to denote the factors as scored in the decisive phase of each case.
7. While praising innovations made by comparativists, including Ragin, Michael Shalev (Citation2007) proposed three main alternatives to both QCA and multiple regression. Among these alternatives are technical refinement, triangulation (combine regression with case studies), and multivariate tables and graphs and factor analysis (substitution). Lane Kenworthy (Citation2007) built on Shalev's work to argue for greater transparency in the presentation of regression findings; a more explicit discussion of the type of variation macro-comparative analysts seek to explain; a more concerted focus on the direction, size, and robustness of regression coefficients instead of statistical significance; and the use of regression to inform the discussion of cases.
8. Boolean algebra was developed in 1854 by George Boole (Citation2003). Boolean algebra differs from standard high school algebra in two ways. First, values are logical values instead of numerical values. These are true or false, present or absent, and are represented as 1 or 0. Second, logical values dictate slightly different mathematical operations obeying slightly different mathematical laws. Many readers will be familiar with Boolean search operators, such as and, or, and not, as they can be used in some search engines. The application of Boolean algebra here has two implications: It requires us to structure our data with logical values (true or false, or, in our case, present or absent), and it allows complex patterns of data to be reduced to the minimum set of factors necessary to determine a pattern, called prime implicants.
9. For the distinction between fuzzy sets and crisp sets, see Ragin (2000).
10. shows that 13 approaches received strong support. One of them, however, “continuation and contestation,” is an insurgent-side approach. We excluded it from the QCA because our emphasis is on successful practices for COIN forces, not insurgents.
11. Conceivably, if we had more cases, we (and QCA) would better be able to discriminate. The addition of even a few cases that were COIN wins but had fewer of the positive COIN practices or that were COIN losses but had more of the positive COIN practices would help eliminate some of the good practices from being possible prime implicants and perhaps allow us to at least identify tiers of good practices, with the top tier being most important or essential and a second tier being beneficial but less critical. Of course, if in a larger set of data a single case drives the outcome of the analysis, we should rightly be concerned with the possibility of a truly exceptional narrative that does not, in some sense, belong to the population of interest.
12. A reviewer asked if some of these factors or practices are not in fact tautologically related to COIN force success. Certainly, some may have that appearance. This is once again a situation in which the narrative analyses show their strength. Each case narrative includes the sequence in which events occurred, supporting our assertion that factors that emerged before the outcome was decided are contributors to, and not themselves results of, those outcomes; this is so in these 30 cases for all the factors identified as good factors/practices.