After World War II and the Vietnam War, the American government and military twice buried counterinsurgency policy, focusing instead on conventional war. After the draw down of forces in Afghanistan and Iraq, and the concurrent attempt to pivot toward Asia, the lessons learned from the two earlier wars seem to have been forgotten or never learned to begin with. Yet counterinsurgency remains among the most common of large-scale US military operations since 1945. Thus, America's Modern Wars by historian Christopher Lawrence, executive director of The Dupuy Institute (TDI), comes at a propitious time. It uses quantitative data to assess successful insurgencies and to address "the failures of the studies and analysis community, the U.S. Army, and the national defense leadership to properly study, understand and appreciate the complexities of guerilla warfare" (298).

Its disjointed nature reflects the book's roots in a dozen reports written for TDI, a self-described "organization dedicated to scholarly research and objective analysis of historical data related to armed conflict and the resolution of armed conflict."[1] The introduction and first chapter concern the origins and findings of TDI's Iraq Casualty Estimate. The second chapter discusses the work of certain counterinsurgency theorists "crudely lump[ed] … into ten rough 'schools' of thought" (41). The next fourteen chapters identify the quantitative data needed to determine what makes insurgencies succeed or fail. Chapters 17–18 compare TDI's findings to those of other theorists and to insurgent leaders' writings. Chapter 19 is a quantitative analysis of withdrawal and conflict termination. Chapters 20–21 explain the force ratio model developed in Iraq, Afghanistan, and Vietnam respectively. Chapters 22–24 summarize the book's findings and suggest areas for future research.

Unfortunately, America's Modern Wars does not achieve its major goals. Its quantitative methodology, particularly the operationalization of variables and cases, is never explicitly spelled out. The dependent variable, victory, is explained only late in the book, and even there, Lawrence admits that the causes of counterinsurgent wins are often unclear (237–38). Similarly, the discussion of outside support (78–82) neglects to indicate whether only the actual deployment of foreign forces is meant or the provision of armaments counts as well. Nor does the author specify the sources of his data, including those for civilian casualties (92) and wounded-to-killed ratios (154), omissions that preclude any replication of the study and undermine the cogency of his arguments.

Though he lists his case-studies in appendix III (316–18), Lawrence does not state his criteria for inclusion. Only in a description of a subset of cases does he set out the following five requirements:

But what precisely does Lawrence understand to be an "intervention," or a "developing" versus "developed" country? What, for that matter, qualifies as an "insurgency?" Such semantic imprecision is fatal to good quantitative analysis.

Lawrence goes on to acknowledge that his included "83 cases are not the complete database of post-World War II cases, which may be well over 300 cases, if all peacekeeping and interventions are included…. Nor is it a representative sample…. It does represent the most important and major cases, and many better known smaller cases" (150). Since his selection comprises both conventional and unconventional conflicts, it is often hard to know exactly what sort of war is being discussed. Moreover, by focusing on "better known" cases, he risks the appearance of selection bias, successful counterinsurgency wars of shorter duration being omitted as unlikely to be better known.

The author's quantitative findings appear to be based on cross-tabulations typically displaying (a) a table of counted results, (b) a table of corresponding percentages, and (c) the p-value or degree of statistical significance.[2] While cross-tabs can demonstrate a relationship between a (potentially) causal variable and a given outcome, they have serious limitations. The use of cross-tabs alone cannot establish whether interacting variables affect the likelihood of victory by either side in a war. But Lawrence hints that such interaction is determinative in his discussion of the effectiveness of barriers—"We suspect that there is another factor in play here" (108) and in population relocation (110).

Lawrence does undertake a pair of logistic regressions, a technique that can, by controlling for factors' effects, clarify the interplay of variables. In the first (65, 183), he tests the effect of force ratios and political concepts on the probability of counterinsurgent success. Indeed, his principal thesis is that counterinsurgents must outnumber the insurgents and that much depends on the goal of the insurgents. In the second instance of logistic regression, he tests the effects of population size, roughness of terrain, and a less than 20-percent urban population (144). While the first regression study reports coefficients, standard errors, and p-values, the second provides no such information; neither does Lawrence provide essential goodness-of-fit indicators.

The author's conclusions do not always jibe with his data. When examining the effect of force ratios on "insurgencies opposing outside intervening forces" (57), he finds that his results correlate to those for other subgroups and for the whole sample of insurgencies, but exiles to a footnote (59n5) the fact that the results are not statistically significant. This means his findings may be due to chance rather than a true causal relationship. Elsewhere, he dismisses findings that a relationship between intervening force commitment and victory is not statistically significant in cases of insurgencies against foreign intervention[3]—"we have reasons for not believing this is the case due to the tendencies seen in all tests we have developed" (163).

Some topics are mentioned only to be dropped without elaboration (torture, 87); others appear without context (for example, countries most willing to take significant losses, 173–77). Many findings are based on small fractions of cases: only four of "more than 50 peacekeeping missions since World War II" (151) are adduced. Regarding "decapitations" (targeted killings or capture of leaders), the author observes in a footnote that ten cases were not yet examined, adding that "It would be worthwhile doing a complete examination of this subject" (160n5).

Too often, Lawrence treats issues as if he were the first to discuss them: in the chapter "Rules of Engagement and Measurements of Brutality" he ignores valuable earlier quantitative analyses[4] as well as many studies on the question of war termination and victory.[5] In his conclusion, he stresses the need for research in three areas: pre-insurgency, early-stage insurgency, and developed insurgency. But many scholars have already explored these subjects, often through quantitative analysis of substantial datasets, as a quick perusal of periodicals like the Journal of Conflict Resolution, Civil Wars, and International Security reveals. What analysis Lawrence offers of other authors' work is perfunctory.[6]

Billed as "a theoretical work based upon the studies and analysis that The Dupuy Institute did for the Defense Department and intelligence agencies" (299), America's Modern Wars will put off general readers by its statistical analyses, and recondite terminology and acronyms,[7] while specialists will dislike its suspect statistical analyses and inattention to germane work by other scholars.

The subject of American counterinsurgency policies is a critically important one, especially today vis-à-vis escalating efforts against ISIL. Given further research and more meticulous analysis and presentation, the work of Christopher Lawrence and the Dupuy Institute could have done much more to deepen our understanding of counterinsurgency.[8]