Much has changed since James Quinliven kicked off discussion over manpower and counterinsurgency. One of the most significant differences is the availability now of useful collections of historical data for analysis.
Detailed below are the lineages for the data sets used in six of the seven analyses I have discussed. The cases used by Libicki and Friedman were drawn from databases created by several academic organizations and work by James Fearon and Daivd Laitin [1] and, Jason Lyall and Isaiah Wilson III [2]. Both Libicki and Friedman contributed additional research of their own to complete their datasets.
The data used by Lawrence, CAA, and IDA was all researched and compiled by The Dupuy Institute (TDI). Both TDI’s Modern Insurgencies Spreadsheets (MISS) Database and CAA’s Irregular Warfare Database contain data on at least 75 variables for each historical case.
The details of the dataset created and used by Hossack at Dstl have not been addressed in public forums, but it is likely to be similar.
Future directions for research
Given the general consensus of all of the studies that counterinsurgent manpower levels do correlate with outcome, the apparent disagreement over force ratio and troop density measures may not be as relevant as previously thought. More data collection and testing should be done to verify the validity of the postulated relationship between counterinsurgent force levels and the local population within an active area of operation.
Though there was consensus on the advantage of counterinsurgent manpower, there was no agreement as to its overall importance. More analysis is needed to examine just how decisive manpower advantages may be. Hossack and Goode suggested that a counterinsurgent manpower advantage may be important largely to prevent insurgent military success. Hossack and Friedman suggested that there may be points of diminishing manpower returns and Lawrence indicated that a force ratio advantage was decisive only against insurgencies with broad popular support. Given the potential difficulties in generating significant additional counterinsurgent manpower, it may be applicable and useful only under particular circumstances.
Due to the limitations of the available data, all of the studies based their analysis on data averages. The figures used for insurgent and counterinsurgent force sizes were usually selected from the highest annual totals across years or decades. All of the studies indicated the need to obtain more detailed data on individual cases to allow for more discreet and dynamic analysis to look for undetected links and patterns. Lawrence in particular called for examination of conditions before insurgencies begin and when they are just getting underway.
Friedman noted the value of quantitative analysis in helping to drive forward discussion and debate on defense and security topics. Research and analysis on insurgency and counterinsurgency was left to languish during the Vietnam War, only to be exhumed under the dire circumstances of the U.S. war in Iraq. It would be deeply unfortunate if promising new lines of inquiry were abandoned again.
I just wanted to do a little ‘back of the envelope” comparison between these two air campaigns. In the case of Kosovo, if you believe the casualty figures provided virtue of Wikipedia (which are not always incorrect), they flew 38,004 sorties and killed 956 supposed hostiles (that is 956 killed, 5,173 wounded and 52 missing for a total of 6,181 casualties). Or, maybe that should be 10,484 “strike sorties.” Regardless, this was either 38 sorties per person killed or 10 “strike sorties” per person killed (missing are counted among the killed for this calculation). Or if based on total casualties, 6 sorties per casualty or 1.7 “strike sorties” per casualty. Now, only 35% of the bombs and missiles used were precision guided.
If you look at the link in my post “Bleeding an Insurgency to Death” you could surmise that in 2015 in Iraq and Syria, the U.S. and its allies dropped 28,714 “munitions.” They claim 25,500 killed. This is 1.13 “munitions dropped” per person claimed killed. So, one bomb kills one person.
Kosovo was 23,614 “air munitions” for 1,008 deaths or 6,181 casualties. This is 23 “air munitions” per person killed or 3.8 “air munitions” per casualty. So, Kosovo in 1999 is 23.42 ‘air munitions” per person killed while Syria and Iraq in 2015 is 1.13 “munitions dropped” per person claimed killed. This is an effectiveness improvement of over 20 times! Of course, these campaigns were conducted against different terrain and somewhat different circumstances that may favor one over the other. We have not evaluated those factors (after all, this is just “back-of-the-envelope” calculations).
Now, in Kosovo, only 35% of the bombs and missiles used were precision guided. Don’t know what the figure is now, but if it was 100%, and if we assumed that only the precision guided munitions in Kosovo hit anything (a questionable assumption), then we still end up with an effectiveness improvement of over seven times.
But maybe the 25,500 killed really means 25,500 killed and wounded (of which the majority would be wounded). In that case using the Kosovo figures for total casualties you end up with 3.82 “air munitions” per casualty versus 1.13 “munitions dropped” per casualty for Syria and Iraq. Again, if we completely discount the effectiveness of non-precision guided munitions in Kosovo, and assume that in Syria and Iraq 100% of the munitions are precision guided, then we end up with similar levels of effectiveness per casualty (1.34 “air munitions” versus 1.13 “munitions dropped” per casualty). There are a lot of “ifs” to get to this point.
Now, one should not put to much stock in the “back of the envelope” calculations, but something doesn’t quite line up here.
A lot of time could and should be spent examining this campaign. It was example of trying to change of countries’ policy by an extended air campaign. It is such an odd case that I left it out of my original article (see blog post “Defeating an Insurgency by Air”, which is also posted to the History News Network). I simply did not want to address it. I probably should have.
As with the U.S. in Afghanistan in 2001, this was a case where U.S. and NATO airpower was supporting the insurgency. The mostly Muslim Kosovars were revolting against the established Yugoslavian government. In this case through, there was not much of an insurgency and the actions on the ground consisted mostly of Yugoslavia attempting to firm up their control over the area, in part by driving the local Kosovars out of the area. At one point, hundreds of thousands of people were dislodged or migrating from the homes.
The United States and NATO were really not providing air support for the insurgents, so much as attempting a punishing bombing campaign against the Yugoslavian Army without taking any risk of losing aircraft. This was an odd aerial bombardment that continued for 78 days at a loss of five aircraft (two due to combat). Finally, on 3 June 1999, the Yugoslavian government agreed to cease their operations in Kosovo and withdraw their forces. Their losses according to some sources were 14 tanks, 18 APCs, 20 artillery pieces and 121 airplanes and helicopters (see Wikipedia: NATO Bombing of Yugoslavia). This was surprisingly low ground losses for a 78 day air campaign that included 38,004 sorties and involved over a thousand aircraft. This included 10,484 “strike sorties; that released 23,614 “air munitions” over Kosovo (these figures are from NATO covering the period from 24 March to 9 June 1999). This was 6,303 tons of munitions and 35% of the bombs and missiles used were precision guided.
Many commentators, including the NATO commander General Wesley Clark, consider that the primarily reason they withdrew is because the Yugoslavian leader, Milosevic, believed that the U.S. was about to insert ground forces in the campaign.
So it was a successful air campaign, but the reason it was successful was that the Yugoslavian leader thought that it was about to turn into a ground campaign. While this is one of the few cases of a pure air campaign actually achieving its stated political goal, it is not a case of an insurgency being defeated primarily by air, especially as the air campaign was done in support of the insurgency.
Hossack [Dstl] (2007), Blaho & Kaiser [CAA] (2009), and Lawrence [TDI] (2015) found no statistically meaningful relationship between troop density and insurgency outcome.
Just to clarify, I went only far enough to prepare the following table:
This was work I did back in 2004 using the original 28 cases we developed the Iraq Casualty Estimate from. At that time, Quinliven had only used 13 cases to establish his theme, and none had a density above 23.7. We did not rigorously test and analyze this theme, as it was clear from even a quick look that there was little value to it “…as a relationship that can be used analytically for predictive purposes for an insurgency.” The original purpose of Quinliven’s estimate was to look at “stability operations.” This is discussed in pages 70-71 of my book.
To summarize the findings of the seven large-N case studies of the relationship between manpower and counterinsurgency:
Troop Density (troops per inhabitant)
Goode [CAA] (2009) asserted a statistically meaningful relationship between troop density and insurgency outcome.
Hossack [Dstl] (2007), Blaho & Kaiser [CAA] (2009), and Lawrence [TDI] (2015) found no statistically meaningful relationship between troop density and insurgency outcome.
Kneece, et al [IDA] (2010) and Friedman (2011) found a statistically meaningful relationship between troop density in defined areas of operation and insurgency outcome
Friedman (2011) asserted that there was no discernible statistical support for a benchmark troop density level (i.e. 20 troops/1,000 inhabitants).
Force Ratios (counterinsurgents per insurgent)
Hossack [Dstl] (2007), Libicki [RAND] (2008), Blaho & Kaiser [CAA] (2009), and Lawrence [TDI] (2015) asserted a statistically meaningful relationship between force ratios and insurgency outcome.
Goode [CAA] (2009) and Kneece, et al [IDA] (2010) rejected the validity of a relationship between force ratios and outcome due to an inherent unreliability of relevant data.
Friedman (2011) identified a statistically meaningful relationship between force ratios and outcome when controls were applied to the data.
Lawrence [TDI] (2015) found a strong relationship between force ratios, the nature of an insurgency, and insurgency outcome.
Manpower and insurgency
At first glance, it would appear that despite the recent availability of historical data on insurgencies, the debate over the relationship of force ratios and troop density to outcomes remains an open one. Amidst the disagreement, however, one salient conclusion stands out: all of the studies generally agree that there is a positive correlation between counterinsurgent force strength and the outcome of an insurgency. The collective analysis suggests that the commitment of larger numbers of counterinsurgent forces has historically correlated with more successful counterinsurgency campaign outcomes. What remains open to dispute is just how significant this finding is, or whether it matters at all.
Troop density (countrywide) vs. troop density (AO)
Another broad conclusion from these studies is that there appears to be no statistical support for the original Quinlivan troop density construct as measured by the number of counterinsurgents per inhabitant. The only study to support this without qualification was Goode [CAA] (2009). The extensive data collection and testing conducted by Friedman (2011) also cast serious doubt on the validity of the notion of force level benchmarks.
However, Kneece, et al [IDA] (2010) and Friedman (2011) both made a compelling case for the usefulness of troop density as measured by the number of counterinsurgents per inhabitant within a defined area of counterinsurgent operations. When measured in this manner, a clear correlation was found to exist between troop density and insurgency outcome. This notion has considerable qualitative appeal. Insurgencies generally do not occur uniformly throughout an entire country. Insurgent activity usually takes place within a specific region or area. Consequently, counterinsurgent forces are not deployed uniformly throughout a country, but rather to areas with the highest insurgent activity. This revised troop density construct definitely merits further study.
Force ratios and data reliability
One argument raised against the applicability of using force ratios is that data relating to insurgents is based on either counterinsurgent estimates or imprecise counts, or unreliable information from insurgent sources. This uncertainty therefore simply renders the data invalid for analysis. This claim seems overstated. While it is relatively certain that there are inaccuracies in such data, it is implausible to think that it is all hopelessly flawed or fictitious. In nearly all the datasets, the data are collected as reported. This variety in sourcing would seem to auger against systematic bias, which would truly render the data invalid.
Data collected in an unsystematic way is definitely going to be fuzzy or noisy, but again, this does not invalidate its usefulness. As my colleague Chris Lawrence contends, even if the insurgent force strength data is inaccurate, it is not incorrect by an order of magnitude. The range of error is probably more like +/- 50%. Random changes in insurgent force size by +/- 50% still produce similar analytical results after regression analysis. Insurgent force size data may be noisy, but that in itself is an insufficient reason alone to discount it.
Sensitivity of results to coding choices/definitions
Given the general agreement that there is a relationship between manpower and outcome, it seems odd that there is still deep disagreement over specific aspects of this. One possible explanation for this is the wide variation in definitions of terms and variables. It should be noted that despite the very large body of research and scholarship on insurgency and counterinsurgency, there is very little consensus on how to define such conflicts. Both Kneece, et al [IDA] (2010) and Friedman (2011) pointed out that analytical outcomes are sensitive to how the variables are defined. Kneece, et al [IDA] (2010) did a quick check on how winning or losing was scored among 36 of the same cases in five different data sets and found agreement on only 11.
Some of the variations in the conclusions may be due to case selection. There are no universally accepted definitions for what insurgency or counterinsurgency are, or any meaningful distinction between these types of conflicts and less violent variants such as peacekeeping operations, interventions, or stabilization operations. The authors of each of the studies established clear but differing criteria for case selection, resulting in analyses of similar datasets with some overlap in common cases.
My next and final post in this series will address the origins of the various datasets and potential future directions for research on this subject.
The most meaningful quote I know of about the value of historical study is “The lessons of history are that nobody learns the lessons of history.” Some may write this off as just cynicism, but unfortunately, “history repeats itself,” and we have seen this all too often. There is 3,400 years of documented military history, and this rather extensive data base of material is often ignored; and when it is accessed, often it is to grab an example or two that supports whatever pre-conceived notion that the user already has. It is a discipline that has been poorly used and often abused. Part of our interest in quantified historical analysis is that we want to study the norms, not the exceptions; not the odd case or two, but what are the overall pattern and trends. Sometimes I think the norms get lost in all the interesting and insightful case studies.
Anyhow, there was a posting in another blog that my fellow blogger, Dr. Woodford, brought to my attention that included the formula 30,000 – 25,000 = 30,000. The link to his post is below:
The problem with this “kill-em’-all with airstrikes” rule, is that it is not working. Pentagon officials claim that at least 25,000 Islamic State fighters have been killed (an anonymous official said 23,000 in November, while on Wednesday, Warren added “about 2,500” more were killed in December.) Remarkably, they also claim that alongside the 25,000 fighters killed, only 6 civilians have “likely” been killed in the seventeen-month air campaign. At the same time, officials admit that the size of the group has remained wholly unchanged. In 2014, the Central Intelligence Agency (CIA) estimated the size of the Islamic State to be between 20,000 and 31,000 fighters, while on Wednesday, Warren again repeated the 30,000 estimate. To summarize the anti-Islamic State bombing calculus: 30,000 – 25,000 = 30,000.
This post brings back a few memories of our work on Iraq in 2004-2006. If you note in my book America’s Modern Wars there is an entire chapter on “Estimating Insurgent Strength” (pages 115-120). Part of our concern, which we briefly documented on page 116, was that the officially release estimated of insurgent strength remained at 5,000 forever. It was a constant figure, no matter how nasty the situation got. We really did not believe it. Then, when everything fell apart and the insurgents grabbed Mosul (sound familiar?), the estimate was revised upwards to 20,000. This was better, but it still seemed too low to us, especially as the U.S. was claiming something like 12,000 insurgents killed a year. Needless to say, if they were killing 60% of the insurgents a year, this was an insurgency that was going to quickly be bled to death. As we now know with a decade of hindsight, this did not happen.
This was the reason for section in my book called “Bleeding an Insurgency to Death (pages 156-158). Needless to say, something was wrong with the math somewhere, and our own estimate of insurgent strength was something like 60,000 (see page 116). As Mr. Zenko’s blog post points out, something remains wrong with the math in the air war against ISIL.
The previously quoted Air University Review article discussing air power and the Mad Mullah clearly overplayed the impact of air power on that campaign and rather significantly downplayed the importance of the 3,600 ground troops involved. Oddly enough, various Wikipedia and other articles on the internet make similar claims. The article then goes on to address Iraq in 1920. It states:
By 1920, Churchill had asked Trenchard to plan much more ambitious project, to control Mesopotamia (Iraq) by air. The British were, at the time, nurturing a new Arab government in Iraq, a government not popular among the Arab tribes populating the country. These tribes seemed totally unimpressed with British-sponsored progressive government, which included rules about taxation and standards of acceptable behavior. In late 1920, a serious rebellion against British rule was in progress; the 80 British and Indian battalions (120,000 troops) garrisoning the country were being hard pressed to maintain order. An additional 15,414 men sent from India were quickly absorbed in trying to control an insurrection of at least 131,000 armed men. The British forces were scattered throughout the country, protecting population centers and vulnerable lines of communication. With simultaneous outbreaks of violence in several areas, the British force proved too weak in any single spot to deal effectively with the problem. Even with 63 aircraft working with the army, putting down the 1920 insurrection in Iraq was a costly business: about 38 million pounds.
Now this was a revolt of 131,000 against 120,000, with an additional 15,414 reinforcements according to the same account (other accounts give slightly different figures, see Haldane (link below), pages 64 & 325). So, over 135,000 trained troops against the 131,000 rebelling Iraqis. Not surprising, this did not go well for the rebelling Iraqis, with or without air power. In fact, the entire revolt was over in less than four months. In the case of Somalia, the RAF has 12 planes supporting 3,600 troops or one plane per 300 troops. Now the ratio appears to be one plane per 2,150 troops. This would imply a much lower level of air support than in Somalia.
The British commander in Iraq, Lt. General Haldane, did write a book on the subject, The Insurrection in Mesopotamia 1920:
Haldane does address the support he received from the Royal Air Force squadrons (see pages 300-301), and details the various operations and engagements fought. He credits the air force on one occasion with greatly assisting in the defense of an isolated garrison at Rumaithah. This is pretty far from a case of an insurgency being defeated primarily by air. It is clear that statements like those found in Wikipedia that “The use of aircraft shifted the advantage to the British and played a huge role in ending the revolt,” seem to grossly overstate the case (see Wikipedia “Iraq Revolt against the British”). These types of claims are made in several other articles on the net. Haldane’s account certainly does not support these claims.
The revolt started with peaceful protests in May 1920. The armed revolt started in July and was pretty much over by mid-October of that year. It was a 3 1/2 month long revolt. It was never able to establish itself in the larger cities like Baghdad, Mosul and Basra, where the British maintained garrisons. The British then assigned the Iraqi’s a king and established a protectorate over Iraq. This insurrection was followed, of course, by an insurrection in Kurdistan in 1922.
The figure of 131,020 revolting armed Iraqis comes from Haldane, page 124. He states that there were 59,805 armed with modern rifles (also see page 298 and 328). We have no idea how accurate these estimates are (but remain suspicious). Still, it was a significant enough revolt that British casualties were 2,269. This was 426 killed (312 killed, 113 died of wounds and died while prisoner of war), 615 missing (451 missing and 164 prisoners of war) and 1,228 wounded (British figures are from 2 July to 17 October). Haldane states that “Most of the missing were killed, a few only rejoining.” I gather this means that the British lost over 800 troops killed during this campaign (55 British killed and 141 British missing, and 354 Indians killed and 282 Indians missing, and 17 “followers” killed and 28 “followers” missing for a total of 877). Haldane claimed that 8,450 Iraqis killed and wound in the rebellion (see Haldane, page 331). A four-to-one exchange between insurgents and professional military is not out of line with other figures we have seen.
After the revolt, the British, with Winton Churchill as Secretary of State of the Colonies, decided to “police” Iraq from the air. It was this effort, which included air campaigns against recalcitrant villages (and therefore civilians), that led many to focus on the air aspects of this revolt. The 1920 revolt was clearly not put down primarily by air power. There is probably good reason to examine the subsequent efforts to use air power in Iraq, as this was an attempt to keep a potential insurgency off balance by an extended air campaign.
This was also one of those ‘brief insurgencies” which we have not invested a lot of time examining. One of the results of our work was that insurgencies tend to last 4 ½ years or less, or last for 7 years or more. We really had no examples of insurgencies lasting between 4 ½ years in length of 7 years in length in the 100+ cases we studies. In effect, they are either over quickly, in a couple of years, or they drag on for seven years or longer (see America’s Modern Wars, pages 24-25 and 148). This did lead me to consider whether I should categorize or classify “revolts” as something different than “insurgencies.” (for example, would the overthrow of Batista in Cuba in 1958 be classified as a “revolt” or as an “insurgency”?). But, we did not have strong reason to do so for of our analysis at the time, so therefore passed on this subject.
Even as James Quinliven’s assertion of the importance of force density became axiomatic during the dark days of the U.S. war in Iraq, other researchers began to take a closer look at the role force strength played in past insurgencies. What quickly became evident was that the lack of available empirical data to analyze limited the validity of the analysis. Quinliven’s conclusions had been based on just 13 cases. The persuasiveness of conclusions of similar studies were also limited by small sample sizes.[1] Larger data sets, such as those provided by the Correlates of War Project or the Empirical Studies of Conflict Project were available but did not contain the detailed strength data needed for force ratio analysis. The urgency of the Iraq conflict led to government funding to collect the specific data needed to facilitate large-N case studies based on samples that approached statistical significance.
The Dupuy Institute (TDI) undertook one of the earliest efforts to gather a significant sample of historical insurgency data for serious analysis. Between 2004 and 2009, sponsored by the U.S. Army Center for Army Analysis (CAA); Office of the Secretary of Defense, Net Assessment; and the National Intelligence Council of the Central Intelligence Agency, TDI amassed data on 109 cases of post-World War II insurgencies, interventions, and peacekeeping operations. 83 cases were used for analysis.
Led by Chris Lawrence, TDI’s analytical effort found that insurgency outcomes closely tracked the force ratio of counterinsurgents to insurgents. While overwhelming numbers were not required to defeat an insurgency, force ratios above 10-to-1 nearly always guaranteed a counterinsurgent victory. Conversely, lower force ratios did not preclude success, but conflicts with two or fewer counterinsurgents per insurgent greatly favored an insurgent victory.
When force ratios were assessed together with the nature of the motivation for the insurgency, TDI found that force ratios had little impact on the outcome of insurgencies with a limited political or ideological basis. However, when facing broadly popular insurgencies, counterinsurgents lost every time they possessed a force ratio advantage of 5-1 or less, failed half the time with odds between 6-1 and 10-1, but succeeded three-quarters of the time when outnumbering the insurgents by 10-1 or more.
TDI also examined the relationship between force density and conflict outcome, but was unable to find any correlation in the data.
In 2004, the Defence Science and Technology Laboratory (Dstl) of the British Ministry of Defence initiated a multi-year research program into counterterrorist/counterinsurgency campaigns as an extension of previous research on asymmetric warfare. The three-year effort, led by Andrew Hossack, involved data collection on 44 post-World War I asymmetric conflicts, of which 34 were used for analysis.
Hossack’s study concluded that there is a relationship, albeit weak, between between the odds of military campaign success and force ratios as measured in terms of a ratio of median annual force strengths. The advantages were relatively modest. Each tenfold increase in the ratio of counterinsurgents to insurgents improved the chances of success by only 30%. A 10-1 manpower advantage yielded a 47% chance of success. Improving the ratio to 100-1 still only afforded a 77% chance of winning.
In 2006, the Office of the Secretary of Defense contracted RAND’s National Defense Research Institute for a comprehensive study of insurgency and counterinsurgency. In support of the project, Martin Libicki directed the compilation of data on 89 post-World War II insurgencies to analyze why they began, grew, and were resolved. The original list of cases was based upon a dataset of 127 insurgencies created by James D. Fearon and David D. Laitin[5] and modified.
With regard to force ratios, Libicki concluded that “as a general rule, the greater the government-to-insurgent force ratio, the lower the odds of an insurgent victory,” but that “the relationship is not particularly strong.”
CAA
In 2008, CAA contracted with TDI to code the data for it had collected for 102 post-World War II conflicts and design a database for it, which was designated the CAA Irregular Warfare Database. The availability of the Irregular Warfare Database spawned two CAA analytical efforts.
The first was conducted by Justine Blaho and Lisa Kaiser in 2009, who sought to build a model to predict the outcome of irregular wars. They tested 34 independent variables from 74 of the cases in the database and found 11 with significant correlations to outcome, including the peak counterinsurgent-to-insurgent ratio for each conflict. Blaho and Kaiser removed the variable for counterinsurgent-per-civilian (force density) from their analysis because it did not demonstrate a significant correlation to outcome. Among other conclusions, they determined that counterinsurgents had a greater probability of winning an irregular war if the peak year counterinsurgent-to-insurgent ratio is high.[6]
In 2009, CAA undertook another study aimed at providing historical context for U.S. force levels in Iraq and Afghanistan. Their goal was to find the number of counterinsurgent forces required to arrest and reverse a given level of insurgent violence. The analysts whittled the number of cases from the Irregular Warfare Database to be analyzed to 42 by eliminating those not deemed to be insurgencies, those lacking complete data, and third-party interventions on the side of the insurgents.
In contrast to the previous work using data collected by TDI, the study analysts rejected the validity of counterinsurgent to insurgent force ratios. As justification, they cited the unreliability of data on insurgent force size, insurgent and civilian casualties, and counts of insurgent attacks. Instead, they adopted the force density construct “in accordance with doctrine and previous studies.” In contrast to Quinliven’s 20/1,000 ratio, the CAA analysis concluded that “the minimum counterinsurgent force is 2.8 soldiers per 1,000 residents, with more forces required as the violence level increases.”[7]
In preparation for the 2010 Quadrennial Defense Review, Office of the Director, Cost Assessment and Program Evaluation (CAPE), Department of Defense asked the Institute for Defense Analysis to update a previous study estimating ground force requirements for stability operations. For the revised study, the IDA analysts also made use of the CAA Irregular Warfare Database, selecting 41 cases “with the objective of establishing a relationship between force density and outcomes.”
Like the second CAA study, the IDA analysts deemed force ratios to be a “flawed and unreliable indicator,” stating that “we do not believe that information about the size of insurgent forces is sufficiently credible to be subjected to statistical analysis.” Acknowledging previous studies that did not find a statistically meaningful link between force density and outcome, the IDA analysts adjusted the dataset in three ways: to use force densities based on their estimates for the local population of the actual areas of operation in historical cases (as opposed to the entire country); to count indecisive outcomes as military operational successes for the counterinsurgents; and rescoring several outcomes as indecisive, rather than failure for counterinsurgents.
Using this adjusted data, the IDA analysts found a statistically significant, though not robust, relationship between force density and outcome. They concluded that a density of 20‐25 troops per thousand population was consistent with a 50‐60% chance of success. Successful outcomes decreased greatly for force densities below 15 troops per 1,000, and for operations with ratios above 40 troops per 1,000, the success rate exceeded 75%.
In 2011, Jeffery Friedman, then a doctoral candidate in the Harvard Kennedy School, published a large-N academic study on insurgency force strengths, which addressed the basic question “how does manpower affect counterinsurgency?” As his starting point, Friedman adopted the dataset created by Jason Lyall and Isaiah Wilson III of 173 post-World War I insurgencies.[10] After dropping two cases, he added his own data from 400 additional sources on foreign and indigenous counterinsurgent force strengths, the size and population of areas of operations, and insurgent force size.
Friedman subjected his data to a broader range of statistical testing than the previous studies. He concluded that there was no statistical correlation between overall numbers of counterinsurgents or counterinsurgents per area and outcome. He found that counterinsurgent to insurgent force ratios were significant only with controls on the data. However, troop density – measured as the ratio of counterinsurgents per inhabitant in the specific area of operations that the counterinsurgents sought to control – demonstrated a consistent, positive correlation with insurgent success.
Perhaps of greater significance, Friedman also found that manpower did not necessarily have a decisive impact on insurgency outcome. He could find no discernible empirical support for Quinliven’s 20/1,000 ratio. A counterinsurgent troop density of 80 troops per 1,000 inhabitants was less than twice as likely to succeed as a counterinsurgent force sized an order of magnitude lower. In a model derived from his analysis, as force size increased from 5 to 80 troops per 1,000 inhabitants in the area of operations, the probability of success rose by less than 15 percent.
Conclusion
The quality of quantitative analyses of force strengths in insurgencies and counterinsurgencies improved greatly between 2003 and 2011, but the debate remains unresolved. My next post will look at the conclusions of these studies and evaluate whether they tip the scales one way or another in favor of force ratios or troop density.
Notes
[1] John J. McGrath, Boots on the Ground: Troop Density in Contingency Operations. Fort Leavenworth, KS: Combat Studies Institute, 2006; James Dobbins, Keith Crane, and Seth G. Jones, The Beginner’s Guide to Nation-Building. Santa Monica, CA: The RAND Corporation, 2007; Joshua Thiel, “COIN Manpower Ratios: Debunking the 10 to 1 Ratio and Surges,” Small Wars Journal, 2011; Riley M. Moore, “Counterinsurgency Force Ratio: Strategic Utility or Nominal Necessity?” Small Wars & Insurgencies, 24:5, 2013
[2] Christopher A. Lawrence, “The Analysis of the Historical Effectiveness of Different Counterinsurgency Tactics and Strategies,” Presentation at the Cornwallis Group XIII: Analysis in Support of Policy, The Pearson Peacekeeping Centre Cornwallis Park, Nova Scotia, Canada, 17-20 March 2008; Christopher A. Lawrence, America’s Modern Wars: Understanding Iraq, Afghanistan and Vietnam, Philadelphia, PA: Casemate, 2015
[3] Andrew Hossack, “Security Force & Insurgent Success Factors in Counter-Insurgency Campaigns,” Presentation at the Cornwallis Group XII: Analysis for Multi-Agency Support, The Pearson Peacekeeping Centre Cornwallis Park, Nova Scotia, Canada, 2-5 April 2007; Andrew Hossack, “Strategic Success Factors in Counter-Insurgency Campaigns: With Discussion of the Modelling Approach,” Presentation to ORS Defence Special Interest Group, 24 February 2010
[4] Martin C. Libicki, “Eighty-Nine Insurgencies: Outcomes and Endings,” Appendix A in David C. Gompert and John Gordon IV, War by Other Means: Building Complete and Balanced Capabilities for Counterinsurgency, Santa Monica, CA: RAND Corporation, 2008
[5] James D. Fearon and David D. Laitin, “Ethnicity, Insurgency, and Civil War” American Political Science Review 97, 1, Feb 2003
[6] Justine Blaho and Lisa Kaiser, “A Predictive Model for Irregular Wars Based on a Structured Historical Database.” Presentation at the 48th AORS Symposium, Working Group 5 – Irregular Warfare and Counterinsurgency, 14-15 October 2009
[7] Steven M. Goode, “A Historical Basis for Force Requirements in Counterinsurgency,” Parameters, Winter, 2009-10
[8] R. Royce Kneece, Jr., David A. Adesnik, Jason A. Dechant, Michael F. Fitzsimmons, Arthur Fries, and Mark E. Tillman, “Force Sizing for Stability Operations,” Arlington, VA, Institute for Defense Analysis, 2010
[9] Jeffery A. Friedman, “Manpower and Counterinsurgency: Empirical Foundations for Theory and Doctrine,” Security Studies, 20:556–591, 2011
[10] Jason Lyall and Isaiah Wilson III, “Rage against the Machines: Explaining Outcomes in Counterinsurgency Wars,” International Organization 63, no. 1, Winter 2009
Much has changed since James Quinliven kicked off discussion over manpower and counterinsurgency. One of the most significant differences is the availability now of useful collections of historical data for analysis.
![By Perhelion [CC0], via Wikimedia Commons](https://i0.wp.com/dupuyinstitute.org/wp-content/uploads/2016/01/1024px-Balanced_scale_of_Justice.svg_.png?resize=584%2C459&ssl=1)