¹ Valentino, Benjamin A., Huth, Paul, and Balch-Lindsay, Dylan, “Draining the Sea: Mass Killing and Guerrilla Warfare,” International Organization 58 (Spring 2004CrossRefGoogle Scholar). Long civil wars, however, tend to display relatively low levels of intensity. See Mukherjee, Shivaji, “Low Intensity, Long Duration Conflict: The Maoist Insurgency in India” (Manuscript, Yale UniversityGoogle Scholar,).

² Olson, Mancur, The Logic of Collective Action: Public Goods and the Theory of Groups (Cambridge:Harvard University Press, 1965Google Scholar); Tullock, Gordon, “The Paradox of Revolution,” Public Choice 11 (September 1971CrossRefGoogle Scholar).

³ For example, Jean Wood, Elisabeth, Insurgent Collective Action and Civil War in El Salvador (New York:Cambridge University Press, 2003CrossRefGoogle Scholar); Petersen, Roger D., Resistance and Rebellion: Lessonsfrom Eastern Europe (New York:Cambridge University Press, 2001CrossRefGoogle Scholar).

⁴ For example, Mursheb, S. Mansoob and Gates, Scott, “Spatial-Horizontal Inequality and the Maoist Insurgency in Nepal,” Review of Development Economics 9 (February 2005Google Scholar); Collier, Paul, “Economic Causes of Civil Conflict and their Implications for Policy,” in Crocker, Chester A., Hampson, Fen Osier, and Aall, Pamela, eds., Turbulent Peace: The Challenges of Managing International Conflict (Washington, D.C.:United States Institute of Peace Press, 2001Google Scholar). For an extensive review, see Sambanis, Nicholas, “Poverty and the Organization of Political Violence: A Review and Some Conjectures,” in Collins, Susan M. and Graham, Carol, eds., Brookings Trade Forum 2004 (Washington, D.C.:Brook-ings Institution, 2005Google Scholar).

⁵ Collier (fn. 4), 150.

⁶ It is implied, for instance, that low state capacity may take the form of occasional state incursions accompanied by indiscriminate violence that minimizes the rebels' collective action problem. In James Fearon and David Laitin's formulation: “Financially, organizationally, and politically weak central governments render insurgency more feasible due to weak local policing or inept and corrupt counterinsurgency practices. These often include a propensity for brutal and indiscriminate retaliation that helps drive noncombatant locals into rebel forces.” Fearon, and Laitin, , “Ethnicity, Insurgency, and Civil War,” American Political Science Review 97 (February 2003CrossRefGoogle Scholar).

⁷ Insurgency, defined as “a technology of military conflict characterized by small, lightly armed bands practicing guerrilla warfare from rural base areas,” has been the dominant type of civil war in the post-World War II era; Fearon and Laitin (fn. 6), 75. This definition outlines the scope conditions for our argument. On the one hand, we expect wars fought as insurgencies to conform to the logic we specify in this article; this includes all types of civil wars (for example, ethnic and nonethnic), as well as insurgencies against occupation regimes. On the other hand, we exclude conventional interstate or civil wars (for example, Spain 1936-39). Although civilians may be systematically targeted in conventional wars and though this targeting is likely indiscriminate (typically, aerial bombing), its logic and implications diverge from the type of violence common in insurgencies for a straightforward reason: while civilians subject to aerial bombing can be demoralized, they do not have the option of “switching” their support toward a rival actor, an option available in insurgencies where frontlines are often fluid. See Kalyvas, Stathis N., The Logic of Violence in Civil War (New York:Cambridge University Press, 2006CrossRefGoogle Scholar).

⁸ Tarrow, Sidney, Power in Movement: Social Movements and Contentious Politics (New York:Cambridge University Press, 1998CrossRefGoogle Scholar).

⁹ Tullock (fn. 2); Lichbach, Mark Irving, The Rebel's Dilemma (Ann Arbor:University of Michigan Press, 1995CrossRefGoogle Scholar); Popkin, Samuel L., The Rational Peasant: The Political Economy of Rural Society in Vietnam (Berkeley:University of California Press, 1979Google Scholar); Wood (fn. 3).

¹⁰ Collier (fn. 4), 150.

¹¹ Note, however, that (prewar) public order is also a collective good that individuals may prefer not to sacrifice. Failure to join an insurgency is a de facto contribution to the provision of this good. However, scholars have so far failed to give adequate consideration to the value of public order.

¹² Fearon, James D., “Why Do Some Civil Wars Last So Much Longer than Others?” Journal of Peace Research 41 (May 2004CrossRefGoogle Scholar).

¹³ Wood (fn. 3); Collier (fn. 4).

¹⁴ Gould, Roger V., Insurgent Identities: Class, Community, and Protest in Parisfrom 1848 to the Com mune (Chicago:University of Chicago Press, 1995), 204Google Scholar.

¹⁵ This is different from the distinction between onset and duration of civil war. In the empirical literature, onset is computed on the basis of large fatalities, a computation that assumes already substantial operational rebel organizations.

¹⁶ Elster, Jon, The Cement of Society: A Study of Social Order (Cambridge:Cambridge University Press, 1989CrossRefGoogle Scholar).

¹⁷ Collier (fn. 4), 143.

¹⁸ “Selective incentives” is the term used for the individual side payments that are dispensed by organizations to overcome the Olsonian logic of collective action. For evidence of the widespread existence of the private benefits that accrue to rebels, see Lichbach (fn. 9), 215–38.

¹⁹ Muller, Edward N. and Opp, Karl-Dieter, “Rational Choice and Rebellious Collective Action,” American Political Science Review 80 (June 1986CrossRefGoogle Scholar).

²⁰ Tilly, Charles, From Mobilization to Revolution (Reading, Mass.:Addison-Wesley, 1978Google Scholar); Brocket, Charles D., Political Movements and Violence in Central America (New York:Cambridge University Press, 2005CrossRefGoogle Scholar).

²¹ Wood (fn. 3).

²² Petersen (fn. 3).

²³ Tullock(fn. 2), 90.

²⁴ Gurr, Ted Robert, Why Men Rebel (Princeton:Princeton University Press, 1970Google Scholar); Muller, Edward N. and Weede, Erich, “Cross-National Variation in Political Violence: A Rational Action Approach,” Journal of Conflict Resolution 34 (December 1990CrossRefGoogle Scholar).

²⁵ Lichbach, Mark Irving and Gurr, Ted, “The Conflict Process: A Formal Model,” Journal of Conflict Resolution 25 (March 1981CrossRefGoogle Scholar); Mason, David T. and Krane, Dale A., “The Political Economy of Death Squads: Toward a Theory of the Impact of State-Sanctioned Terror,” International Studies Quarterly 33 (June 1989CrossRefGoogle Scholar). Quoting Greene, Lichbach conjectures that violence used by a government against its own citizens may be seen as arbitrary, which would tend to “lower the government's legitimacy and raise the society's revolutionary potential.” He concludes that, as a result, “the apathetic become politicized, the reformers become radicalized, and the revolutionaries redouble their efforts. Thus, when the government follows a policy of coercion, the policy itself may become the target of dissent by new challenging groups, thereby spreading conflict and engulfing the entire nation”; in short, repression “radicalizes” previous “free riders” to the revolution. In this formulation of the effects of repression, Li-chbach does not consider the violence of repression per se, which may alter an individual's cost-benefit calculation. See Lichbach, Mark Irving, “Deterrence or Escalation? The Puzzle of Aggregate Studies of Repression and Dissent,” Journal of Conflict Resolution 31 (June 1987), 269CrossRefGoogle Scholar.

²⁶ Goodwin, Jeff, No Other Way Out: States and Revolutionary Movements, 1944–1991 (New York:Cambridge University Press, 2001CrossRefGoogle Scholar); Lichbach (fn. 9); Wood (fn. 3); Fearon and Laitin (fn. 6), 75–76.

²⁷ For example Lichbach (fn. 26); Francisco, Ron, “Coercion and Protest: An Empirical Test in Two Democratic States,” American Journal of Political Science 40 (November 1996CrossRefGoogle Scholar); Moore, Will H., “Repression and Dissent: Substitution, Context, and Timing,” American Journal of Political Science 42 (July 1998CrossRefGoogle Scholar).

²⁸ Gutiérrez Saín, Francisco, “Criminal Rebels? A Discussion of Civil War and Criminality from the Colombian Experience,” Politics and Society 32 (June 2004Google Scholar).

²⁹ This empirical variation likely accounts for the contradictory results of recent microlevel studies that rely on higher-quality data: Lyall supplies evidence that indiscriminate state violence may be effective against Chechen rebels, whereas Do and Iyer conclude the opposite for Nepal. See Jason Lyall, “Disturbing Fire: A Randomized Evaluation of the Impact of Indiscriminate Violence on Insurgent Reprisals” (Paper presented at the annual meeting of the International Studies Association, Chicago, February 28-March 3, 2007); Do, Quy-Toan and Iyer, Lakshmi, “Poverty, Social Divisions and Conflict in Nepal” (Manuscript, World Bank, 2006Google Scholar).

³⁰ Wood (fn. 3), 8–10.

³¹ Quite obviously, this claim holds only for those who can choose whether or not to participate, not for those targeted during ethnic riots and pogroms.

³² While it is true that civilians are sometimes targeted systematically in conventional wars (in particular through area bombing), the relative impermeability of frontlines generally leaves them with no options in terms of participation. For this reason, we are unlikely to encounter the kind of endogenous, risk-based logic of participation contemplated here.

³³ Kalyvas (fn. 7).

³⁴ Brinkley, Douglas, “Tour of Duty: John Kerry in Vietnam,” Atlantic Monthly 292 (December 2003Google Scholar).

³⁵ As opposed to selective targeting based on tangible evidence about a specific individual's actions. See Kalyvas (fn. 7).

³⁶ For example, see Goodwin (fn. 26); Mason and Krane (fn. 25).

³⁷ See Kalyvas (fn. 7).

³⁸ Why should we assume this? Experience with the peacetime court systems of advanced industrial countries teaches that distinguishing guilt from innocence is an inexact science. In civil war the scale of the problem alone should increase the proportion of false positives. The logic of malicious denunciation leads us to anticipate an even higher proportion ofjunk intelligence in insurgencies. The selection mechanism we elaborate adds insult to injury.

³⁹ This discussion draws on Kalyvas (fn. 7).

⁴⁰ Valentino, Huth, and Balch-Lindsay (fn. 1).

⁴¹ Kalyvas, Stathis N., “The Paradox of Terrorism in Civil War,” Journal of Ethics 8 (March 2004Google Scholar).

⁴² Respectively: Condit, D. M., Case Study in Guerrilla War: Greece during World War II (Washington, D.C.:Special Operations Research Office, American University, 1961Google Scholar), 268; Browning, Christopher R., “Germans and Serbs: The Emergence of Nazi Antipartisan Policies in 1941,” in Berenbaum, Michael, ed., A Mosaic of Victims: Non-Jews Persecuted and Murdered by the Nazis (New York:New York University Press, 1990), 68Google Scholar; Kedward, H. R., In Search of the Maquis: Rural Resistance in Southern France, 1942–1944 (Oxford:Oxford University Press, 1993), 190Google Scholar.

⁴³ Mazower, Mark, Dark Continent: Europe's Twentieth Century (London:Allen Lane, 1998Google Scholar); Li, Lincoln, The Japanese Army in North China, 1937–1941: Problems of Political and Economic Control (Tokyo:Oxford University Press, 1975Google Scholar).

⁴⁴ Kalyvas (fn. 41).

⁴⁵ Two snapshots from El Salvador: when asked why he joined, an insurgent answered that he “had no choice. .. . It was a matter of survival. Those were the days when not to go meant getting killed.” Likewise, a woman said that “in this war, you don't get involved because you want to, but because you have to. Because if you don't, they kill you. Even though you didn't know anything about the war.” Quoted respectively in Lee Anderson, Jon, Guerrillas:Journeys in the Insurgent World (New York:Penguin, 2004), 222Google Scholar; and Viterna, Jocelyn, “Pulled, Pushed, and Persuaded: Explaining Women's Mobilization into the Salvadoran Guerrilla Army,” American Journal of Sociology 112 (July 2006), 24CrossRefGoogle Scholar.

⁴⁶ Aussaresses, Paul, Services Spéciaux Algérie 1955–1957: Mon temoignage sur la torture [Algeria Special Services 1955–1957: My Testimony on Torture] (Paris:Perrin, 2001), 62Google Scholar; Hayden, William, “The Kosovo Conflict: The Strategic Use of Displacement and the Obstacle to International Protection,” Civil Wars 2 (Summer 1999), 57Google Scholar.

⁴⁷ Wou, Odoric Y. K., Mobilizing the Masses: Building Revolution in Henan (Stanford, Calif.:Stanford University Press, 1994), 231Google Scholar; Wickham-Crowley, Timothy P., Exploring Revolution: Essays on Latin American Insurgency and Revolutionary Theory (Armonk, N.Y.:M. E. Sharpe, 1991), 43Google Scholar; Lansdale, Edward G., “Viet Nam: Do We Understand Revolution?” Foreign Affairs 43 (October 1964), 85CrossRefGoogle Scholar.

⁴⁸ Elliott, David W. P., The Vietnamese War: Revolution and Social Change in the Mekong Delta, 1930–1975 (Armonk and London:M. E. Sharpe, 2003), 873Google Scholar.

⁴⁹ Stoll, David, Between Two Armies: In the Ixil Towns of Guatemala (New York:Columbia University Press, 1993Google Scholar); Ivan Degregori, Carlos, “Harvesting Storms: Peasant Rondas and the Defeat of Sendero Luminoso in Ayacucho,” in Stern, Steve J., ed., Shining and Other Paths: War and Society in Peru, 1980–1995 (Durham and London:Duke University Press, 1998Google Scholar); Tishkov, Valery, Chechnya: Life in a War-Torn Society (Berkeley:University of California Press, 2004CrossRefGoogle Scholar); Lyall (fn. 29).

⁵⁰ Kalyvas (fn. 7).

³¹ For instance, a U.S. marine was given a sentence of eight years for killing an civilian, Iraqi“who was known to support the American occupation”; New York Times, February 19, 2007, A6Google Scholar.

⁵² Binford, Leigh, The El Mozote Massacre: Anthropology and Human Rights (Tucson:University of Arizona Press, 1996), 115Google Scholar.

⁵³ Svolos, Alexandras, Andartis sta vouna tou Moria: Odoiforiko (1947–49) [Guerrilla in the Mountains of Moria: A Journey (1947–49)] (Athens:Self-published, 1990), 22Google Scholar.

⁵⁴ Papakonstantinou, Michalis, To chroniko tis megalis nichtas [Chronicle of the Long Night] (Ath ens:Estia, 1999), 313Google Scholar.

⁵⁵ Johnson, Ralph William, Phoenix/Phung Hoang:A Study of Wartime Intelligence Management (Ph.D. diss., American University, 1982Google Scholar).

⁵⁶ Thayer, Thomas C., War without Fronts: The American Experience in Vietnam (Boulder, Colo.:Westview Press, 1985), 208Google Scholar. Note that the intended targets of the Phoenix Program were not full-time combatants but rather were clandestine operatives.

⁵⁷ Hernngton, Stuart A., Stalking the Vietcong: Inside Operation Phoenix-A Personal Account (No-vato, Calif:Presidio Press, 1997), 69Google Scholar.

⁵⁸ Johnson (fn. 55), 307.

⁵⁹ National Archives and Records Administration (NARA), National Police Infrastructure Analysis Subsystem-II (Records Group 330. Accession no. 3–349–79–002-D).

⁶⁰ It is important to keep in mind that the Phoenix Program counted all of the individuals on the list as Vietcong agents. It is our assumption, based on a reasonable understanding of insurgency processes, that many of these individuals were incorrectly identified. Our task is to determine how many.

⁶¹ Note, as well, that we have no way of ascertaining the effect of this perverse selection on insurgent recruitment. Kalyvas (fn. 7) suggests that a key variable in the efficacy of selective violence is not accurate selection per se but, rather, is the perception among the population that targeting is based on accurate selection. Such analysis would require data currently not available in the case of the Vietnam War.

⁶² According to NPIASS-II, Phoenix was aware of the “current address” of nearly 64 percent of confirmed persons but less than 1 percent of unconfirmed persons. Likewise, nearly 23 percent of confirmed persons but less than 1 percent of unconfirmed persons were the subjects of individualized arrest warrants. These data strongly suggest that significantly greater effort was oriented toward those under higher suspicion.

⁶³ p_vn stands for “proportion of Vietcong neutralized” and p_in stands for “proportion of innocents neutralized.”

⁶⁴ There is some disagreement in the methodological literature over the use of the odds ratio to express the size of effects. Note that this issue has no bearing on our analysis, because we do not attempt to use the odds ratio as an absolute measure of anything. We merely equate the odds ratios of two different comparisons.

⁶⁵ Why use killed + captured instead of just killed? Both capture and assassination require that an individual be physically located and identified by forces sufficient to take action against him. Still, this is a conservative assumption. Since defection was a good way to avoid being killed or captured (and sentenced to prison), we might expect innocent but threatened individuals to take this option at higher rates than would highly committed Vietcong agents, essentially shrinking the pool of potential unconfirmed victims faster than the pool of potential confirmed victims.

⁶⁶ Thayer (fn. 56). As an important Defense Department analyst during the Vietnam War, Thayer had access to all the data we use here, as well as to much that remains classified. His book does not consider the possibility we suggest, perhaps because it was simply inconceivable to him that the Phoenix Program was not merely unproductive but was actually counterproductive. Although it is true that the majority of people “neutralized” by Phoenix were believed to be low ranking, this result follows straightforwardly from the pyramidal form of any military or bureaucratic organization.

⁶⁷ Indeed, the files of captured persons were updated an average of 1.3 times, much more often than were the files of those who were killed (0.33 times), who rallied (0.33), or who remained at large (0.59).

⁶⁸ By “extreme” we mean the two solutions with the largest and smallest number of innocents victimized. In this context there are various ways to conceive of an intermediate solution. We chose a solution in which the number of innocents versus Vietcong victimized is intermediate between the extremes.

⁶⁹ Models 1 and 2 differ only in their operationalization of age. Model 1 collapses age into four categories (under 18,18–31, 32–44, 45 and older). Model 2 uses a binary indicator for “military age,” equal to 1 if a person was 18–44 years old, 0 for those both younger and older. We focus our discussion on model 2.

⁷⁰ Jones, Adam, “Gendercide and Genocide,” Journal of Genocide Research 2 (June 2000CrossRefGoogle Scholar)

⁷¹ Note also that there is evidence of sex and age profiling for selection into the Phoenix list as a whole: over 70 percent of the persons on the list were of military age; over 75 percent were men.

⁷² Thayer (fn. 56).

⁷³ Elliott (fn. 48), 1137.

⁷⁴ “Certain CF military intelligence officers told the ICRC that in their estimate between 70% and 90% of the persons deprived of their liberty in Iraq had been arrested by mistake. They also attributed the brutality of some arrests to the lack of proper supervision of battle group units.” See International Committee of the Red Cross (ICRC), Report ofthe International Committee of the Red Cross (ICRC) on the Treatment by the Coalition Forces of Prisoners of War and Other Protected Persons by the Geneva Conventions in Iraq during Arrest, Internment and Interrogation (Geneva:ICRC, 2004), 8Google Scholar. Even the highly selec tive process by which individuals were shipped to Guantanamo appears to have suffered from similar problems: it turns out that 92 percent of the 517 Guantanamo detainees were not al-Qaeda fighters, while 95 percent of them were not captured by the Americans themselves; some 86 percent were handed over in Afghanistan and Pakistan after a widespread campaign in which big financial bounties were offered in exchange for anyone suspected of links to al-Qaeda and the Taliban. Simpson, John, “No Surprises in the War on Terror,” BBC News, February 13, 2006Google Scholar, http://news.bbc.co.uk/2/hi/ middle_east/4708946.stm (accessed February 14,2006).

⁷⁵ One indication that local politics played a key role in the Phoenix Program stems from our knowledge of the sources of information. The intelligence used to identify Vietcong agents was attributed in nearly 56 percent of all cases to the Regional Forces, Popular Forces, or Civilian Irregular Defense Group militias, rather than to police or regular military forces of the U.S. or South Vietnam.

⁷⁶ Moyar, Mark, Phoenix andthe Birds ofPrey: The CIA'S Secret Campaign to Destroy the Viet Cong napolis:Naval Institute Press, 1997), 114Google Scholar.

⁷⁷ Ibid., 116.

⁷⁸ Ibid., 293.

⁷⁹ Binford (fn. 52), 107.

⁸⁰ Moyar (fn. 76), 122.

⁸¹ Elliott (fn. 48), 947.

⁸² Quoted in Maass, Peter, “The Way of the Commandos,” New York Times Magazine, May 1, 2005, 47Google Scholar.

⁸³ The rosters of the local prison in the town of Nafplio show over one thousand individuals held there during the same period. Hundreds more were sent to a concentration camp in the neighboring town of Konnthos, while a smaller but unspecified number were sent to slave labor camps in Germany. However, the Germans did not take rebel prisoners. Indeed, a significant proportion of the rebels killed in action were shot after being captured. A common practice of German occupation troops was to list most civilian victims as rebel fatalities. See H. F. Meyer, Von Wien nach Kalavryta: Die blutige Spur der 117:Jdger-Division durch Serbien und Griechenland [From Vienna to Kalavryta. The Bloody Trail of the 117 Jaeger Division through Serbia and Greece] (Moehnesee: Bibliopolis, 2002). Careful disaggregation based on extensive archival and field research confirms the intuition of many historians that most of these “partisan fatalities” were in fact civilians fleeing the German advance.

⁸⁴ On top of the twenty fatalities suffered during the occupation, the regiment lost fifteen more local fighters during the battle of Athens (December 1944), when the communists attempted to seize power. They are not included in the analysis since violence against civilians subsided after the end of the occupation. Our source on rebel fatalities is Vazeos, Emmanouil, Ta agnosia paraskinia tis Ethnikis Antistaseos eis tin Pehponnison [The Unknown Backstage of the National Resistance in the Pelopon-nese] (Korinthos:Self-published, 1961Google Scholar). The 6^th Regiment of ELAS, which was active in the Argolid and Korinthia areas and recruited primarily from these two regions, reached thirty-five hundred men in October 1944, after the occupation's end; many of these men were recruited or conscripted after the German evacuation. The regiment's full force prior to this was closer to five hundred men. Data on militiamen were collected from archival sources and civil registries.

⁸⁵ Stoll (fn. 49).

⁸⁶ In light of this analysis it should not comes as a surprise that mass displacement is so common in civil wars. Nevertheless, it is rarely the first choice of civilian populations: rural populations depend on land for their livelihood and abandon their villages only under tremendous pressure. Flight is not a form of free riding. It carries substantial costs and is often not an option for military-age men who may be shot attempting to flee (hence the predominance of women, children, and the old in refugee camps).

⁸⁷ Tullock (fn.2),93.

⁸⁸ McNeill, William H., The Greek Dilemma: War and Aftermath (Philadelphia:J. B. Lippincott, 1947), 80–81Google Scholar.

⁸⁹ An implication is that club goods should be common across all types of armed groups. Jeremy Weinstein has suggested instead that only well-funded rebel organizations recruit via club goods (mainly loot). Consequently, they attract opportunistic and undisciplined individuals who abuse the civilian population. Aside from the assumption that individuals joining an armed group to acquire club goods cannot be socialized to become motivated and disciplined combatants later, this argument requires a key condition: that the state facing the rebels must be exceedingly weak or even nonexistent. Otherwise, an army with these characteristics has little chance of surviving. See Weinstein, Jeremy M., Inside Rebellion: The Politics of Insurgent Violence (New York:Cambridge University Press, 2007Google Scholar).

⁹⁰ Why, then, would individuals join “go nowhere” insurgencies? An answer is that they are attracted by access to club goods.