Why does #terrorism   so rarely succeed in policy goals compared to guerrilla warfare, and trigger backfire effects? Because use of terrorist tactics triggers the horns effect; given the same hypothetical, they are perceived as fundamentally hateful, unable to keep negotiated commitments, & intrinsically preferring violence. From "The Credibility Paradox: Violence as a Double-Edged Sword in International Politics", Abrahms 2013:

"Implicit in the rationalist literature on bargaining over the last half-century is the political utility of violence. Given our anarchical international system populated with egoistic actors, violence is thought to promote concessions by lending credibility to their threats. From the vantage of bargaining theory, then, empirical research on terrorism poses a puzzle. For nonstate actors, terrorism signals a credible threat in comparison with less extreme tactical alternatives. In recent years, however, a spate of studies across disciplines and methodologies has nonetheless found that neither escalating to terrorism nor with terrorism encourages government concessions. In fact, perpetrating terrorist acts reportedly lowers the likelihood of government compliance, particularly as the civilian casualties rise. The apparent tendency for this extreme form of violence to impede concessions challenges the external validity of bargaining theory, as traditionally understood. In this study, I propose and test an important psychological refinement to the standard rationalist narrative. Via an experiment on a national sample of adults, I find evidence of a newfound cognitive heuristic undermining the coercive logic of escalation enshrined in bargaining theory. Due to this oversight, mainstream bargaining theory overestimates the political utility of violence, particularly as an instrument of coercion.

Why might escalation fail to produce superior, even commensurate gains when it boosts the credibility of a threat? Almost in passing, Schelling provides a potential clue. He mentions that for coercion to work, the challenger must signal not only a credible threat to inflict pain when concessions are withheld, but also a credible promise to remove the pain in the event concessions are forthcoming. Otherwise, no incentive exists for complying with the demands (Schelling 1966:75-76). 3 Schelling and his rationalist disciples analyze escalation, however, only in terms of its positive effect on the threat -not the potentially negative effect on the promise. Neglected is what I call the Credibility Paradox: In international politics, the very escalatory acts that add credibility to a challenger's threat can subtract credibility from his promise.

The Credibility Paradox is due to the Correspondence of Means and Ends bias, a newfound cognitive heuristic in international affairs substantiated via an experiment. Contrary to the strategic choice literature underlying mainstream bargaining theory, the experiment indicates that defenders are apt to infer the extremeness of a challenger's preferences directly from the extremeness of his tactics notwithstanding the nature of his actual demands. Because of this human tendency to confound the extreme means of the challenger with his presumed ends, escalation can discredit his vow to remove the pain regardless of whether the defender were to accommodate his demands.

In the third section, I present the anomaly-mounting empirical evidence across disciplines and methodologies that escalating to terrorism or with terrorism actually decreases the odds of bargaining success, despite elevating the threat to the targets of the pressure. In the fourth and fifth sections, I describe my theory to account for the anomalous empirics and test it with an experiment embedded in a survey on a large national sample of American adults. Fielded by the survey research firm YouGov/Polimetrix, the experiment demonstrates how escalating to terrorism or with terrorism adds credibility to a challenger's threat and yet subtracts credibility from his promise to ever remove it, undermining support for concessions. In the sixth section, I demonstrate the generalizability of the Credibility Paradox beyond the tactic of terrorism

For nonstate challengers, employing terrorism is unquestionably a credible signal of resolve based on the standard arguments in bargaining theory. Perpetrating terrorist acts is costly in blood and treasure compared to relying on less extreme tactical alternatives. In her historical investigation of protest, Chenoweth and Lawrence find "The likelihood of being killed while carrying out one's duties as an armed insurgent is high, whereas many lower-risk tactics are available to participants in a nonviolent resistance campaign" (Chenoweth and Lawrence 2010:257). So accepted are the relative dangers of terrorism that scholars commonly analyze its usage as a collective action problem (for example, Wood 2003). Clearly, the high likelihood of expending terrorist members reveals their own commitment. Yet it also exhibits that of the larger organization from which they hail-one evidently prepared to sacrifice not only critical manpower, but the very cadres whose resolve would have made them valuable in other key roles (Berman and Laitin 2008). Such determined members are always in precious supply regardless of what leaders of these groups may say. Gould captures the essence of this point: "While activists might have little trouble persuading a casual acquaintance to sign a petition, they would have great difficulty convincing such a person to risk injury, death, or imprisonment" (1995:204). The moral repugnance of killing civilians drains the pool of potential terrorists, adding to the costs of losing even one (Chenoweth and Lawrence 2010). And naturally, the costs of utilizing terrorism include financial ones in terms of both conducting operations and bearing the response. In addition to these sunk costs from using terrorism, Mueller's (2006) research details how adopting this tactic leaves something to chance by provoking capricious government overreactions. The historical record is replete with aggrieved parties escalating to or with terrorism, aware that the additional pain to the target would boost the odds of paying a prohibitively steep price. Afghan tribal leaders, for instance, warned Osama bin Laden against committing the September 11 attacks due to uncertainty over the costliness of the American reaction (9/11 Commission 2004:251). Finally, nonstate actors that employ terrorism are manifestly willing to countenance the costs from other parties as well, viz. audience costs. Perhaps more than any other tactic, terrorism offends constituencies beyond the target of the pressure (Hultman 2005; Laitin and Shapiro 2008). As Weinstein notes: "Undoubtedly, groups that deploy violence against noncombatants incur significant costs in consequence. Indiscriminate violence...damages the reputation of the group both within the country and outside of it" (2007:206). In fact, such post-attack losses of support are among the most common ways for terrorist groups to end (Cronin 2009).

Terrorism is a "weapon of the weak," but only in the sense that its practitioners are nonstate actors and therefore less capable than their government foes. In a recent review, Fortna (2012) finds no empirical evidence that weaker groups are prone to terrorism. Consistent with bargaining theory, she shows that the strongest rebel groups in civil wars are the most likely to engage in terrorism by attacking the population. Similarly, Asal and Rethemeyer (2008) demonstrate that membership size and other organizational resources are significant predictors of terrorist lethality. A case study on al-Qaida illustrates this point by detailing how its production of terror peaked with organizational capacity (Eilstrup-Sangiovanni and Jones 2008).

For decades, terrorism specialists have expressed skepticism that attacking civilians helps nonstate actors to achieve their strategic demands. In the 1970s, Laqueur (1976) published a paper entitled "The Futility of Terrorism," in which he asserted that terrorist groups seldom attain their political platforms. In the 1980s, Crenshaw also observed that terrorists do not obtain their given political ends, and "Therefore one must conclude that terrorism is objectively a failure" (1988:15). Similarly, RAND remarked, "Terrorists have been unable to translate the consequences of terrorism into concrete political gains... It is a fundamental failure" (Cordes, Hoffman, Jenkins, Kellen, Moran and Sater 1984:49). Since the September 11, 2001, attacks, several large-n observational studies have offered a firmer empirical basis. These reveal that only a handful of terrorist groups have managed to accomplish their political platforms (Abrahms 2006; Jones and Libicki 2008; Cronin 2009). All of the authors conclude that terrorism does not encourage concessions. Abrahms (2006) shows that none of the politically successful groups in his sample relied on terrorism as the primary coercive tactic. Jones and Libicki note that in the few cases in which terrorist groups have triumphed, terrorism had "nothing to do with the outcome" (2008:33). And Cronin concludes that the victorious achieved their demands "despite the use of violence against innocent civilians [rather] than because of it," and that "The tactic of terrorism might have even been counterproductive" (2009:203). Hard case studies have inspected the limited historical examples of clear-cut terrorist victories, determining that these salient events were idiosyncratic, unrelated to the harming of civilians, or both (for example, Neumann and Smith 2007; Rose, Murphy and Abrahms 2007; Cronin 2009; Dannenbaum 2011)

Gaibulloev and Sandler analyze a data set of international hostage crises from 1968 to 2005. They exploit variation in whether the perpetrators escalate by killing the hostages. The study finds that hostage-takers significantly lower the likelihood of achieving their demands by inflicting physical harm in the course of the standoff. The authors conclude that terrorists gain bargaining leverage from restraint, as escalating to "bloodshed does not bolster a negotiated outcome" (2009:19). There is no research consensus on whether the definition of terrorism requires the violent act to inflict physical pain. Terrorism data sets often include all hostage incidents in which an aggrieved person or group issues a demand. Yet many scholars count an act as terrorism only in the event that a measure of physical harm is inflicted, usually in the form of a civilian death (for example, Horowitz 2010). Those logistic regression results may therefore be interpreted as showing that either escalating to terrorism or with terrorism hinders bargaining success. Chenoweth and Stephan (Stephan and Chenoweth 2008; Chenoweth and Stephan 2011) provide additional empirical evidence that meting out pain hurts nonstate actors at the bargaining table. Their studies compare the coercive effectiveness of 323 violent and nonviolent resistance campaigns from 1900 to 2006. Like Gaibulloev and Sandler, the authors find that refraining from bloodshed significantly raises the chances of government accommodation even after correcting for selection issues. Chenoweth and Stephan employ an aggregate measure of violence that includes both indiscriminate attacks on civilians and discriminate attacks on military personnel or other government officials, which are frequently differentiated from terrorism as insurgent, guerrilla, or militant attacks (for example, Ganor 2002). Other statistical research suggests that when terrorist attacks are combined with such discriminate violence, the bargaining outcome is not additive; on the contrary, the pain to the population significantly decreases the odds of government concessions (Abrahms 2012; Fortna 2012), especially against democracies (Abrahms 2007). Getmansky and Sinmazdemir (2012) find that the Israeli government in particular is significantly less likely to cede land to the Palestinians when they have perpetrated terrorism.

Without exception, this research shows that terrorism does not intimidate citizens of target countries into supporting more dovish politicians. Quite the opposite, terrorism systematically raises popular support for right-wing leaders opposed to concessions. In a couple of statistical papers, Berrebi and Klor (2006, 2008) find that terrorist fatalities within Israel significantly boost local support for anti-accommodation right-bloc parties, such as the Likud. Other quantitative work demonstrates that the most lethal terrorist incidents in Israel are the most likely to induce this rightward electoral shift. The authors conclude that heightening the pain to civilians tends to "backfire on the goals of terrorist factions by hardening the stance of the targeted population" (Gould and Klor 2010:1507). These trends appear to be the international norm, not Israel-specific. Chowanietz (2010) analyzes variation in public opinion within France, Germany, Spain, the United Kingdom, and the United States from 1990 to 2006. In each target country, terrorist attacks have shifted the electorate to the political right in proportion to their lethality. More anecdotally, similar observations have been registered after mass casualty terrorist attacks in Egypt, Indonesia, Jordan, the Philippines, Russia, and Turkey (for example, Wilkinson 1986:52; Mueller 2006:184, 587). In a summary of the literature, RAND concludes: "Terrorist fatalities, with few exceptions, increase support for the bloc of parties associated with a more intransigent position. Scholars may interpret this as further evidence that terrorist attacks against civilians do not help terrorist organizations achieve their stated goals" (Berrebi 2009:189-190). Psychologists are replicating these results in laboratory experiments, further ruling out the possibility of a selection effect (for example, Pyszczynski, Rothschild and Abdollahi 2008). Perhaps unsurprisingly, the most notorious rebel leaders in modern history, from Abdullah Yusuf Azzam to Regis Debray, Vo Nguyen Giap, Che Guevara, and Carlos Marighela admonished their foot-soldiers against targeting the population because the indiscriminate violence was proving counterproductive (Wilkinson 1986:53, 59; Rapoport 2004:54-55).

Research on terrorism and violent crime is surprisingly distinct, but citizens of target countries may be subject to the same biases as victims of assault, battery, robbery, and other violent offenses. Baumeister finds that victims of violent crime characteristically suffer from "The Myth of Pure Evil," a cluster of interrelated biases about the intentions of the perpetrator. In reality, the modal perpetrator is motivated by specific grievances; he victimizes the target as a means to an end rather than as an end in itself. Because his violence is instrumental, he will usually desist upon obtaining his demands or discovering alternative avenues to achieve them (Baumeister 1999:101-103, 127). From the standpoint of the victim, however, the means of the perpetrator are endogenous to his desired ends. The attacker is ostensibly motivated not to achieve his particular demands, but by the sadistic pleasure of harming the victim (Baumeister 1999:71, 90). Baumeister's research on violence extends the work of other psychologists (for example, Darley 1992).

In the next section, I examine whether citizens of target countries likewise view the extreme means of terrorists as endogenous to their ends. Specifically, I test for evidence of the Correspondence of Means and Ends bias, a new cognitive heuristic in international affairs.

The survey research firm, YouGov/Polimetrix, fielded my experiment over the Internet in May 2010 on a large, national sample of voting-age American citizens. By using random digit dialing to recruit participants and providing free Internet access to households that lack it, Polimetrix can administer survey instruments online to nationally representative samples. This technique is now standard in survey experiments, as the samples generated tend to surpass in quality those from conventional telephone polling (Tomz 2008). The empirical strategy is propitious for both methodological and substantive reasons. First, it enables the investigator to vary the extremeness of the challenger's tactics in a controlled environment unencumbered by selection issues, facilitating precise assessments of how the tactical change alone affects perceptions of the actor's intentions. Second, the subjects were members of the most relevant constituency for assessing the coercive value of terrorism. Terrorists target democracies to convince electorates that accommodating their demands is cheaper than resisting them (Pape 2003). If, however, citizens of target countries tend to conclude that perpetrators of terrorism are perforce unappeasable extremists, then the violence would create a credible commitment problem, empower leaders opposed to compromise, and thereby lower the odds of government compliance.
All subjects were presented with a simple vignette of an unidentifiable group issuing a traditionally moderate preference through the American media-the release of its imprisoned leaders from US custody in exchange for permanently demobilizing. Subjects were randomly assigned, however, to two conditions that differed along a tactical dimension. In the control condition, the group surrounds a bunch of American civilians, takes them hostage, but does not physically harm anyone in the course of the confrontation. The same information was presented in the treatment condition, except the group escalates tactically by killing the civilians in its custody. To minimize framing issues, I paid attention to the formal aspects of the instrument by avoiding any derivatives of the word "terror" or any other emotive labels to describe either the coercive acts or the actors themselves. The two conditions were thus duplicates, except that in the painful treatment, the moderate group adopts a more extreme method by killing the civilians instead of releasing them unharmed. Subjects in both conditions were presented with a series of identical multiple choice and ordinal scale questions designed to assess both directly and indirectly the perceived extremeness of the group's preferences. Specifically, all subjects were asked the following set of questions: (i) to evaluate whether the group is motivated to achieve its demand of freeing the imprisoned leaders in US custody or to harm Americans out of hatred toward them; (ii) to rate the group's preferences from 1 to 7 along this continuum; 5 (iii) to judge whether the group would in fact demobilize upon achieving its demand to free the imprisoned leaders; (iv) to appraise whether the group would derive satisfaction from Americans physically harmed in an unrelated incident that would not contribute to winning back the imprisoned leaders in US custody; and (v) to ascertain whether the group would continue to engage in the same actions against Americans even after discovering a less extreme method that promised to free the imprisoned leaders (Appendix 1).
Following convention in experimental research, I then applied a two-tailed difference of means test to determine whether the tactical manipulation alone yields significant variation in the perceived extremeness of the self-described moderate group's preferences. Answers to each of the five questions strongly confirm the Credibility Paradox and are statistically significant at the .01 level or better (Table 1 and Figure 1). Compared to subjects in the control condition in which no civilians were physically harmed, those exposed to the painful treatment were on average:
(i) 27% more likely to believe the group is motivated not to free the imprisoned leaders in US custody, but to harm Americans out of hatred toward them; (ii); 20% more likely to rate the group's preferences as the most extreme on a standard 7-point ordinal scale; 6 (iii) 23% more likely to believe the group would not demobilize upon achieving its demand to free the imprisoned leaders; (iv) 33% more likely to believe the group would derive satisfaction from Americans physically harmed in an unrelated incident that would not contribute to winning back the imprisoned leaders; and (v) 22% more likely to believe the group would continue to engage in the same actions against Americans even after discovering a less extreme method to free its leaders from
US custody.

This is the first controlled experiment on the mechanism of coercion. A concern inherent to this methodology is the sacrificing of external validity for precision. The results appear externally valid, however, in the following ways. First, the vignettes in the experiment are not based on hypothetical scenarios. Each tracks closely with the most common international events from a leading data set on nonstate coercion. International Terrorism: Attributes of Terrorist Events, 1968-2005 (ITERATE 5) contains detailed information from over a thousand international hostage incidents between 1968 and 2005. By far the two most common demands issued are for governments to cede money or prisoners, which occurred in 16% and 11% of the cases, respectively. The perpetrators demand prisoners in vignette 1 and money in vignette 2 in accordance with both the relative frequency of these moderate demands and the definition of terrorism itself as an extremism of means though not necessarily ends (Oberschall 2004:26). 7
Second, case studies confirm that the publics of target countries are in fact prone to inferring the extremeness of challengers' preferences directly from their tactics, empowering anti-accommodation hardliners in the face of terrorism. Within Israel, polls show that respondents who perceive the tactics of Palestinians as "mainly violent" are more likely to believe their intent is to "destroy Israel." Conversely, respondents who perceive the tactics of Palestinians as "mainly nonviolent" are more likely to believe their intent is merely to "liberate the occupied territories" (Kaufman 1991). 8 Consistent with the Credibility Paradox, Palestinian terrorism thus erodes Israeli popular support for the peace process, strengthens hawkish leaders opposed to a two-state solution, and lowers the likelihood of government concessions (for example, Berrebi and Klor 2006). The Russian public also infers the extremeness of Chechen preferences directly from their tactics, fortifying hardliners against terrorist appeasement. Before the terrorism commenced, Russians favored granting an independent Chechen state. When terrorism erupted in the late 1990s, however, the Russian public concluded that the Chechens were apparently bent on harming it, shifting popular support away from concessions, while bolstering Vladimir Putin to instead bomb Grozny (Abrahms 2006). Al-Qaida's stated grievances about US-Middle East policies fell on deaf ears for the same reason. Bin Laden and his associates stressed that the purpose of the September 11, 2001, attack was to coerce the United States into withdrawing from the Middle East (for example, Clarke 2004). Polls show, however, that most Americans thought the point was "to harm them" as an end in itself (Harris Poll, 19-24 September), a perception that facilitated George W. Bush's counterterrorism response in the Arab-Muslim world (see Pronin, Kennedy and Butsch 2006).
Third, studies of media coverage of terrorism confirm that it struggles to amplify the political demands of the perpetrators. In their content analysis of terrorism articles in the New York Times and Times of London, Kelly and Mitchell find that "Less than 10% of the coverage in either newspaper dealt in even the most superficial way with the grievances of the terrorists" (1981:287). Terrorists struggle to broadcast their demands even when the perpetrators emphasize them, are highly educated, and speak the majority language of the target country. As Schmid and De Graaf (1982:111) illustrate via the Weather Underground, "The terrorists could bomb their names on to the front pages, but they could do next to nothing to make sure that the message intended by their bombings was also the message transmitted." Cordes et al. (1984:1) observe that "Although terrorism is often described as a form of communication, terrorists are rather poor communicators," as "The violence of terrorism is rarely understood by the public". Hewitt notes that instead of clarifying their preferences, perpetrators of terrorism are seen as engaging in "senseless bestiality" (1993:52), similar to how victims of violent crime view the motives of their attackers (Baumeister 1999). Although most content analyses have a Western bias, the Correspondence of Means and Ends bias does not appear to be culturally contingent. 9

Pape (2005) shows that even the most lethal terrorists express surprisingly moderate political aims.

Similarly, Lebow's case study on the Vietnam War reveals that the indiscriminate bombings and roundups failed to communicate the costs of resisting American demands; instead, the bloodshed inadvertently signaled to the Vietnamese people that the Johnson administration was bent on "hostility and convince[d] its target that it will be the victim of aggression, now and in the future, unless it stands firm" (1996:562). 10 This undiagnosed inference is sometimes modeled for extreme tactics in addition to terrorism. In a prominent application, terrorists bait governments into attacking the local population in order to exhibit their hostility toward it and thus inability to credibly bargain (Lake 2002)

challenger states also struggle to convey their intentions. In Conflict among Nations, Snyder and Diesing find in their sample of 181 escalatory threats that only 40% were "correctly interpreted" by the target. The authors underscore how these empirics are problematic for bargaining theory, as "To treat all these characteristics of crisis bargaining merely as deviations from an ideal of perfect rationality would mean that the norm itself would be lost" (1977:237). Case studies of nineteenth- and twentiethcentury warfare supply more detailed empirical evidence that the intent was not automatically received, leading Lebow (1989), Snyder (1989), Jervis et al. (1989), and others to conclude that such misperceptions are the central weakness of bargaining theory. Second, the political psychology literature suggests that in conflicts between states, defenders are also known to overrate the extremeness of their adversaries' intentions, impeding cooperation (for example, Larson 1997)."

#psychology