HLSS522Wk1

Regular Articles

Breaking taboos: Why insurgents pursue and use CBRN weapons

Victor Asal

University at Albany, SUNY

Nazli Avdan

University of Kansas

Gary Ackerman

University at Albany, SUNY

Abstract

Much ado has been made about the specter of unconventional weapons in the hands of militant groups. Despite the alarmism of the policy community, the pursuit of chemical, biological, radiological, and nuclear capabilities by non- state actors is rare. What explains why some violent non-state organizations pursue chemical, biological, radiological, and nuclear weapons while most do not? Using new data on organizational behaviors and attributes from the Big Allied and Dangerous 2 Insurgency project, we analyze 140 insurgent actors, from 1998 to 2012. We expand previous work by Asal, Ackerman & Rethemeyer by studying the phenomenon longitudinally and attending to organizational behavior rather than only to structural and environmental factors. We find that organizations that attack cultural sites are more likely to pursue and use chemical, biological, radiological, and nuclear weapons. We interpret the latter as a willingness to transgress upon accepted mores, for the sake of sensationalism and shock value, which coheres with a willingness to pursue unconventional weapons. Our results reflect that chemical, biological, radiological, and nuclear pursuit and use is part of a larger strategy for some violent non-state organizations, one predicated on generating mass fear and exacting a psychological toll.

Keywords

CBRN weapons, insurgent groups, non-state groups, terrorism, WMD

Introduction

In 2018, the White House released a strategy calling for ‘new aggressive steps to counter the growing threat posed by WMD terrorism’ (White House, 2018), a concern that has been echoed in almost all United States national strategy documents post-911. In fact, all administrations since Clinton have aimed to dampen the risk of so-called Weapons of Mass Destruction (WMD) terrorism (Mauroni, 2018). The specter of chemical, biological, radiological, and nuclear (CBRN) weapons in the hands of terrorists and other violent non-state actors (VNSAs) also figured in debates and prognostications about the possible resurgence of ISIS. To some extent, this concern

has been shared internationally, as evidenced by such multilateral efforts such as United Nations Security Council Resolution 1540 and the Global Initiative to Combat Nuclear Terrorism.

The alarmism of the security establishment may, however, seem puzzling in light of the empirics. On the nuclear weapons side of things, in the last 25 years, there have only been around 20 incidents where stolen weapons-grade nuclear material was seized by terrorist groups (Bunn & Roth, 2017). Depending on how one

Corresponding author: [email protected]

Journal of Peace Research 2023, Vol. 60(2) 193–208 ª The Author(s) 2022 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/00223433211057021 journals.sagepub.com/home/jpr

measures such phenomena, there have also only ever been a handful of truly mass-casualty terrorist attacks using chemical or biological agents, including the 1984 dissemination of food-borne Salmonella typhimurium, which sickened 751 people in Oregon (Török et al., 1997; Carus, 2000) and the attack by the Japanese Aum Shinrikyo cult with the nerve agent sarin in 1995 (Dan- zig et al., 2012). Indeed, of all 181,691 terrorist inci- dents recorded in the Global Terrorism Database (GTD) from 1970 to 2017, only 433 incidents (0.24% of all incidents) utilize CBRN technology (GTD, 2019).

Yet, while the employment of CBRN technology by VNSAs is a low-probability event, its consequences

could be so devastating that the scenario arguably justi- fies hypervigilance on the part of the security community (Bunn & Roth, 2017). Additionally, the broader pat- terns displayed in the GTD miss plots to use CBRN weapons. For example, there are estimates of 40 cases of radiological materials pursuit by non-state actors (Campbell & Murdie, 2018; James Martin Center, 2016), more than 60 high-threat attacks or plots against nuclear facilities (Ackerman & Halverson, 2016), and several expressions of interest by terrorist groups in acquiring nuclear weapons. For the period of 1990 to 2017, the POICN Database records a total of 224 uses and 517 total cases of CBRN pursuit (Binder &

Figure 1. CBRN events, 1990–2017, N ¼ 517

Figure 2. CBRN uses, 1990–2017, N ¼ 224

194 journal of PEACE RESEARCH 60(2)

Ackerman, 2021). Moreover, Figures 1 and 2 here show that, although CBRN cases have diminished since a peak in the latter 1990s, there are still periodic cases of both pursuit and use.

Our article offers several interlinked contributions to the state of knowledge in this area. Our analysis high- lights that the majority of CBRN attacks tend to be smaller scale and not highly lethal. Nonetheless, we maintain that the preoccupation with event magnitude may mask the psychological impetus to seek out uncon- ventional weapons simply for the sake of sowing fear and intimidation. Indeed, our article is timely insofar as the global coronavirus pandemic has brought the potency of biological weapons into sharp relief. While SARS COV- 2 is a natural disaster, the economic devastation wrought by the outbreak may animate VNSAs to ramp up their pursuit of biological weapons, or more generally CBRN technology. The ineptitude of even economically devel- oped states to stymie the spread of the virus has exposed the numerous weaknesses of global public health sys- tems. The allure of paralyzing advanced democratic states and instilling disorder may thus impel the more strategically adept groups on the cusp of pursuing bio- terrorism past the ‘tipping point’ and ‘across the Rubi- con’ (Ackerman & Peterson, 2020: 64). Some may contend that the international community’s focus should now be on biological weapons above all other types of CBRN, but we believe this to be too narrow in focus. Only by analyzing all CBRN weapons under one umbrella, are we able to exploit sufficient cases to gen- erate insights about incentives that would be directly fungible for professionals concerned with just biological weapons.

Even while we stress the important distinction between WMDs and CBRN technologies, we alert read- ers to the significance of studying the latter. We enlarge the scope of the discussion beyond highly fatal weapons that may remain out of reach for technologically limited groups. We likewise do not remain tethered to solely successful deployment of CBRN weapons but study pur- suit and use under the same theoretical framework. In so doing, we generate insights that are also applicable to pursuit and use of weapons that fall on the highly lethal end of the CBRN spectrum. Thus, our study notably informs scholars and pundits alike on the potential per- petrators for major catastrophic attacks using CBRN technology. A key takeaway we offer is that tying like- lihood of CBRN pursuit and use solely to militant groups’ capabilities would be akin to judging a book by its cover. There is more to militant groups than their size and membership. Organizations’ behaviors offer

important clues on the beliefs and practices that propel the pursuit of unconventional weapons.

Beyond its theoretical relevance, our findings are of policy importance insofar as practitioners can heed cer- tain types of behaviors as early-warning signs of groups’ goals to seek out these weapons. On that note, a beha- vioral metric that we scrutinize in this article is whether groups attack sites of cultural and symbolic value. Anec- dotal and news stories treat these cases as abhorrent inci- dents that violate a cherished heritage, but we hold that these attacks also spotlight groups’ incentives to think outside the box in a manner that breaks taboos. Put differently, attacks upon cultural sites may be indicative of the same belief systems as well as strategies that com- pel groups to pursue CBRN weapons.

CBRN weapon usage by non-state actors

From 2014 on, ISIS launched a chemical warfare cam- paign, and was responsible for 37 chemical attack inci- dents in Iraq and Syria, including using self-produced mustard agent, some of which led to injuries and deaths (Binder, Quigley & Tinsley, 2018). ISIS is the most recent and notorious example of chemical weapons pur- suit and use by a violent non-state actor. Other groups have launched CBRN attacks against security forces. For example, in Colombia the FARC utilized cyanide and tear gas in its guerrilla war against Bogota (O’Grady, 2002). Yet, why did these violent actors pursue and utilize CBRN weapons, while so many others do not?

To date, the topic has been explored mainly through anecdotal accounts and case studies that point to the real and imagined dangers of CBRN use by non-state actors, with only a handful of large-sample work on the topic. Many of these quantitative studies have empha- sized the role of environmental factors, uncovering how state level traits affect the likelihood of CBRN inci- dents. Asal, Ackerman & Rethemeyer (2012) deviate from the environmental focus by showing that organi- zational attributes significantly shape whether or not VNSAs pursue CBRN technology. In a similar vein, we study how characteristics of organizations link to CBRN pursuit.

However, we argue that organizations’ choices are guided more directly by what organizations do rather than what they are. Our study makes a further empirical contribution by expanding beyond Asal, Ackerman & Rethemeyer’s (2012) cross-sectional design and bringing in temporal dynamics to the study of weapon choice. By using new, yearly data contained within the Big Allied and Dangerous 2 Insurgency (BAADI2) project, which

Asal et al. 195

covers 140 insurgent actors from 1998 to 2012, we are able to harness information about organizational beha- vior over time.1

It behooves us to study how behavior affects decisions to pursue and use CBRN because some group attributes such as ideology are fairly invariant across time whereas other attributes such as size and strength change slowly. In contrast, organizations are innovative; they learn and adopt new tactics and may shift strategies over time. These tactical choices may serve as indicators of change in the propensities of groups to pursue CBRN weapons beyond what such proxies for capabilities as size and strength offer. There is also cross-organizational and temporal variation in CBRN pursuit that a focus on organizational attributes and environmental conditions may miss. For instance, Palestinian Islamic Jihad (PIJ) has pursued CBRN while another organization of similar size and age, also operating on the same territories in Israel, the Popular Resistance Committees, has not. Moreover, PIJ changed tack in 2001 to pursue CBRN and then abandoned the cause in subsequent years (Post, Sprinzak & Denny, 2003).

It should be noted at the outset that, while common in popular parlance, labeling CBRN weapons as WMDs is a misnomer. A common shortcoming in the literature on CBRN terrorism is the failure to clearly distinguish between CBRN weapons able to generate only a handful of casualties from those rare WMDs that exact catastrophic harm (see Ackerman & Rethemeyer, 2009). There is a vast spectrum of consequences, depending on the scale, delivery method, sophistication and choice of CBRN agent. Additionally, CBRN pur- suit is not the sole province of large and formal organi- zations. Looser cadres and lone actors have pursued these weapons but typically engaged in cruder, smaller scale, and less frequent CBRN plots and attacks (Acker- man & Pinson, 2014).

In this article, we are concerned with the basic deci- sion by a VNSA to pursue unconventional weapons, irrespective of whether the weapon constitutes a true

WMD or merely a small-scale use of CBRN agents. Therefore, it would be erroneous to assume that all CBRN technology is aimed at generating mass casualties. By this logic, broader group lethality on its own cannot comprise a compelling explanation for CBRN pursuit and use. Lethality is a poor determinant of organizational choice also because groups may be driven by the quest for publicity, more so than they are by the desire to generate mass casualties. More generally, arguments about group capabilities often fall short not only because CBRN technology that qualifies as WMDs is unattain- able for most groups, but also because not all groups seek true WMDs. Moving beyond capabilities, we offer an alternative explanation hinging on motivations. One of our core findings is that organizations that attack highly symbolic targets that embody ‘sacred values’ are more likely to embrace CBRN weapons. We posit that this stems from a similarity in motivations for attacking highly symbolic targets and using CBRN weapons. More importantly, it serves as confirmation that at least part of the puzzle as to which subset of VNSAs is most likely to select CBRN can be answered by examining the actors’ prior and concurrent behavior.

Aside from a handful of exceptions, there is limited large-sample empirical inquiry into why groups pursue mass casualty CBRN weapons, largely due to paucity of data. Most initial work in this area is case-study focused, with little thought to tying the cases together into a broader theoretical framework (Allison, 2004a, 2004b). Tucker’s comparative case study (2000) is an exception and concludes that organizations with an apocalyptic ideology, that possess charismatic leadership, and ones that are insular are more likely to pursue CB weapons.

A strictly motivational story misses the mark because while ideology and leadership style may render CBRN appealing, organizations weigh the pursuit of unconven- tional weapons against conventional weapons. Uncon- ventional weapons may not be any more effective and the consequences of wielding them are often indetermi- nate (Sinai, 2005), for example, organizations that pur- sue them can run the risk of alienating supporters. Contrary to popular wisdom, except in the most extreme cases, conventional weapons kill more people on a per incident basis than CBRN weapons (Rapoport, 2004).

Empirical scholarship follows two trajectories. The first focuses on country level determinants, looking at state level characteristics that facilitate the acquisition and use of CB weapons by groups (Ivanova & Sandler, 2006, 2007; Early, Fuhrman & Li 2013; Campbell & Murdie, 2018). The relatively more recent second stream shifts focus to organizational traits, with the social

1 The violent non-state organizations (VNSAs) in the BAAD2I dataset cannot be defined strictly as terrorist organizations, although many do utilize the tactic of terrorism (i.e. targeting non- combatants for coercive purposes). The BAAD2I project identifies insurgent organizations as non-state groups engaged in armed combat, which perpetrated at least 25 battle-related deaths according to the Uppsala Conflict Data Program (UCDP) and Battle Deaths dataset for the period 1998–2012. Therefore, the extent to which our results are directly comparable with research that is solely focused on terrorist organizations remains an open question.

196 journal of PEACE RESEARCH 60(2)

networks in which violent groups operate bridging these two tracks (Asal, Ackerman & Rethemeyer, 2012). In order to foreground our theoretical argument, we briefly survey and outline these two strains of literature.

Looking at the first stream, we find that the environ- ments in which organizations function shape their choices. Ivanova & Sandler (2006, 2007) find that a strong rule of law, democracy, wealth, and corruption are all predictors of CBRN incidents. Economic globa- lization heightens the likelihood by facilitating skill and personnel transfer across borders (Asal, Ackerman & Rethemeyer, 2012; McComb, 2013). At the same time, civil turmoil imposes logistical costs on groups, impeding their effective functioning. Other scholars stress that cor- ruption also raises the likelihood that organizations will pursue CBRN technology (Ivanova & Sandler, 2007; Early, Fuhrmann & Li, 2013) by making it easier for illicit actors to pay to get their hands on CBRN material. Campbell & Murdie (2018) argue that state repression can expose countries to the threat of CBRN acquisition by hindering intelligence gathering and counter- terrorism efforts, and that countries with better human rights records have a lower likelihood that CBRN mate- rial will make it into the hands of non-state actors (Campbell & Murdie, 2018).

The second strain of the literature shifts focus to the organizational level. Asal, Rethemeyer & Anderson (2009) and Asal, Ackerman & Rethemeyer (2012) argue that organizational behavior is dictated to a large extent by capabilities, in addition to threat environments, rather than by ideology. Ivanova & Sandler (2007) also explore the impact of terrorist organization ideology and con- clude that non-fundamentalist groups are just as likely to stage CBRN attacks as fundamentalist networks. The key plank of Asal, Ackerman & Rethemeyer’s (2012) argument is that network ties and resources pay divi- dends: wider alliance networks elevate the probability that groups will pursue and develop CBRN technology, by providing more access to materials and expertise and allowing for the diffusion of justifications for otherwise taboo behavior.

We contribute primarily to the second strain, focusing on those factors that prompt VNSAs to seek out these weapons, rather than seeking to uncover factors that lead to their successful use. We extract several insights from the broader scholarship before proceeding to our hypoth- eses. First, ideology presents a partial story insofar as many organizations of different ideological stripes have considered acquiring or using CBRN weapons (Asal, Ackerman & Rethemeyer, 2012). In a related vein, contrasting conceptual and empirical accounts exist on

how religious ideology relates to the propensity of orga- nizations to pursue unconventional weapons (Post, Sprinzak & Denny, 2003; Blum, Asal & Wilkenfeld, 2005; Campbell & Murdie, 2018).

Second, pursuing CBRN weapons is a strategic as well as tactical decision. Possessing CBRN may shift the bar- gaining range outside the bounds of previously available options on the table (Fearon, 1995). High-casualty attacks aid VNSAs by wresting coercive leverage through the civilian costs of ongoing conflict (Pape, 2003). To reiterate, CBRN weapons are not necessarily WMDs, but for those that do have the potential to cause cata- strophic levels of damage, they could potentially exert a high degree of coercive leverage. At the same time, since the pursuit and threat of CBRN weapons by groups is likely to incite wide-scale reprisals, organizations may jettison CBRN as a riskier and on average less successful way to achieve lethal outcomes. Also, insofar as CBRN weapons collide with prevailing norms, they arouse social apprehension and disapproval and may cost the organi- zation the support of the populace. In short, organiza- tions face tradeoffs when weighing CBRN against conventional weapons (Gressang, 2001; Palfy, 2003).

Third, terrorists often think beyond the tangible effects of their actions. That is, decisions do not simply reflect tactical concerns with event magnitude, imme- diate damage, or fatalities. Given that most CBRN attacks involve lower-level agents and toxins such as ricin, tear gas, and cyanide (Koehler & Popella, 2018), it is plausible that many organizations who pur- sue them are drawn to their psychosocially disruptive effects, beyond the casualty metric. As Stern (1999: 48) notes, CBRN weapons ‘ . . . are inherently terrorizing. They evoke moral dread and visceral revulsion out of proportion to their lethality’. Moreover, although the probability that an organization will successfully wield a true WMD may be small, the anticipated destruction and disruption that would ensue from a WMD attack involving CBRN is so immense that merely the pursuit of these weapons can exact a psychological toll on potential victims. Faced with small probability events of high magnitude, people become preoccupied with the consequences and forget that these events are unlikely to transpire (Sunstein, 2002).

Fourth, beyond their purely strategic or tactical appeal to VNSAs, the pursuit of CBRN weapons can be driven by individual psychology, status-related or internal group dynamic factors. For example a leader displaying an affi- nity for innovation or even an idiosyncratic ‘technofe- tishism’ (Bale, 2017) can be drawn to one or more types of unconventional weapons, as demonstrated by Shoko

Asal et al. 197

Asahara, the leader of the Japanese doomsday cult Aum Shinrikyo who went so far as to produce odes to the nerve agent sarin, which his group subsequently used on the Tokyo subway in March 1995 (Danzig et al., 2012; Revill, 2017).

Fifth, organizational connectedness plays an impor- tant role in affecting tactics (Asal & Rethemeyer, 2008). We know that tactical diffusion is facilitated by alliance networks such that VNSAs adopt and modify tactics they deem to be successfully employed by other violent organizations in their network (Asal, Ackerman & Rethemyer, 2012). VNSAs in the same network can also transfer skills, resources and the legitimacy of certain actions by cooperating logistically, such as training each other’s operatives.

Variables and hypotheses

The preceding discussion suggests that we can categorize potential determinants of CBRN pursuit and use into four broad categories: (1) environmental factors (usually at the country level), (2) structural organizational attri- butes, (3) organizational behavior and (4) connections between the organization and other actors. It is upon the third category – the prior or concurrent behavior of the VNSA organization – which we focus here, especially the effects of organizational tactics such as target selec- tion, and tactical effectiveness. We do so for two reasons. The first is that factors within this category have not received much attention in past research. While previous studies have included some variables that reflected the behavior of the organization, they have focused primarily on either the environmental variables (Ivanova & Sand- ler, 2006; Early, Fuhrmann & Li, 2013; Campbell & Murdie, 2018) or organizational attributes (Asal, Acker- man & Rethemeyer, 2012). Second, organizational behavior is both specific to the group (unlike the breadth of country-level factors that apply to all organizations in the same country) and is often more observable than other types of organizational attributes, such as the size or leadership structure. If such factors can be shown to be correlated with CBRN pursuit, then they can serve as particularly useful indicators for counterterrorism authorities, in that they can both be easily observed and can discriminate between different organizations in the same country.

It should be noted here that a major methodological advance over previous quantitative analyses of CBRN terrorism is the ability to test organizational variables yearly. Therefore, we explicitly retest some of the vari- ables analyzed by Asal, Ackerman & Rethemeyer (2012).

We add to the theoretical arguments underlying some of the previously hypothesized variables’ effects on CBRN pursuit or use, especially those in the fourth category (networks), and test several completely new variables. In addition, we include variables that represent the other two categories (environment and organization attributes) as control variables, for which we simply reference the previous study’s rationale for their inclusion rather than enter into detailed discussion.

Organizational behavior Targeting cultural sites. The link between attacking cultural sites on the one hand, and CBRN weapons pur- suit on the other is based on a hypothesized similarity in motivational impetus. On one level, this similarity reflects merely one manifestation of the search for pub- licity, a goal inherent to terrorists but cherished by var- ious violent organizations the world over. Jenkins (1974) aptly refers to terrorism as theater, a metaphor which reflects the symbolic nature of violence. Perpetrators engage in violence to communicate with an audience. In an increasingly desensitized media environment, novel and distinct signals are needed to garner the atten- tion that VNSAs crave. While several decades ago timing attacks with important events (Aksoy, 2014) or attacking a well-known personage might have been all that was required to guarantee the media spotlight, over time this requires more novel or outlandish attacks to attain high levels of publicity.

To illustrate how attacking cultural sites might achieve this, in 2015 ISIS destroyed ancient ruins at Palmyra, triggering international outrage at the destruc- tion of invaluable artifacts. In the same year, ISIS released a video depicting the demolition of Hatra, an ancient Greco-Roman city that was declared a World Heritage site in 1985. These attacks served as part of the group’s larger religiously motivated propaganda and cul- tural cleansing campaign to vilify pre-Muslim civiliza- tions and other Muslim sects (Curry, 2015). Previously, in 2001, the Taliban destroyed the Bamiyan Buddha statues in Afghanistan, triggering international condemnation and outrage at the destruction of invalu- able cultural artifacts. These symbolic acts simultane- ously capture imaginations and elicit considerable media coverage, which in turn sensationalizes the event (Picard, 1993).

The pursuit of CBRN by itself carries a taste of the sensational. Groups that are able to successfully launch CBRN attacks can disrupt governance, undercut trust in government agencies, and thereby further exacerbate

198 journal of PEACE RESEARCH 60(2)

a sense of doom or helplessness. In addition, the some- times gruesome effects of CBRN agents have long been ingrained in a variety of popular culture media, from techno-thriller novels to science fiction movies. Thus, if ‘[t]errorists want a lot of people watching’ (Jenkins, 1974), both attacking sites with monumental and senti- mental value and pursuing CBRN weapons are guaran- teed to grab headlines.

On another level, however, these behaviors do not represent just a greater ‘volume’ or diffusion potential of the perpetrators’ message but a specific qualitative character that is embodied within the message. Both attacking highly symbolic targets and pursuing CBRN weapons are in and of themselves psychologically potent strategies. Cultural targeting is an attack against often deeply held societal intangibles. ‘By undermining the cultural identity of local communities, destruction of religious monuments in political violence and theft of national collections hold tactical value for insurgencies and terrorist groups’ (Nemeth, 2010: 10). These orga- nizations are willing to go the extra mile by overtly tram- pling upon cherished treasures. As in the case of ISIS, violent actors gain notoriety by targeting heritage sites.

Similarly, CBRN weapons are qualitatively distinct from conventional weapons and thereby lend a greater psychological edge to organizations (Ackerman & Pin- son, 2014). The public imagination is captivated by the notion of an ‘excruciating and horrible death and sec- ondary effects like the creation of uninhabitable spaces’ (Koehler & Popella, 2018: 4), which also renders CBRN agents effective in delivering impact with comparable low casualty rates. Insofar as the coercive effectiveness of terrorism, including that carried out during an insur- gency, pivots on the ability to foment and disseminate fear to wider audiences than the immediate victims (Braithwaite, 2013), CBRN weapons lend organizations a stronger psychological edge. Moreover, the pursuit of WMD weapons is widely characterized as a normative taboo (Tannenwald, 1999) and the common conflation among both VNSA’s supporters and targeted publics of CBRN with WMD can result in any CBRN pursuit having a similar effect. Early, Fuhrmann & Li (2013) write that simply by plotting nuclear and radiological attacks, terrorists amass notoriety.

The specter of CBRN weapons in the hands of a militant organization therefore evokes a sense of dread and awe and exacts a psychological cost, while guarantee- ing high levels of publicity, independent of casualties, just as the violation of sacrosanct assets by targeting cultural heritage sites does. While most VNSAs strive for publicity, terror and breaking taboos (Schmid,

2011), both of the modalities considered here arguably break taboos beyond ‘normal’ terrorist attacks and in doing so, evoke fear and achieve sensationalism without necessarily exacting a high death toll. Such behavior can up the ante so to speak, in order to jar publics that are relatively inured to ‘ordinary’ VNSA violence. Therefore, we expect cultural targeting to be in lockstep with the pursuit of CBRN technology.

Our variable for cultural targeting comes from the BAADI2 project. The variable tracks insurgent attacks on heritage sites designated by the UNESCO World Heritage List or clearly identified in existing sources as having a cultural or symbolic value. The list of attacks contains targeting of any cultural, natural, or mixed site in a given country, by a VNSA.2 The original data iden- tifies 832 cultural sites and 206 natural sites. The vari- able codes for any attacks in a year by an organization that were either directly aimed at the site according to the Global Terrorism Database GTD (START, 2018) or were geographically identified (based on longitude and latitude) as being in the boundaries of the heritage site and then externally validated from newspapers that the attack was directed at the site. The indicator reports 1 if the heritage site was the intended target of the insurgent group and 0 otherwise. When insurgent groups are con- sidered, we have 11 organization-years with explicit heri- tage site targeting, attributed to eight groups, two of whom have attacked the same site more than once. Table A1 in our Online appendix lists these cases, including names of World Heritage-designated attack sites.

H (Culture): Organizations that target heritage sites are more likely to engage in pursuit of CBRN capabilities.

Criminal activity. Insurgent organizations often engage in organized criminal activity (outside of their ideologi- cally motivated violence), usually for purposes of raising funds to support the organization (Shelley & Picarelli, 2002; Hutchinson & O’Malley, 2007; O’Neil, 2007; Williams, 2009). Groups which persist in criminal activ- ities for some time invariably become adept at moving goods and finances illicitly within and across country borders. They also tend to make contacts with networks of middlemen and illicit suppliers in the criminal under- world who can act as key nodes in the identification and procurement of a wide variety of materials and expertise. The greater the number and diversity of criminal

2 The World Heritage list by country can be obtained here: https://whc.unesco.org/en/list/.

Asal et al. 199

activities the organization is involved in, the greater this effect is likely to be, lending greater confidence in suc- cessful acquisition and deployment of CBRN weapons- related materials and thus encouraging pursuit. We draw on BAADI2 for an ordinal variable counting up the types of distinct criminal activity each organization was engaged in, from zero to five.

H (Crime): The greater the number of different crim- inal activities an organization is involved in, the more likely it is to pursue CBRN weapons.

Connections between the organization and other actors State sponsorship. In their previous analysis, Asal, Ack- erman & Rethemeyer (2012) posited two factors that might make state-sponsored terrorist organizations more likely to engage in CBRN terrorism: (a) the greater tech- nical and material-related resources that state sponsor- ship often bestows on a terrorist organization and (b) such organizations are less beholden to a wider pop- ular constituency and thus would experience fewer stra- tegic constraints. The cross-sectional data did not show any significant relationship in this regard, leading the authors to speculate that state sponsors would exert just as many restraints, if not more so, on their protégés’ tactical choices, with a particularly cautious attitude when it comes to CBRN weapons (Asal, Ackerman & Rethemeyer, 2012). We retest the original hypotheses by constructing a binary variable capturing whether an orga- nization received foreign state support in the year in question.

H (State sponsor): Organizations known to receive some form of state sponsorship will be more likely to engage in CBRN terrorism.

Alliances. We add to Asal, Ackerman & Rethemeyer’s (2012) prior theoretical argument the insight that VNSA networks might not only assist in the provision of capability-related goods, but might also augment the motivation to pursue CBRN. If any element of the net- work, such as an ideologue or strategist in a connected group, becomes interested in CBRN weapons or pro- motes or justifies their use, then the existing trusted connections can facilitate the spread of these ideas to the organization itself. Therefore, all else being equal, those organizations with multiple alliances are more likely to be exposed to motivating influences for CBRN weapons and therefore are more likely to themselves make the decision to pursue. To measure alliances that each orga- nization has we use a variable in BAADI2 that is a count of allies that each organization has in a year.

H (Alliances): The more VNSAs that an organization is allied with, the more likely it is to pursue CBRN.

Control variables

In order to test the effect of organizational behavior relative to other types of factors, we include in our analysis a number of environmental and organizational attribute variables as controls. We incorporated environ- mental variables related to the country in which the VNSA is based, that either have been tested by other scholars in connection with terrorism or are commonly examined in the literature on political violence. These include: economic wealth, as measured by Gross Domes- tic Product per capita (GDP/capita in USD at PPP) derived from the Quality of Government dataset (Teorell et al., 2011); levels of civil conflict (Asal, Ackerman & Rethemeyer, 2012), proxied by the number of battle deaths drawn from UCDP/PRIO data (Pettersson & Eck Pettersson, 2018); and three variables linked to the behavior of the state. These latter variables measure the amount of host country repression (Avdan & Uzonyi, 2017) for the country and year in question through the inferred Polity IV score (fh_ipolity2) as defined by the Quality of Government Project (Marshall, Jaggers & Gurr, 2011; Dahlberg et al., 2020), the Political Terror Scale ( Henschke & Legrand, 2017), and a binary vari- able for whether the state has embraced a repressive coun- terinsurgency (COIN) strategy, including routine police work, investigating crimes, arresting members, bringing cases to courts and any military actions against the group. This indicator, labeled ‘sticks’, is sourced from the BAADI2 project (Asal, Rethemeyer & Schoon, 2018). The last environmental control variable is related to CBRN directly and tests the impact of the presence of a state-level nuclear weapons program by coding a binary variable based on information obtained from the Bulletin of the Atomic Scientists (2019).3 Arguably, while the link between a nuclear weapons program and groups’ pursuit of CBRN may be tenuous, terrorist groups have championed existing nuclear weapons programs in order to gain political influence (Joshi, 2020).

We also considered several variables previously tested by Asal, Ackerman & Rethemeyer (2012), mainly related to organizational attributes. First, we include an indica- tor for total fatalities, incorporated from the Global Ter- rorism Database (GTD), which records all terrorist attacks on a yearly basis. Our variable is matched to

3 See Bleek (2017) for a comprehensive coding of state nuclear programs.

200 journal of PEACE RESEARCH 60(2)

BAADI2 groups and represents attacks orchestrated by these groups in a given year. In an alternative model, we disaggregate the total fatalities indicator into two components: (1) a sum of total fatalities aimed at civi- lians and (2) a sum of total police and military deaths. We are motivated to disaggregate the fatalities counts given the conjecture that the targeting of civilians con- veys that groups have far-reaching aims (Abrahms, 2006). If this argument is correct, then organizations with extreme preferences may be more predisposed to pursue unconventional weapons, in an effort to turn the tide in their favor, at all costs. As another group-level facet, we account for organizational ideology, by includ- ing dummy variables for three ideological categories: religious, far left and separatist,4 with categories drawn from BAADI2. The other variables retested were size, age, and a binary variable for territorial control, all drawn from BAADI2.

Data To test which factors lead insurgent organizations to pursue or use CBRN weapons, we use the Big Allied and Dangerous Insurgency 2 (BAAD I2) dataset (Asal, Rethe- meyer & Schoon, 2018) married to the Profiles of Inci- dents involving CBRN and Non-state actors (POICN) dataset (Binder & Ackerman, 2021).5 The BAADI2 insurgency dataset covers 140 organizations drawn from the Uppsala Armed Conflict Data Program’s dataset,

which collects data on organizations that have been involved in at least 25 battle deaths in a year (Gleditsch et al., 2002; Pettersson & Wallensteen, 2015). These organizations comprise 95% of all non-state actors engaged in armed insurgency during the 1998 to 2012 time period (Asal, Rethemeyer & Schoon, 2019). The BAADI2 dataset contains information on organizational ideology, structure and behavior beyond political violence.

The POICN dataset is an extensive collection of information on ideologically motivated plots, attempts and actual uses involving CBRN materials since 1990. The POICN database, in addition to scrupulously eval- uating the credibility of all its sources and capturing multiple uncertainties in metadata, also distinguishes between low-end CBR terrorism and more serious events. Our dependent variable is a binary variable coded as a 1 if an organization plotted to use CBRN weapons, threatened an attack while possessing a CBRN weapon, or attempted a CBRN attack (note that ‘pure’ threats without the actual acquisition of a weapon are excluded). As noted earlier, this is not that common an occurrence: of the 1,386 organizational years in the dataset there are only 50 organizational years or 3.61% where an

Table I. Organizations and number of years organization was involved in CBRN activity

No. of years Organization name

3 Al-Aqsa Martyrs Brigade 9 Al-Qa’ida 2 Al-Qa’ida in the Arabian Peninsula (AQAP) 1 Ansar Al-Islam 1 Armed Islamic Group (GIA) 1 Communist Party of India – Maoist (CPI–M) 1 Eastern Turkistan Islamic Movement (ETIM) 6 Hamas (Islamic Resistance Movement) 1 Hizballah 3 Islamic State of Iraq and al Sham (ISIS) 1 Kurdistan Workers’ Party (PKK) 2 Liberation Tigers of Tamil Eelam (LTTE) 1 Mujahedin-e Khalq (MEK) 2 National Liberation Army of Colombia (ELN) 1 National Union for the Total Independence of Angola

(UNITA) 1 Oromo Liberation Front 1 Palestinian Islamic Jihad (PIJ) 1 Popular Liberation Army (EPL) 1 Real Irish Republican Army (RIRA) 4 Revolutionary Armed Forces of Colombia (FARC) 2 Salafist Group for Preaching and Fighting (GSPC) 5 Taliban

4 We are not asserting that only these ideologies might be associated with CBRN pursuit, but the impact of other ideologies is not testable with the available insurgent data. For example, recent work has highlighted the importance of far-right – either of a neo-Nazi or white supremacist flavor – in driving CBRN plots, particularly in Western states (Koehler & Popella, 2018). Unfortunately, our data contain only four observation years with a group coded as having a right-wing ideology, the Somali Reconciliation and Restoration Council (SRRC), 2001–2004. This data limitation precludes us from analyzing the effects of far-right ideology on CBRN pursuit and use. It bears mentioning, however, that the sparsity of organization-years corresponding to right-wing ideology is congruent with recent work, which has noted that far-right plots are more common among lone actors rather than organizations (Koehler & Popella, 2018). The authors further underscore the paucity of right-wing organization plots in existing datasets of terrorism, including the GTD and the Monterey database. In addition, the BAAD2I definition of right-wing ideology is premised on the economic right–left spectrum and consequently does not allow a direct comparison against Koehler & Popella (2018). 5 POICN is open-source; interested parties may obtain access to the data by emailing [email protected]. Please see https://www.start.umd. edu/research-projects/profiles-incidents-involving-cbrn-and-non- state-actors-poicn-database.

Asal et al. 201

organization was coded as a 1. Table I lists the 22 orga- nizations which were involved in such behavior from 1998 to 2012. As can be seen, most organizations were only involved in one year of CBRN activity. That being said, there are several that engaged in this activity far more often. The organization that was most active was al-Qaida, which was engaged in such activity for nine years, followed by Hamas for six years and then the Taliban for five years and FARC for four years.

Table II provides descriptive statistics for our variables.

Analysis and findings

Because our dependent variable is a binary variable, we use logistic analysis; and to interpret the impact of the variables, we provided discrete changes in predicted probabilities (Long & Freese, 2006). To control for the possible impact of previous CBRN behavior we lag our dependent variable forward one year and include the dependent variable from the year before as one of our independent variables. We also cluster on the organiza- tional identifier to account for non-independence of observations within each organization. Table III presents the logistic regression along with percentage changes in

probability from their minimum value to their maxi- mum value. Model 1 presents the main model including total GTD fatalities whereas Model 2 disaggregates GTD fatalities by police/military, and civilian fatalities.

Table III shows that only seven of the variables in Model 1 and seven variables in Model 2 have a statisti- cally significant impact on the dependent variable. First, and unsurprisingly, if the organization is engaged in the behavior the year before, they are more likely to continue to do so the following year, though this effect is quite small (3.15%).

Turning to the organizational behavior variables, there are several interesting results.

First, organizations that attack cultural sites are more than 13% more likely to engage in CBRN behavior. This suggests that there might be some credence to the notion that organizations willing to cause harm to a community’s core values in a way that guarantees sensa- tionalism and deep psychological resonance will be less constrained in their pursuit of CBRN weapons. Although there is obviously a signaling element to such attacks, there might also be a deeper phenomenon at work. Destroying cultural heritage carries connotations of going beyond what is a strategic or instrumental neces- sity. Acting out of ‘spite’ is probably too crude a

Table II. Descriptive statistics

Variable Observation Mean Std. Dev. Min. Max.

CBRN plot or use lagged forward 1,247 0.034 0.181 0 1 CBRN plot or use 1,386 0.036 0.187 0 1 Attacked cultural heritage site 1,386 0.006 0.076 0 1 GTD fatalities 1,386 35.427 156.664 0 2,996 GTD police and military fatalities 1,386 12.715 68.797 0 1,676 GTD civilian fatalities 1,386 22.71138 110.7174 0 2,807 Criminal activity 1,386 0.524 0.990 0 5 Foreign state support 1,386 0.082 0.274 0 1 Alliance count 1,386 0.818 1.596 0 15 Wealth (GDP/cap) 1,270 4317.403 6298.057 199.01 33,580.82 Conflict (UCDP battle deaths) 1,386 219.328 825.214 0 14,716 Home base has nuclear weapons 1,386 0.269 0.444 0 1 Leftist ideology 1,386 0.212 0.409 0 1 Separatist ideology 1,386 0.467 0.499 0 1 Foreign state support 1,386 0.082 0.274 0 1 Criminal activity 1,386 0.524 0.990 0 5 Territorial control 1,386 0.247 0.431 0 1 Stick COIN strategy applied 1,386 0.452 0.498 0 1 Age 1,386 17.436 14.359 0 65 Size 1,386 2.688 0.705 1 4 Alliance count 1,386 0.818 1.596 0 15 Polity IV dataset inferred 1,386 5.071 2.840 0.25 10 Political terror scale Amnesty Intl. w/inferred data 1,386 4.038 0.896 1 5

202 journal of PEACE RESEARCH 60(2)

moniker, but there might thus be an association between an organization’s desire to ‘show the world the middle finger’ and a willingness to pursue CBRN weapons, which are often viewed as being beyond the pale. This is all the more telling when one considers that the num- ber of fatalities a group has inflicted was not a significant variable in the analysis. This differs theoretically from the cultural attack variable in that there might be a greater number of strategic or tactical reasons to cause large numbers of fatalities than to attack cultural sites.

Second, the diversity of criminal activity was not sig- nificant, perhaps because the suspicion and compart- mentalization that is the hallmark of criminal activity does not in practice allow for many of the hypothesized network effects to operate. Third, when considering more direct network effects – that is, the connections with other actors – aligning with past research on the impact of alliance connections on VNSA behavior, orga- nizations with more alliances are more likely to engage in CBRN behavior with a 25% impact on the likelihood from a minimum to a maximum value of alliance counts. State sponsorship assumes significance only in Model 2 but even then, only increases the probability of CBRN pursuit and use by less than 1%.

Turning to the control variables, first, neither the overall fatality rate in Model 1, nor its components in Model 2 register significant effects. Although group leth- ality indicators were not shown to be significantly related to CBRN pursuit, they might still be linked to WMD pursuit and use, which was not testable given the (for- tunate) paucity of WMD cases. Moreover, the actual number of fatalities is often more closely related to the efficacy with which an attack is conducted, rather than the intention of the attacker (for instance, when an orga- nization seeks to kill hundreds but only ends up injuring a few individuals). Perhaps a better variable to test in future analyses, if data describing it can be obtained, is the intended number of casualties, rather than the actual number of fatalities.

Second, organizations in richer countries are much more likely to engage with CBRN by over 26%. Also, the greater the level of conflict that an organization is involved in (i.e. the greater the number of battle deaths on all sides), the more likely they are to engage with CBRN, having a 23.5% change from minimum to max- imum probability.

With respect to organizational attributes, as in previ- ous empirical analyses and challenging much of the lit- erature on CBRN pursuit and use by VNSAs, the religious ideology variable does not have an impact (nor does the separatist ideology control). This draws

Table III. Table Effects of organizations’ tactics and lethality on CBRN pursuit and use

%

(1) (2) Change in probability

Pursuit and use of CBRN (lagged)

1.822** 1.813** 3.21%

(0.618) (0.609) GTD total fatalities 0.002* 43.01%

(0.001) GTD police and

military fatalities 0.007* 22.38%

(0.003) GTD civilian fatalities 5.78e-05 NS

(0.001) UCDP battle deaths 0.0004* 0.0004* 17.72%

(0.0002) (0.0002) Attack cultural site v2 2.775* 2.807* 9.65%

(1.360) (1.185) Religious ideology 0.283 0.256 NS

(0.427) (0.410) Leftist ideology 1.557** 1.595** 1.9%

(0.554) (0.548) Separatist ideology �0.360 �0.413 NS

(0.678) (0.684) State sponsorship 0.498 0.552 NS

(0.470) (0.463) Criminal activity 0.261 0.321y 2.29%

(0.185) (0.194) Territorial control �0.311 �0.437 NS

(0.521) (0.561) COIN strategy

applied 0.590 0.558 NS

(0.365) (0.360) Organization age �0.004 �0.001 NS

(0.015) (0.014) Organization size �0.581 �0.667y �1.83%

(0.381) (0.378) Alliance count 0.398** 0.382** 60.82%

(0.140) (0.138) Polity IV score �0.0994 �0.144 NS

(0.0996) (0.102) Political terror scale

(Amnesty) 0.399y 0.411y 0.83%

(0.238) (0.242) Real GDP per capita 0.0001*** 0.0001*** 17%

(2.81e-05) (2.97e-05) Country nuclear

program �0.880 �0.668 NS

(0.775) (0.747) Constant �5.781*** �5.549***

(1.600) (1.577) Observations 1,238 1,238

Robust standard errors in parentheses. *** p<0.001, ** p<0.01, * p<0.05, y p<0.1.

Asal et al. 203

attention back to the discussion in Asal, Ackerman & Rethemeyer (2012) regarding the potential subsuming of the ideology variable in other variables like alliances, as well as the possible mischaracterization of ideology using simple one-dimensional labels (see also Ackerman & Burnham, 2021). However, somewhat surprisingly, hav- ing a leftist ideology does have a positive impact, albeit a very small one (1.88%). There are no theoretical reasons for why leftist organizations should be more interested in CBRN than organizations espousing alternative ideolo- gies, so one might speculate that the observed effect is potentially capturing an unmeasured covariate that is more often seen among leftist groups.

As for the remaining control variables, the analysis confirmed the lack of significance of several variables found in Asal, Ackerman & Rethemeyer’s (2012) study, including territorial control, organization size, and age. Moreover, none of the additional environmental vari- ables related to the actions and type of the host country regime tested (political terror, or a repressive counter- insurgency strategy) proved to be significant. Last, the nuclear weapons status of the host country showed no significant association with CBRN pursuit, although this factor might still affect the pursuit of RN weapons spe- cifically. Possibly, as well, there may be a stronger link between a civil nuclear power program, rather than a nuclear weapons program, and the pursuit of RN. The number of RN weapons’ pursuit cases on their own is too small to test for statistical significance.

In addition to the changes in probability from mini- mum to maximum for each significant variable in Table III, in Table IV we present the changes in probability from minima to maxima by alliance connections as this variable interacts with the organization attacking cultural sites. As one can see, the combination is very powerful. Organizations that have no connections and have not attacked a cultural site in the previous year have less than a 1% probability of engaging in CBRN activity. If they have no alliances but have attacked a cultural site their probability jumps up to almost 6% and if they have done so and have more than ten network connections the likelihood of doing so jumps to nearly 50%.

Finally, as sensitivity analyses, we re-evaluated both models using rare events logit. As our data section dis- cussed, our dependent variable takes on the value of 1 only 3.61% of the time. Given the rarity of CBRN pur- suit and use, coefficients may be biased downward due to the zero dominance of the outcome variable. As a con- sequence, predicted probabilities are biased downward (King & Zheng, 2001). The rare events results are pre- sented in the supplementary document, Table AI,

followed by first differences in probabilities, in Table AII. In brief, these analyses reveal mostly parallel patterns to the results presented in Table III, lending credence to our hypotheses on targeting cultural sites and lethality.

Conclusion

The theoretical contribution of our article is to shift the focus from what organizations are to what organizations do. Rather than focusing on static organizational traits that pertain to capabilities, we uncover the effects of prior tactics and target choice on propensities to pursue CBRN.

Our article is the first to empirically re-evaluate the results of previous scholarship (Asal, Ackerman & Rethe- meyer, 2012) on organizations’ attributes on CBRN pursuit and use by using longitudinal data. We show that some of the keystone findings obtain statistical sup- port when modeling organizations’ decisions over time. Notably, we confirm that network connectedness signif- icantly shapes the probability of CBRN pursuit over time. To this finding we add the expected organizational attribute of prior pursuit of CBRN weapons and the more surprising choice to attack cultural sites, as impor- tant indicators of CBRN pursuit.

Unlike Asal, Ackerman & Rethemeyer (2012), we also find substantially more support for environmental factors, primarily the wealth of the country in which the organization is based and the severity of overall conflict occurring in the country. Our results showcase that groups that do pursue CBRN are not necessarily larger or older or more fatal organizations. But rather, these are

Table IV. Change in probability for alliance count and attack- ing cultural sites on likelihood of CBRN Activity

Alliance Count No Yes

0 0.18% 2.92% 1 0.27% 4.22% 2 0.39% 6.06% 3 0.57% 8.63% 4 0.83% 12.15% 5 1.21% 16.84% 6 1.76% 22.88% 7 2.56% 30.29% 8 3.70% 38.89% 9 5.33% 48.24%

10 7.62% 57.72% 11 10.77% 66.66% 12 15.03% 74.55% 13 20.57% 81.10% 14 27.50% 86.27% 15 35.72% 90.20%

204 journal of PEACE RESEARCH 60(2)

organizations that are ensconced in alliance networks. In fact, the effects of alliance ties trump the effects of state sponsorship, which provides a trivial boost to the prob- ability of CBRN pursuit. Policy-makers should thus keep abreast of the broader strategic environment in which organizations operate, not necessarily vis-à-vis state actors, but with respect to allied VNSAs.

A further insight that our findings yield concerns venue choice. Our data comprise eight organizations that chose to attack heritage sites. While the corresponding organiza- tional years constitute only a small proportion of the data, the results do suggest that the same hunger for publicity that compels violent groups to attack sites of symbolic and cultural value may also fuel them to pursue CBRN tech- nology. Beyond being publicity stunts, both strategies strive for sensationalism and arguably break taboos beyond even those that VNSAs (especially terrorists) flout using their usual forms of violence. This opens up an interesting future research question on the roles of actor reputation and legiti- macy in CBRN pursuit. Previous work underscores that organizations may compromise legitimacy in order to establish reputations for brazenness (Asal & Rethemeyer, 2008; Asal, Ackerman & Rethemeyer, 2012; Tokdemir & Akcinaroglu, 2016). Perhaps organizations willing to be defined by negative reputations and are willing to engage in certain odious acts – such as child recruitment and sexual enslavement – are more likely to pursue and use CBRN weapons. In any event, the door has been opened to explor- ing behavioral (rather than only structural or environmen- tal) aspects of VNSA behavior as forward indicators of CBRN pursuit.

Replication data

The dataset, codebook, and do-files for the empirical analysis in this article, along with the Online appendix, are available at https://www.prio.org/jpr/datasets/. All analyses were conducted using Stata Version 16.

Acknowledgements We would like to thank Karl Rethemeyer and Suzan Wheedon Levy, participants at the panel on terrorism at the International Studies Association 2019 meeting in Tor- onto, ON, the three anonymous reviewers, and the editor of JPR for their helpful feedback on previous drafts.

Funding The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: this research was supported by the Defense

Threat Reduction Agency Basic Research Award HDTRA1-10-1- and by the Department of Homeland Security Science and Technology Directorate’s Office of University Programs through award no. 2012-ST-061- CS0001, made to the National Consortium for the Study of Terrorism and Responses to Terrorism.

ORCID iD Nazli Avdan https://orcid.org/0000-0002-0984-6120 References Abrahms, Max (2006) Why terrorism does not work. Interna-

tional Security 31(2): 42–78. Ackerman, Gary & Michael Burnham (2021) Towards a def-

inition of terrorist ideology. Terrorism and Political Violence 33(6): 1160–1190. DOI: 10.1080/09546553.2019. 1599862.

Ackerman, Gary & James Halverson (2016) Attacking nuclear facilities: Hype or genuine threat? In: Brecht Volders & Tom Sauer (eds) Nuclear Terrorism: Countering the Threat. New York: Routledge, 111–141.

Ackerman, Gary & Hayley Peterson (2020) Terrorism and COVID-19. Perspectives on Terrorism 14(3): 59–73.

Ackerman, Gary & Lauren Pinson (2014) An army of one: Assessing CBRN pursuit and use by lone wolves and auton- omous cells. Terrorism and Political Violence 26(1): 226–245. DOI: 10.1080/09546553.2014.849945.

Aksoy, Deniz (2014) Elections and the timing of terrorist attacks. Journal of Politics 76(4): 899–913.

Allison, Graham (2004a) Nuclear Terrorism: The Ultimate Pre- ventable Catastrophe. New York: Times Books/Henry Holt.

Allison, Graham (2004b) Nuclear terrorism: How serious a threat to Russia? Russia in Global Affairs Online Edition, September/October (https://www.belfercenter.org/publica tion/nuclear-terrorism-how-serious-threat-russia). Accessed 3 December 2021.

Asal, Victor & R Karl Rethemeyer (2008) The nature of the beast: Organizational structures and the lethality of terrorist attacks. Journal of Politics 70(2): 437–449.

Asal, Victor & R Karl Rethemeyer (2009) Islamist use and pursuit of CBRN terrorism. In: Gary Ackerman & Jeremy Tamsett (eds) Jihadists and Weapons of Mass Destruction. Boca Raton, FL: CRC, 335–358.

Asal, Victor; Gary Ackerman & R Karl Rethemeyer (2012) Connections can be toxic: Terrorist organizational factors and the pursuit of CBRN weapons. Studies in Conflict & Terrorism 35(3): 229–254.

Asal, Victor; R Karl Rethemeyer & Ian Anderson (2009) Big allied and dangerous (BAAD) database 1-Lethality data, 1998–2005, Project on Violent Conflict (https://data verse.harvard.edu/dataset.xhtml?persistentId¼hdl%3 A1902.1/16062). Accessed October 2021.

Asal, Victor; R Karl Rethemeyer & Eric Schoon (2018) Repli- cation data for: Crime, conflict and the legitimacy tradeoff:

Asal et al. 205

Explaining variation in insurgents’ participation in crime. Harvard Dataverse, V1. DOI: 10.7910/DVN/JT6GFR.

Asal, Victor; R Karl Rethemeyer & Eric Schoon (2019) Crime, conflict, and the legitimacy trade-off: Explaining variation in insurgents’ participation in crime. Journal of Politics 81(2): 399–410.

Avdan, Nazli & Gary Uzonyi (2017) V for vendetta: Govern- ment mass killing and domestic terrorism. Studies in Con- flict & Terrorism 40(11): 934–965.

Bale, Jeffrey (2017) The Darkest Sides of Politics, II State Ter- rorism, Weapons of Mass Destruction, Religious Extremism, and Organized Crime. London: Routledge.

Binder, Markus & Gary Ackerman (2021) Pick your POICN: Introducing the profiles of incidents involving CBRN and non-state actors (POICN) database. Studies in Conflict and Terrorism 44(9): 730–754.

Binder, Markus; Jillian Quigley & Herbert Tinsley (2018) Islamic State chemical weapons: A case contained by its context? Combating Terrorism Center CTC Sentinel 11(3) (https://ctc.usma.edu/islamic-state-chemical-weapons- case-contained-context/). Accessed 3 December 2021.

Bleek, Philipp (2017) When Did (and Didn’t) States Prolifer- ate? Chronicling the Spread of Nuclear Weapons. Cambridge: Belfer Center for Science and International Affairs.

Blum, Andrew; Victor Asal & Jonathan Wilkenfeld (2005) Non-state actors, terrorism, and weapons of mass destruc- tion. International Studies Review 7(1): 133–137.

Braithwaite, Alex (2013) The logic of public fear in terrorism and counter-terrorism. Journal of Police and Criminal Psy- chology 28(2): 95–101.

Bulletin of the Atomic Scientists (2019) Nuclear notebook multimedia page (https://thebulletin.org/). Accessed Octo- ber 2021.

Bunn, Matthew & Nicholas Roth (2017) The effects of a single terrorist nuclear bomb. Bulletin of the Atomic Scientists, 28 September (https://thebulletin.org/2017/09/ the-effects-of-a-single-terrorist-nuclear-bomb/. https://the bulletin.org/nuclear-notebook-multimedia). Accessed 6 December 2021.

Campbell, Blake & Amanda Murdie (2018) Keep the infor- mants talking: The pursuit and use of CBRN weapons by terrorist organizations. Studies in Conflict and Terrorism. Published online 27 December. DOI: doi.org/10.1080/ 1057610X.2018.1531517.

Carus, W Seth (2000) The Rajneeshees (1984). In: Jonathan Tucker (ed.) Toxic Terror: Assessing Terrorist Use of Chem- ical and Biological Weapons. Cambridge, MA: MIT Press, 115–138.

Curry, Adam (2015) Here are the ancient sites ISIS has dam- aged and destroyed. National Geographic, 1 September (https://news.nationalgeographic.com/2015/09/150901- isis-destruction-looting-ancient-sites-iraq-syria-archaeol ogy/). Accessed 6 December 2021.

Dahlberg, Stefan; Sören Holmberg, Rothstein Bo, Natalia Alvarado Pachon & Sofia Axelsson (2020) The Quality of

Government basic dataset, version Jan20. University of Gothenburg: The Quality of Government Institute (http://www.qog.pol.gu.se).Accessed 3 December 2021. DOI: 10.18157/qogbasjan20.

Danzig, Richard; Marc Sageman, Terrance Leighton, Lloyd Hough, Hidemi Yuki, Rui Kotani & Zachary M Hosford (2012) Aum Shinrikyo: Insights into How Terrorists Develop Biological and Chemical Weapons. 2nd edn. Washington, DC: Centre for a New American Security.

Early, Brian; Matthew Fuhrmann & Quan Li (2013) Atoms for terror? Nuclear programs and non-catastrophic nuclear and radiological terrorism. British Journal of Political Science 43(4): 915–936.

Fearon, James (1995) Rationalist explanations for war. Inter- national Organization 49(3): 379–414.

GDT (2019) Global Terrorism Database (shttps://www.start. umd.edu/gtd/). Accessed 31 December 2021.

Gleditsch, Nils Petter; Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg & Håvard Strand (2002) Armed con- flict 1946–2001: A new dataset. Journal of Peace Research 39(5): 615–637.

Gressang, Daniel (2001) Audience and message: Assessing terrorist WMD potential. Terrorism and Political Violence 13(3): 83–106.

Henschke, Adam & Legrand Timothy (2017) Counterterror- ism policy in liberal-democratic societies: locating the ethi- cal limits of national security. Australian Journal of International Affairs 71(5): 544–561.

Hutchinson, Steven & Pat O’Malley (2007) A crime-terror nexus? Thinking of some of the links between terrorism and criminality. Studies in Conflict Terrorism 30(12): 1095–1107.

Ivanova, Kate & Todd Sandler (2006) CBRN incidents: Polit- ical regimes, perpetrators, and targets. Terrorism and Polit- ical Violence 18(3): 423–448.

Ivanova, Kate & Todd Sandler (2007) CBRN attack perpe- trators: An empirical study. Foreign Policy Analysis 3(4): 273–294.

James Martin Center (2016) James Martin Center for Non- proliferation Studies (https://nonproliferation.org/). Accessed 31 December 2021.

Jenkins, Brian (1974) International Terrorism: A New Kind of Warfare. Santa Monica, CA: The Rand Corporation.

Joshi, Sharad (2020) How terrorist actors in Pakistan use nuclear weapons for political influence. Asian Security 16(2): 221–242. DOI: 10.1080/14799855.2019.15 82522.

King, Gary & Langche Zeng (2001) Logistic regression in rare events data. Political Analysis 9(2): 137–163.

Koehler, Daniel & Peter Popella (2018) Mapping far-right chemical, biological, radiological, and nuclear (CBRN) ter- rorism efforts in the west: Characteristics of plots and per- petrators for future threat assessment. Terrorism and Political Violence. Published online first 15 August 2018.

206 journal of PEACE RESEARCH 60(2)

(printed version 32(8): 1666–1690 (2020)). DOI: 10. 1080/09546553.2018.1500365.

Long, Scott & Jeremy Freese (2006) Regression Models for Categorical Dependent Variables Using Stata. College Sta- tion, TX: Stata.

Marshall, Monty; Keith Jaggers & Ted Robert Gurr (2011) Polity IV Project: Dataset Users’ Manual. Arlington, VA: Polity IV Project.

Mauroni, AL (2018) The new US strategy to tackle WMD terrorism is old wine in new wineskins. War on the Rocks, 14 December (https://warontherocks.com/ 2018/12/the-new-u-s-strategy-to-tackle-wmd-terrorism- is-new-wine-in-old-wineskins/). Accessed 6 December 2021.

McComb, Jonathan (2013) Closing Pandora’s box: The threat of terrorist use of weapons of mass destruction. Global Security Studies 4(1): 71–92.

Nemeth, Erik (2010) Market Value of Culture: Quantifying the Risk of Antiquities Looting. Santa Monica, CA: RAND Corporation.

O’Grady, Mary Anastasia (2002) Capitol Hill leftists side with Colombian terrorists. Wall Street Journal, 23 August (https://www.wsj.com/articles/SB103006768884072435). Accessed 6 December 2021.

O’Neil, Siobhan (2007) Terrorist Precursor Crimes: Issues and Options for Congress. Congressional Research Service Report CR34014. Washington, DC: Congressional Research Service.

Palfy, Arpad (2003) Weapon system selection and mass-casualty terrorism. Terrorism and Political Violence 15(2): 81–95.

Pape, RA (2003) The strategic logic of suicide terrorism. American Political Science Review 97(3): 343–361.

Pettersson, Therése & Kristine Eck Pettersson (2018) Orga- nized violence, 1989–2017. Journal of Peace Research 55(4): 535–547.

Pettersson, Therése & Peter Wallensteen (2015) Armed con- flicts, 1946–2014. Journal of Peace Research 52(4): 536–550.

Picard, Robert (1993) Media Portrayals of Terrorism: Functions and Meaning of News Coverage. Ames, IA: Iowa State Uni- versity Press.

Post, Jerold; Ehud Sprinzak & Laurita Denny (2003) The terrorists in their own words: Interviews with 35 incarcer- ated Middle Eastern terrorists. Terrorism and Political Vio- lence 15(1): 171–184.

Rapoport, David (2004) The Four Waves of Modern Terror- ism. In: Audrey Kurth Cronin & James Ludes (eds) Attack- ing Terrorism: Elements of a Grand Strategy. Washington DC: Georgetown University Press, 46–73.

Revill, James (2017) Past as prologue? The risk of adoption of chemical and biological weapons by non-state actors in the EU. European Journal of Risk Regulation 8(4): 626–642.

Schmid, Alex (2011) The literature on terrorism. In: Alex Schmid (ed.) The Routledge Handbook of Terrorism Research. New York: Routledge, 457–474.

Shelley, Louise & John Picarelli (2002) Methods not motives: Implications of the convergence of international organized crime and terrorism. Police Practice and Research 3(4): 305–318.

Sinai, Joshua (2005) Forecasting terrorists’ likelihood to embark on conventional to CBRN. International Studies Review 7(1): 151–153.

START – National Consortium for the Study of Terrorism and Responses to Terrorism (2018) Global Terrorism Database [1970–2015] (https://www.start.umd.edu/gtd). Accessed 6 December 2021.

Stern, Jessica (1999) The Ultimate Terrorists. Cambridge, MA: Harvard University Press.

Sunstein, Cass (2002) Terrorism and probability neglect. Journal of Risk and Uncertainty 26(2–3): 121–136.

Tannenwald, Nina (1999) The nuclear taboo: The United States and the normative basis of nuclear non-use. Interna- tional Organization 53(3): 433–468.

Teorell, Jan; Marcus Samanni, Sören Holmberg & Rothstein Bo (2011) The Quality of Government dataset, version 6 April 2011. University of Gothenburg, The Quality of Government Institute (https://www.gu.se/en/quality-gov ernment). Accessed October 2021.

Tokdemir, Efe & Seden Akcinaroglu (2016) Reputation of terror groups dataset: Measuring popularity of terror groups. Journal of Peace Research 53(2): 268–277.

Török, Thomas; Robert Tauxe, Robert Wise, John Liven- good, Robert Sokolow, Steven Mauvais, Kristin Birk- ness, Michael Skeels, John Horan & Laurence Foster (1997) A large community outbreak of salmonellosis caused by intentional contamination of restaurant salad bars. Journal of the American Medical Association 278(5): 389–395.

Tucker, Jonathan (2000) Toxic Terror: Assessing Terrorist Use of Chemical and Biological Weapons. Cambridge, MA: MIT Press.

White House Fact Sheet (2018) President Donald J Trump is taking action to keep weapons of mass destruction out of the hands of terrorists. 18 December (https://www.white house.gov/omb/fact-sheets/). Accessed October 2021.

Williams, Phil (2009) Criminals, Militias and Insurgents: Orga- nized Crime in Iraq. Carlisle, PA: Strategic Studies Institute, US Army War College.

VICTOR ASAL, PhD in Political Science (University of Maryland, 2003); Professor of Political Science and Director of the Center for Policy Research, University at Albany-SUNY (2003– ); current interests include gender and security, LGBTQ rights, characteristics of militant organizations, pedagogy. Most recent book: Insurgent Terrorism: Organizational Relationships and the Killing of Civilians (Oxford University Press, forthcoming, 2022).

Asal et al. 207

NAZLI AVDAN, PhD in Political Science (Duke University, 2011); Associate Professor of Political Science and Director of Graduate Studies, University of Kansas (2013– ); current interests include migration and border policies, the migration–security nexus, terrorism and terrorist groups, human trafficking and organized crime–terrorism nexus. Most recent book, Visas and Walls (University of Pennsylvania Press, 2019).

GARY ACKERMAN, PhD in Political Science (King’s College London, 2014); Associate Professor of Homeland Security, University at Albany-SUNY (2017– ); current interests include terrorism, counterterrorism, radicalization, terrorism and technology, terrorism and CBRN weapons. Most recent edited book: Jihadists and Weapons of Mass Destruction (CRC Press, 2009).

208 journal of PEACE RESEARCH 60(2)

<< /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles true /AutoRotatePages /None /Binding /Left /CalGrayProfile (Gray Gamma 2.2) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Warning /CompatibilityLevel 1.3 /CompressObjects /Off /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages false /CreateJobTicket false /DefaultRenderingIntent /Default /DetectBlends true /DetectCurves 0.1000 /ColorConversionStrategy /LeaveColorUnchanged /DoThumbnails false /EmbedAllFonts true /EmbedOpenType false /ParseICCProfilesInComments true /EmbedJobOptions true /DSCReportingLevel 0 /EmitDSCWarnings false /EndPage -1 /ImageMemory 1048576 /LockDistillerParams true /MaxSubsetPct 100 /Optimize true /OPM 1 /ParseDSCComments true /ParseDSCCommentsForDocInfo true /PreserveCopyPage true /PreserveDICMYKValues true /PreserveEPSInfo true /PreserveFlatness false /PreserveHalftoneInfo false /PreserveOPIComments false /PreserveOverprintSettings true /StartPage 1 /SubsetFonts true /TransferFunctionInfo /Apply /UCRandBGInfo /Remove /UsePrologue false /ColorSettingsFile () /AlwaysEmbed [ true ] /NeverEmbed [ true ] /AntiAliasColorImages false /CropColorImages false /ColorImageMinResolution 266 /ColorImageMinResolutionPolicy /OK /DownsampleColorImages true /ColorImageDownsampleType /Average /ColorImageResolution 175 /ColorImageDepth -1 /ColorImageMinDownsampleDepth 1 /ColorImageDownsampleThreshold 1.50286 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages true /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.40 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /ColorImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasGrayImages false /CropGrayImages false /GrayImageMinResolution 266 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Average /GrayImageResolution 175 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50286 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict << /QFactor 0.40 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /GrayImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasMonoImages false /CropMonoImages false /MonoImageMinResolution 900 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Average /MonoImageResolution 175 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50286 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict << /K -1 >> /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox false /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (U.S. Web Coated \050SWOP\051 v2) /PDFXOutputConditionIdentifier (CGATS TR 001) /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org) /PDFXTrapped /Unknown /CreateJDFFile false /Description << /ENU <FEFF005500730065002000740068006500730065002000530061006700650020007300740061006e0064006100720064002000730065007400740069006e0067007300200066006f00720020006300720065006100740069006e006700200077006500620020005000440046002000660069006c00650073002e002000540068006500730065002000730065007400740069006e0067007300200063006f006e006600690067007500720065006400200066006f00720020004100630072006f006200610074002000760037002e0030002e00200043007200650061007400650064002000620079002000540072006f00790020004f00740073002000610074002000530061006700650020005500530020006f006e002000310031002f00310030002f0032003000300036002e000d000d003200300030005000500049002f003600300030005000500049002f004a0050004500470020004d0065006400690075006d002f00430043004900540054002000470072006f0075007000200034> >> /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ << /AsReaderSpreads false /CropImagesToFrames true /ErrorControl /WarnAndContinue /FlattenerIgnoreSpreadOverrides false /IncludeGuidesGrids false /IncludeNonPrinting false /IncludeSlug false /Namespace [ (Adobe) (InDesign) (4.0) ] /OmitPlacedBitmaps false /OmitPlacedEPS false /OmitPlacedPDF false /SimulateOverprint /Legacy >> << /AllowImageBreaks true /AllowTableBreaks true /ExpandPage false /HonorBaseURL true /HonorRolloverEffect false /IgnoreHTMLPageBreaks false /IncludeHeaderFooter false /MarginOffset [ 0 0 0 0 ] /MetadataAuthor () /MetadataKeywords () /MetadataSubject () /MetadataTitle () /MetricPageSize [ 0 0 ] /MetricUnit /inch /MobileCompatible 0 /Namespace [ (Adobe) (GoLive) (8.0) ] /OpenZoomToHTMLFontSize false /PageOrientation /Portrait /RemoveBackground false /ShrinkContent true /TreatColorsAs /MainMonitorColors /UseEmbeddedProfiles false /UseHTMLTitleAsMetadata true >> << /AddBleedMarks false /AddColorBars false /AddCropMarks false /AddPageInfo false /AddRegMarks false /BleedOffset [ 9 9 9 9 ] /ConvertColors /ConvertToRGB /DestinationProfileName (sRGB IEC61966-2.1) /DestinationProfileSelector /UseName /Downsample16BitImages true /FlattenerPreset << /ClipComplexRegions true /ConvertStrokesToOutlines false /ConvertTextToOutlines false /GradientResolution 300 /LineArtTextResolution 1200 /PresetName ([High Resolution]) /PresetSelector /HighResolution /RasterVectorBalance 1 >> /FormElements true /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles true /MarksOffset 9 /MarksWeight 0.125000 /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PageMarksFile /RomanDefault /PreserveEditing true /UntaggedCMYKHandling /UseDocumentProfile /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ] /SyntheticBoldness 1.000000 >> setdistillerparams << /HWResolution [288 288] /PageSize [612.000 792.000] >> setpagedevice