HLSS522Wk2

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=ftpv20

Terrorism and Political Violence

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/ftpv20

Islamic Extremism and CBRN Terrorism

Wesley S. McCann

To cite this article: Wesley S. McCann (2023) Islamic Extremism and CBRN Terrorism, Terrorism and Political Violence, 35:3, 594-614, DOI: 10.1080/09546553.2021.1964964

To link to this article: https://doi.org/10.1080/09546553.2021.1964964

View supplementary material

Published online: 07 Sep 2021.

Submit your article to this journal

Article views: 947

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Islamic Extremism and CBRN Terrorism Wesley S. McCann

Department of Criminology, Law and Society, George Mason University, Fairfax, Virginia, USA

ABSTRACT The extant literature on the relationship between religious ideology and CBRN weapons pursuit lacks a strong empirical basis. To address this, the current study utilizes a new dataset—the CBRN Terrorism Database (CTD)—to evaluate whether Islamic actors are more or less likely to pursue CBRN weapons. Binary logistic regression models are used on several different constrained subsam- ples to assess the relationship between ideology, group-, and country-level factors and CBRN weapons pursuit. Across the models, the strongest predictors of pursuit are whether the actor is an individual, is inspired by an Islamic ideology, and resides in a country with low levels of ethnic fractionalization. Other factors are important under specific contexts, but Islamic groups are significantly more likely to pursue CBRN weapons as well. Implications and recommendations are discussed.

KEYWORDS CBRN terrorism; terrorism; Islam; Islamic extremism

Introduction

The extant literature on chemical, biological, radiological, and nuclear (CBRN) terrorism is vast, but much of it is speculative and lacks strong empirical findings. The studies that are empirical tend to focus on each type of CBRN terrorism committed by all types of groups; thus failing to provide nuance where it is often needed. Nevertheless, the current empirical literature is mostly mixed as to the impact religious ideology has on CBRN weapons pursuit and this finding transcends the type of database used. However, almost nothing empirical has been executed on Islamic actors and their pursuit of CBRN weapons. It can be argued that the apocalyptic visions of many religious-inspired actors—especially groups such as ISIS, Al Qaeda, and the Taliban—coupled with access to advanced weaponry makes them particularly prone to pursue these weapons. Another argument is that religious terrorism is overblown and that we should focus on single-issue terrorists instead.1

Because both arguments have merit, this study delves deeper into the religion and CBRN terrorism debate by utilizing a novel dataset that tracks CBRN terrorism pursuit from 1990–2016, while also including cases by non-CBRN terrorists. Most cases involving “religious” actors are in fact Islamic extremists. As such, no study on “religious CBRN terrorism” can truly be conducted. Since prior studies have concluded that “religious” actors are less dangerous than other types of actors or are no different than them in their pursuit of CBRN weapons, this study seeks to determine if this position holds for Islamic actors.

Thus, this study adds value to the extant literature on CBRN weapons pursuit through: (1) use of a novel merged dataset that includes non-CBRN cases; (2) an analysis of the influence of Islamic ideology on CBRN weapons pursuit; (3) the ability to disaggregate models by actor type and ideology; and (4) the use of expanded group and country-level effects. The following section covers the available literature on the predictors of CBRN terrorism.

CONTACT Wesley S. McCann [email protected] Department of Criminology, Law and Society, George Mason University, 4400 University Drive, 4F4, Enterprise Hall, Rm 354, Fairfax, VA 22030, USA

Supplemental data for this article can be accessed on the publisher’s website.

TERRORISM AND POLITICAL VIOLENCE 2023, VOL. 35, NO. 3, 594–614 https://doi.org/10.1080/09546553.2021.1964964

© 2021 Taylor & Francis

Purpose of study

Given the extant literature, this study needs to delve deeper into “religious” CBRN terrorism. Since the data and prior scholarship indicate that an overwhelming amount of “religious” terrorism is really just Islamic extremism, this study will examine Islamic extremism specifically.2 The central purpose of this study is to analyze Islamic extremism and the pursuit of CBRN weapons, and whether the correlates for their pursuit of said weapons are any different than what is outlined in the extant literature.

This study is justified for the following reasons. First, the extant CBRN terrorism literature obfuscates any real analysis of the role of religion generally, and Islamic extremism specifically, by way of classification.3 Second, many studies have included unreliable cases within their analyses, are focused only on one type of CBRN terrorism or agent, and have relied on datasets with extremely high levels of unknown actors. These limitations impact the inferences that follow. This will be unpacked in the literature review. Third, CBRN terrorism studies often lack proportionality in analyses. Put another way, an overwhelming majority of “cases” involve chemical weapons. However, the actors most likely to pursue more serious agents are religious actors; specifically Islamic extremists (see descriptives).4 As such, the literature ignores the severity of the threat posed by focusing on the odds of “pursuit” or “attack” in general without further specification. This study seeks to overcome these deficiencies. The following serve as the directional hypotheses for this study:

H1: Groups that are more transnational in focus will be more likely to pursue CBRN and BRN weapons.

H2: Actors motivated by an Islamic ideology will be no more likely to pursue CBRN and BRN weapons, as compared to non-Islamic actors.5

H3: Individual actors will be more likely, as compared to groups, to pursue CBRN and BRN weapons.

H4: Ethnic fractionalization will have a significant negative relationship with CBRN and BRN weapons pursuit. This means that each of these types of pursuit are less likely to occur in societies marked by high levels of ethnic diversity.

H5: Polity will have a significant negative relationship with CBRN and BRN weapons pursuit. This means that each of these types of pursuit are less likely to occur in more democratic societies.

Review of literature

“Religion” and CBRN terrorism

The extent to which religion plays a role in pushing terrorist groups to obtain or use CBRN weapons has been the subject of debate amongst terrorism scholars for decades. Some argue that is does not play that big of a role,6 or that actors motivated by other ideologies are just as if not more dangerous.7

Others contend that religious beliefs—especially the apocalyptic variety—enables actors to evade the psychological boundaries that are inherent to other ideologies.8 Thus, religious actors can rationalize almost anything—no matter how horrific—because normal constraints are missing.9 One such Defense Threat Reduction Agency report argued that religious extremists and apocalyptic groups—namely jihadists—were the most likely to use CBRN weapons.10 No empirical scholarship has tested this assertion though.

Regardless, many have wrestled with the question concerning whether terrorists with non- religious (or apocalyptic) motivations are just as if not more likely to use CBRN weapons.11

Several scholars argue that CBRN terrorism is a strategic choice and is not unique to religious actors.12 Others also note that the brand of religion may be of significance, in that more nuance here

TERRORISM AND POLITICAL VIOLENCE 595

may provide more insight when examining “religious” CBRN terrorism.13 However, most studies to date do not focus on the specific brand of religion, or which type of religious actor is most likely to pursue CBRN weapons.

One study by Asal and colleagues found that religious groups were not more likely to pursue CBRN weapons.14 However, almost 75 percent of all cases involved unknown actors; potentially decreasing the ability to detect the influence of “religion” on the pursuit of CBRN weapons.15 Ivanova and Sandler demonstrated that cults and groups that were transnational were more likely to utilize CBRN weapons, but religious fundamentalists were not, despite numerous religious groups (e.g. Al Qaeda) acting transnationally.16 When looking at bioterrorism, Jessica Stern contended that religious and other far-right anti-government actors are the most likely to use bioweapons.17 Recent scholarship on bioterrorism, however, has found that single-issue terrorists were not significantly different than religious actors in terms of their odds of pursuing bioweapons.18

Many studies contend that single-issue terrorists are the biggest concern. Guarrieri and Meisel found that older unemployed individuals promoting a single-issue were more likely to pursue chemical and biological weapons, compared to other actors.19 However, this study used several different data sources whereby demographic profiles were derived only from actors in the U.S., and the dataset was heavily focused on individuals; including some who were a part of cells or groups. Furthermore, group-oriented measures were not used, nor country-level effects for obvious reasons, and the models were not disaggregated on the type of actor (e.g. individual vs. group). While this study was able to include a measure on Islamic ideology, it was only in comparison to single-issue actors, which are overrepresented in the U.S. and constitute most of the “individual” cases; hence why they still found that “group members” will be more likely to pursue CB weapons. A simulation of CBRN weapons pursuit and use by Clark and Guarrieri found that cells and single-issue actors were more likely than others to pursue bioweapons, whereas far-left and single-issue actors were the most successful in obtaining and using CBRN weapons in general.20 While this study provides immense insight into the use of novel techniques to study CBRN terrorism, it also suffered from many of the same issues as the former study by Guarrieri and Meisel; namely the inability to take country or environmental factors into account.

The most comparable study to the current one, examined CBRN terrorism by religious actors using both CBRN and non-CBRN cases. McCann disaggregated CBRN cases by using BRN cases as the dependent variable in many of the models given the reality than an overwhelming number of cases are chemical weapons. The study found that religious actors and individuals were more likely to pursue CBRN weapons as compared to other actors, and this was true of BRN weapons too.21 This study provided a compelling theoretical and empirical basis for allowing anti-abortion terrorists to be included in either the religious or “other” camp depending on the underlying motivation while getting rid of the single-issue terrorism category. Here, single-issue terrorists were not significantly more likely to pursue CBRN weapons.22 The study did conclude, however, that a majority of its “religious” cases seemed to be Islamic extremism, and much of the literature’s contentions about CBRN weapons pursuit may be deceptive given the overreliance on chemical weapons pursuit.23 As such, modeling “religious” CBRN terrorism specifically, and CBRN terrorism in general, may be a bit misleading; thus warranting a more robust analysis of the relationship between CBRN terrorism and Islam as well as researchers partitioning their models by excluding chemical weapons cases.

The role of Islam

While the aforementioned study contended that most religious CBRN cases were likely Islamic extremist groups, a robust exegesis was not conducted, nor was anything included in the models that could test such a relationship. Guarrieri and Meisel did include a “Islam” measure, but found that single-issue individuals were still more likely to pursue chemical and biological weapons than those adopting an Islamic ideology.24

596 W. S. MCCANN

While the empirical literature on Islam and CBRN weapons pursuit is extraordinarily thin, the historical analysis of decision-making process amongst Islamic extremists and jurists is quite a bit more robust. This is especially the case after 9/11, although many studies continue to fail to focus specifically on the unique motivations and intentions of various actors.25 Caves and Carus have argued that a main reason Islamic extremists pose such a large threat is their ability to get their hands on stockpiles of CBRN weapons in countries with weak governments or that are characterized by civil conflict.26 This was the former Soviet Union initially, but has also evolved to include Pakistan, Mali, Iraq, and now Syria. There is a fair amount of historical evidence for such fears, especially given the behavior of Al Qaeda.27 Nevertheless, most of these attempts resulted in abject failures, and none of the extant case studies really go into detail as to the reasoning behind the attempted acquisition or use of CBRN weapons.

It must also be stated that Islamic law may justify the use of CBRN weapons. Under the purview of bin Laden’s authority, a Saudi cleric—Nasir bin Hamd al-Fahd—issued a fatwa justifying the use of such weapons if this is the only way the “infidels can ... be repelled.”28 In fact, this fatwa has not be recanted, despite even the Ayatollah Khamenei arguing that they are illegal under Islamic law.29

Another well-known Islamic writer and strategist—Abu Musab al-Suri—argued that bin Laden should have used WMDs instead, and that the only way to fight the West on a level playing field is through the use of WMDs.30 It is not surprising then that Al Qaeda has been pursuing CBRN weapons for decades, even though such pursuits have often been quite crude or unsuccessful.31 Despite the large techno- logical gap these groups are unlikely to cross, the intent to acquire and use these weapons is there. For example, after 9/11, Al Qaeda pivoted away from the defensive use of such weapons to their use for offensive first-strike purposes.32 However, this may only be true of large groups with immense amount of capital already, as smaller groups may utilize conventional weapons to carry out smaller scale attacks.

Irrespective of group differences, it must also be conceded that Islamic groups demonstrate rationality and strategic decision-making in pursuing these weapons, namely deterrence value against the United States, but that decision-making process is not devoid of irrationality or magical apoc- alyptic thinking.33 In fact, much of the reasoning of contemporary groups has been analogous to the medieval aims of Islam—the return of the Caliphate, expansionist ambitions, and the return of the glory of Islam.34 Similar to the aforementioned fatwa, almost all of the support for the pursuit of CBRN weapons comes directly from the Qu’ran, ahadith (teachings of the prophet Muhammad), and Islamic just war school of thought which emphasizes reciprocity of actions (“eye for an eye”).35 Today, this manifests itself more squarely in the realm of puritanical edicts and the removal of unbelief or jahiliyya, which cannot coexist with the aims of jihad. Much of this has been echoed by Sunni Muslim scholars Hasan al-Banna, Sayyid Qutb, and Abu A’la Mawdudi. Thus, groups that portend a return to the old ways coupled with the reconquest of formerly Muslim lands and the Western world have enormous ends that would justify almost any means; especially in the face of unmatched military superiority by Western countries. Hence the impetus to “secure the triumph of the dar al-Islam (abode of Islam) over the dar al-kufr (abode of unbelief).”36

After the invasion of Afghanistan by American forces, Al Qaeda was severely disrupted. This has led unfortunately to the mass decentralization and thus, globalization, of the movement. Subsequently, issues concerning identity and the need for revenge encompassed more real estate within the jihadist mindset.37 Not all Islamic extremists will support the use of CBRN weapons, and may not need to as suicide bombings allow for cruder forms of attack that satisfy groups short- term goals and resource constraints.38 Furthermore, there are significant pros and cons for groups to weigh in deciding whether to pursue these weapons. For one, it raises the bargaining power and profile of these groups.39 However, amongst the few actual WMD attacks by terrorists, none have been emulated; thus suggesting that the operational advantages may not be that great in terms of long-term strategy.40 Also of note, the materials needed for most CBRN weapons are extremely cost prohibitive and difficult to get let alone weaponize and use. Put simply, the return on investment may not be that great for many groups, unless mass destruction or apocalyptic thinking

TERRORISM AND POLITICAL VIOLENCE 597

is part of their modus operandi.41 Such was the case for the infamous cult Aum Shinrikyo, but it not necessarily the case for many Islamic groups, despite their predilections of world domination. However, many chemical agents can be—and have been—produced and used in a cost effective manner; thereby justifying the need to study chemical weapons pursuit and use separately.

It seems that only technical hurdles delay many Islamic extremists’ efforts. Nevertheless, most of these qualitative studies are hyper focused on Al Qaeda, and for good reason.42 This may, however, bias our understanding of the role of Islam in pushing actors to pursue CBRN weapons. Islamic extremist groups are quite resilient and have shown their capacity to survive over long periods of time, even in face of strong opposition; so there is some rationality to fear of their eventual acquisition.43

While the current study is not an exegesis of Islamic teachings, law, or scripture, the empirical question as to whether Islamic actors pursue CBRN weapons more often than others—including non-religious actors—has yet to be fully answered. Contemporary evidence on terrorism world- wide demonstrates that for the past two decades Islamic extremism has been the most lethal form of terrorism, has been continually growing, and has globalized extensively since 9/11.44 These facts coupled with the ideological predispositions of its adherents and lofty and unrealizable aims seemingly make Islamic extremism the most problematic forms of religious terrorism, and thereby, the most likely to pursue CBRN weapons. More scholarship is needed to affirm this, however.

Role of contextual factors

Several contextual factors have been found to be relevant when explaining CBRN weapons pursuit. For example, the recent study by McCann found that ethnic and religious fractionalization—or level of diversity—were relevant contextual factors in some of the models explaining religious group pursuit.45

These measures are salient since homogenous societies are more likely to squash dissent and inter- mediate levels of fractionalization may provide an environment for actors to carry out their attacks.46

However, Blomberg and colleagues argue that research using ethnic fractionalization is not robust and needs further examination.47

Governance is also an important indicator, namely in the form of polity scores. In fact, this measure has been found to significantly explain biological weapons pursuit,48 fatalities,49 and group longevity.50

Asal and colleagues found that authoritarian countries are also more likely to be victims of CBRN terrorism,51 while Ivanova and Sandler found that wealthier democratic countries experience more CBRN terrorism; although only by lone actors.52 In McCann’s analysis, polity was only significant in very specific scenarios.53 Related to governance is wealth. Hou and colleagues found that income per capita was positively related to the use of CBRN weapons by certain groups,54 and has also been valuable in other studies on group success and lethality.55 However, Asal and colleagues have shown that a country’s level of wealth (energy consumption per capita) did not explain actor’s decisions to pursue CBRN weapons, even though economic interconnectedness with outside world did,56 and even income per capita was found to only explain the use of CBRN weapons by certain groups in the study by Ivanova and Sandler.57 Furthermore, McCann’s recent analysis found no significant relationship between a country’s wealth and CBRN or BRN weapons pursuit,58 and another recent analysis using similar data and techniques to this study found that GDP did not significantly relate to biological weapons pursuit across any of the models.59

As such, further analysis as to whether ethnic and religious fractionalization, indicators of wealth, and polity need to be examined further to determine whether they play a role in explaining CBRN weapons pursuit by Islamic actors. It could be the case that these factors are more important in explaining Islamic-inspired terrorism than in models are focused on religious actors in general, given that Islamic extremism is concentrated in countries with lower polity scores, weaker economies, and less ethnic and religious diversity.

598 W. S. MCCANN

Methods

This study uses data from several sources to create a database of CBRN and non-CBRN cases for both individuals and groups—CBRN Terrorism Database (CTD). While the merging of several open- source databases is rare, it has been carried out by others studying CB terrorism before and is a popular technique in the field of epidemiology.60 These “incidents” were then aggregated to create cases in the year-actor format for years 1990–2016. As such, this study utilizes a case-control design. These types of design are helpful for examining risk factors within a matched sample in hindsight, and is also a popular technique within the medical profession.61 Most importantly, case control designs enable the study of some behavior or occurrence without sampling on the dependent variable; in this case, only CBRN events. These designs do not constitute random assignment, yet they still provide some advantages to studying CBRN terrorism without a control group. For this study, contextual, actor, country, and ideological factors are the theoretically relevant measures for examining CBRN weapons pursuit. CBRN terrorism is extremely rare, as is terrorism in general. Case-control designs provide researchers the ability to make inferences about numerous risk factors using rare event data. While not perfect, the limitations of this design are minimized by the integration of several data quality measures within the respective data to filter out unreliable cases (see Methodological Appendix).

The primary data source is the Profiles of Incidents Involving CBRN and Non-State Actors (POICN) Database. This source covers CBRN terrorism cases involving non-state actors from 1990– 2016. This source is also more desirable over other CBRN terrorism sources because of its robust credibility, inclusion, source evaluation, and uncertainty criterion. For a more in depth overview of this data’s strengths, please see Binder and Ackerman’s exegesis.62 Nevertheless, these quality control measures provide something akin to a total survey of error metric, which is valuable when using open source data.63 The second data source is the Extended Data on Terrorist Groups (EDTG) from Hou and Colleagues.64 This is derived from several public sources, primarily the Global Terrorism Database (GTD), to create group and country-level data in long format, whereas the GTD does not. It also includes measures of group ideology, duration, base country, attack diversity, peak size, as well as numerous country-level factors such as geographical features, polity, and metrics of diversity. Lastly, non-CBRN terrorism cases by individuals were obtained via the GTD. These were constrained to only cases that: (1) that occurred from 1990–2016; (2) that were not in doubt as to their terrorism label (e.g., doubt_terr = 0); (3) that did not involve CBRN weapons or agents; (4) that occurred in the same countries as the single-actor POICN cases; and (5) that were committed by unaffiliated individuals.65 When added together, the new database was comprised of 870 total cases in long format. For a more robust explanation of how these data were merged to create the CTD, please see the Methodological Appendix.

Variables

Dependent variables Two outcome measures were used. CBRN denotes whether the actors pursued CBRN weapons (“1”=yes). Also, BRN denotes whether the actors pursued biological, nuclear, or radiological weapons during the period studied. For both of these outcomes, “pursuit” constitutes everything from a plot to actual use of the agents. Also, the exclusion of chemical weapons vis-à-vis the second outcome measure is important here because, based on descriptive data, many religious-inspired actors pursue the later more often than non-religious inspired actors. Furthermore, BRN weapons have the capacity to be exponentially more dangerous and deadly than chemical weapons, so focusing on their pursuit is essential; especially when constraining cases down to only Islamic extremists. This is why McCann bifurcated their models according to agent pursued as well.66

TERRORISM AND POLITICAL VIOLENCE 599

Independent variables Actor-oriented measures. Each of the independent variables came from merging of EDTG and POICN. It must be noted that several variables are only coded for groups; allowing some models to be disaggregated by actor type (e.g. groups vs. individual). As such, individual indicates whether the actor in question is an individual (1) or group (0). One study found that lone actors are actually more dangerous than groups and carry out CBRN terrorism for reasons others than religion,67 whereas another study on far-right terrorism discovered that cases were predominately indiscriminate, and were carried out by lone middle-aged, well-educated males motivated by non-religious ideologies.68

A recent analysis by McCann also concluded that lone actors were more likely to pursue CBRN weapons than groups,69 and the same finding was true of bioweapons pursuit as well in a different study.70

Ideology is a categorical measure that classifies actors into three ideological groups- “Other” (0), “Nationalism” (1), and “Islam.”71 The “Other” category includes far-right, far-left, other religious actors, and unknown motivations. The CTD contains large amounts of nationalistic and religiously motivated actors, however, recent scholarship by McCann has argued that the latter are overwhel- mingly Islamic.72 Operationalizing the variable in this manner allows for ideological comparisons, and the “other” categories also contain a fair amount of “single-issue” terrorism due to the inclusion of anti-abortion actors along with far-right actors. However the “single-issue” label lacks ideological specificity, despite their grievances being narrow.73 In fact, a fair number of single-issue actors adopt a broader ideological platform over time, and others are often consigned to this category if they don’t fit the religious or ethno-nationalist molds.74 As such, this “other” category will contain all of the single-issue terrorists along with other categories that are not as ideologically refined or are unknown. In later models, ideology will be swapped for Islam to test a more absolute effect of having this specific ideology on pursuit. Islam denotes whether the actor was motivated, specifically, by an Islamic ideology. This measure was constructed using each of the forgoing data sources. Actors coded as being religious (“1”), where flagged for review. Case notes, sources, and motivation were examined to discern whether an Islamic ideology was present. For more information on how this measure was created, consult the Methodological Appendix.

TCH denotes whether an actor seeks a territory change (“1”=yes). All attack-related measures are the aggregated amounts provided over this time. Transnational provides the percentage of an actor’s attacks that are transnational during this period. Prior scholarship determined that increases in the share of attacks that are transnational increase the chances of the group’s demise.75 However, this measure is only used in models focusing specifically on groups, due to a lack of data on individuals. Close to 55 percent of groups have “0 percent” for this metric, whereas only 23 percent of groups have a transnational share of 10 percent or more; and only 10 percent of groups with 23 percent or more. Thus, this measure is recoded as “none,” “some” (1 percent through 9 percent), and “significant” (greater than or equal to 10 percent) in terms of transnational activity. Due to linearity issues total attacks and total casualties are used to compute a casualty rate measure—CPA—which denotes the number of casualties per attack.76 All of these measures were used in the analysis by McCann as well.77

Country-oriented measures. Country-level measures that vary (e.g. economic and political) are coded as the initial values for when the actor or group became active. Measures that do not vary (e.g. elevation, terrain) are reported the same across time, and during aggregation. If more than one country serves as the base of operations for the actor, then an average score is calculated for each country-level measure.

Polity scores are included within the EDTG but are derived from the Polity5 Project.78 Here, scores for each country range from authoritarian (−10) to democratic (10).79 The logarithm of GDP is used here. Thus, this study will explore whether this measure explains pursuit with this novel dataset. Several additional country-level effects are included; both demographic and geographical. Ethnic fraq. and religious fraq. indicate the level of fractionalization within a given country. This is presented as a percentage, with higher values representing more diversity.80 However, each study to date has relied

600 W. S. MCCANN

on transformations that are difficult to interpret and have only found that intermediate levels of ethnic fractionalization relate to survival.81 Furthermore, it is unclear what “intermediate” levels are, as no demarcations are ever provided. Given this reality, the need for parsimony, and the fact that squared measures negatively affected model convergence once the data set it partitioned down to only individuals or groups (see Analytic Strategy), these are transformed into ordinal measures. Each is operationalized into quartiles, since each measure is a percentage from .000 to 1.00. These quartiles approximate three equal (e.g. .333, .667) categories for each metric; defined as “low,” “intermediate,” and “high” levels of fractionalization.

Elevation provides the average elevation of a given country in meters, whereas tropics provides a percent estimate of the entire landmass within a given country that is tropical. These are located within the EDTG, but are originally provided by Gallup and colleagues.82 Landlock also denotes whether a country is landlocked or not (1=“yes”). These measures are useful to determine if geogra- phical heterogeneity plays a role in actor decision-making, as some contend that tropical and mountainous terrains provide more safe havens for terrorist bases, while actors operating in land- locked nations may struggle to import or export their requisite resources.83 Again, McCann included these measures in his recent study as well, and found that these factors play a small role in explaining pursuit in limited contexts.84

Study descriptives Close to 11 percent of all cases involved some form of CBRN weapons pursuit, with about half that for just BRN weapons pursuit (see Table 1). Most actors within the dataset are groups (76 percent), and most are motivated by nationalistic (31.7 percent) or Islamic (28.3 percent) ideologies. Groups adopted a territory change goal close to 30 percent of the time, and most groups were domestic as only some carried out transnational attacks. Casualties per attack (CPA) had a mean of 11.12 for groups as well.

Table 1. Descriptives

Variable n valid cases mean valid %

Individual 208 870 23.9 percent Pursuit (CBRN) 96 870 11.0 percent Pursuit2 (BRN) 51 870 5.9 percent Ideology 863

Other (ref) 348 40.0 percent Nationalism 276 31.7 percent Islam 246 28.3 percent

CPA 626 11.1 Ethnic Fraq. 858

Low 220 25.6 percent Intermediate 459 53.5 percent High (ref) 179 20.9 percent

Religious Fraq 860 Low 334 38.8 percent Intermediate 273 31.7 percent High (ref) 253 29.4 percent

Elevation 862 708.1 Tropics 862 0.276 Landlock 78 869 9.0 percent TCH 189 634 29.8 percent Transnational 621 0.089

None 333 53.6 percent Some 151 24.3 percent Significant 137 22.1 percent

Polity 895 5.14 GDP(log) 829 26.6

TERRORISM AND POLITICAL VIOLENCE 601

Looking at country-level factors, most countries were characterized by intermediate levels of ethnic fractionalization (53.5 percent), with a plurality having low levels of religious fractionalization (38.8 percent). On average, countries were slightly tropical (27.6 percent), had a mean elevation of 708 meters, and were slightly less than democratic (5.1). Nevertheless, a majority of countries were considered democracies (having a polity score of 6 or more). Thus, numerous authoritarian countries are included in this analysis.

Analytic strategy

This study uses twelve binary logistic regression models to assess the influence of contextual, country, ideological, and actor-level factors on CBRN weapons pursuit, with a specific focus on Islamic actors. As such, models 1–6 include the measure ideology to discern whether Islamic extremists are more likely to pursue these weapons compared to other types of actors. Models 1 and 2 include all actors and are regressed on both outcome measures: CBRN and BRN weapons pursuit. Models 3 and 4 are regressed on the same outcomes but include only groups; not individuals. Models 5 and 6 operate similarly to models 3 and 4 but also include additional group-level measures that have more missing values. Models 7–12 subsequently operate in the same exact manner as models 1–6, but ideology is swapped for Islam to allow for a more robust conclusion as to whether having this ideology is a risk factor compare to all other ideologies.

Results

Each model was significant, the correct classification of cases ranged from 89.0 to 96.1 percent, and the models ranged in their discriminative ability (AUC) from modest to excellent. Across the models, individual actors, Islamic-inspired actors, and actors operating in less ethnically diverse societies were significantly more likely to pursue CBRN and BRN weapons.

In models 1–2, only individual was significant (see Table 2). This model included all cases within the dataset. Here, individuals have 2.5 times the odds of pursuing CBRN weapons, but 4.1 times the odds of pursuing BRN weapons.

Table 2. Pursuit by all actors

Model 1 Model 2

Variable B S.E. p Exp(B) B S.E. p Exp(B)

Constant −3.084 0.608 0.000 0.046 −3.738 0.769 0.000 0.024 Individual 0.933 0.374 0.013 2.543 1.428 0.484 0.003 4.170 Ideology 0.169 0.158

Nationalism −0.099 0.346 0.775 0.906 0.106 0.473 0.823 1.112 Islam 0.439 0.283 0.121 1.550 0.654 0.363 0.071 1.924

Ethnic Fraq. 0.331 0.567 Intermediate 0.613 0.440 0.164 1.846 0.587 0.562 0.296 1.798 Low 0.552 0.393 0.160 1.737 0.331 0.508 0.514 1.393

Religious Fraq 0.672 0.965 Intermediate −0.127 0.410 0.756 0.880 0.141 0.529 0.790 1.151 Low 0.159 0.370 0.667 1.173 0.092 0.476 0.848 1.096

Elevation 0.000 0.000 0.277 1.000 0.000 0.000 0.676 1.000 Tropics −0.334 0.435 0.443 0.716 −0.627 0.610 0.304 0.534 Landlocked −0.744 0.730 0.308 0.475 0.160 0.808 0.843 1.173 Polity −0.013 0.025 0.594 0.987 −0.041 0.031 0.191 0.960 AUC 0.674 0.71 Classification 89 93.9 Nagelkerke R-square 0.082 0.09 Chi-square 34.948 28.232 2LL 544.593 355.865 df 11 11 N 836 836

602 W. S. MCCANN

Once cases are constrained to only groups in models 3–4, many of the effects change (Table 3). Ideology was significant in model 3 and 4, but the only significant difference between Islamic extremists and other groups occurred in model 4; meaning Islamic groups are only significantly more likely to pursue BRN weapons (4.63 time the odds) compared to “other” groups. Ethnic fractionalization and religious fractionalization became significant in model 3 but not 4. This means that, when examining CBRN weapons pursuit, countries with low levels of ethnic diversity have 3.6 times the odds of having groups who pursue CBRN weapons, when compared to countries with high levels of ethnic fractionalization. Countries with low and intermediate levels of religious fractionalization (e.g less diversity) are less likely to have groups pursue CBRN weapons.

When more factors are added to the models constrained on groups, several factors change (Table 4). Ideology again is significant, in that Islamic groups have 11.7 times the odds of pursuing BRN weapons (model 6), but this is only marginally significant in model 5. Ethnic fractionalization was significant in both models, with countries with low levels of ethnic fractionalization being more likely to have a group pursue both CBRN and BRN weapons, compared to countries with high levels. It was also true that intermediate levels were significant different as well, but only when BRN weapons is the outcome measure. It seems that the addition of more group-level measures mediates the importance of these country-level effects. Here, religious fractionalization falls out of significance, but elevation becomes significant in explaining CBRN pursuit (model 5), and tropics comes close in model 6. Put another way, for every unit increase (in meters) in average elevation, there is a significant .1 percent increase in odds of groups pursuing CBRN weapons.

The added factors also found significance as well. The desire for territory was not significant, but GDP(log) was marginally significant. More importantly, transnational was also significant in both model 5, in that groups with any level of transnational behavior had 6.7 to 8.5 times the odds of pursuing CBRN weapons. However, this measure could not be included for BRN weapons because it was too perfect a predictor; almost every pursuit of BRN weapons was by groups with some transnational behavior. Lastly, casualties per attack was marginally significant in explaining CBRN weapons pursuit, but significant in explaining BRN pursuit. Again, the exclusion of chemical weapons adds clarity to the model in mediating the effects of CPA and ideology.

Table 3. Pursuit by All Groups

Model 1 Model 2

Variable B S.E. p Exp(B) B S.E. p Exp(B)

Constant −2.900 0.606 0.000 0.055 −4.404 1.184 0.000 0.012 Ideology 0.032 0.009

Nationalism −0.410 0.413 0.321 0.664 0.151 0.636 0.813 1.162 Islam 0.646 0.411 0.116 1.908 1.533 0.606 0.011 4.630

Ethnic Fraq. 0.027 0.076 Intermediate 0.957 0.544 0.079 2.604 1.865 0.845 0.027 6.456 Low 1.298 0.488 0.008 3.661 1.665 0.789 0.035 5.283

Religious Fraq 0.032 0.594 Intermediate −1.123 0.460 0.015 0.325 −0.635 0.687 0.356 0.530 Low −1.064 0.466 0.023 0.345 −0.675 0.708 0.341 0.509

Elevation 0.000 0.000 0.318 1.000 0.000 0.001 0.724 1.000 Tropics −0.546 0.513 0.288 0.580 −0.954 0.801 0.234 0.385 Landlocked −1.253 0.961 0.192 0.286 0.455 1.011 0.653 1.576 Polity 0.012 0.030 0.684 1.012 −0.004 0.038 0.915 0.996 AUC 0.701 0.769 Classification 92.2 96 Nagelkerke R-square 0.103 0.014 Chi-square 28.019 22.125 2LL 316.025 188.05 df 10 10 N 628 628

TERRORISM AND POLITICAL VIOLENCE 603

Once ideology was swapped for Islam, most of the effects did not change across models 7–12. In models 7 and 8 (Table 5), which included all actors again, individual was significant once more. Here, individuals have 2.6 and 4.0 times the odds of pursuit CBRN and BRN weapons respectively. However,

Table 4. Pursuit by All Groups with additional measures

Model 5 Model 6

Variable B S.E. p Exp(B) B S.E. p Exp(B)

Constant −12.504 4.069 0.002 0.000 −10.691 5.308 0.044 0.000 Ideology 0.012 0.010

Nationalism −0.642 0.647 0.321 0.526 0.771 1.072 0.472 2.163 Islam 0.976 0.550 0.076 2.654 2.466 0.976 0.011 11.770

Ethnic Fraq. 0.032 0.049 Intermediate 1.017 0.640 0.112 2.765 1.785 0.912 0.050 5.961 Low 1.482 0.573 0.010 4.400 2.042 0.833 0.014 7.707

Religious Fraq 0.740 0.777 Intermediate −0.528 0.736 0.473 0.590 −0.627 0.991 0.527 0.534 Low −0.304 0.736 0.680 0.738 −0.652 0.942 0.489 0.521

Elevation 0.001 0.000 0.042 1.001 0.000 0.001 0.999 1.000 Tropics −0.556 0.663 0.402 0.574 −2.397 1.333 0.072 0.091 Landlocked −0.742 0.963 0.441 0.476 0.404 1.170 0.730 1.497 Polity −0.059 0.036 0.101 0.942 −0.085 0.045 0.060 0.918 GDP(log) 0.276 0.144 0.055 1.318 0.206 0.185 0.265 1.229 TCH 0.885 0.541 0.102 2.423 0.893 0.752 0.234 2.444 Transnational 0.000

Some 1.889 0.529 0.000 6.612 Significant 2.147 0.549 0.000 8.558

CPA 0.004 0.002 0.087 1.004 0.005 0.002 0.011 1.005 AUC 0.823 0.874 Classification 93.1 96.1 Nagelkerke R-square 0.26 0.251 Chi-square 58.298 39.489 2LL 214.294 138.217 df 15 13 N 524 542

Table 5. Pursuit by all

Model 7 Model 8

Variable B S.E. p Exp(B) B S.E. p Exp(B)

Constant −3.136 0.581 0.000 0.043 −3.694 0.743 0.000 0.025 Individual 0.958 0.365 0.009 2.607 1.402 0.468 0.003 4.063 Islam 0.474 0.254 0.062 1.607 0.619 0.325 0.057 1.857 Ethnic Fraq. 0.325 0.568

Intermediate 0.620 0.440 0.159 1.858 0.587 0.562 0.297 1.798 Low 0.554 0.393 0.158 1.741 0.334 0.508 0.511 1.396

Religious Fraq 0.682 0.964 Intermediate −0.129 0.410 0.753 0.879 0.142 0.530 0.788 1.153 Low 0.153 0.369 0.679 1.165 0.099 0.475 0.835 1.104

Elevation 0.000 0.000 0.266 1.000 0.000 0.000 0.680 1.000 Tropics −0.339 0.436 0.436 0.712 −0.616 0.606 0.310 0.540 Landlocked −0.765 0.729 0.294 0.466 0.172 0.804 0.831 1.188 Polity −0.012 0.024 0.612 0.988 −0.041 0.031 0.186 0.960 AUC 0.677 0.709 Classification 89 93.9 Nagelkerke R-square 0.082 0.09 Chi-square 34.866 26.182 2LL 544.675 355.915 df 10 10 N 836 836

604 W. S. MCCANN

Islam was marginally significant in both models; suggesting that the collapsing of ideological groups still renders Islamic actors as more dangerous than any other ideology. Nothing else was significant across these models, just as in models 1 and 2.

In models 9 and 10 (Table 6), individuals were excluded, and once again ideology, ethnic and religious fractionalization had significant relationships with the outcome measures. However, Islam was now significant in both models whereas ideology was only significant in model 4. In each model, Islamic extremist groups were 2.3 and 4.2 times the odds of pursuing CBRN weapons and BRN weapons, respectively. Similar to models 3 and 4, ethnic fractionalization was marginally significant when regressed on BRN weapons, but significant when the outcome included chemical weapons. Thus, countries with low levels of ethnic diversity have 3.7 times the odds of having groups pursue CBRN weapons, compared to countries with high levels of diversity. The opposite relationship was true for religious fractionalization, just as in model 3, whereby countries with low and intermediate levels of ethnic diversity were less likely to have groups pursue CBRN weapons than countries with high levels of fractionalization. This again disappears when chemical weapons are excluded.

Once the additional covariates are added to models 11 and 12 (Table 7), much of the same effects from models 5 and 6 are replicated. However, GDP(log) becomes significant here when regressed on CBRN weapons, and TCH finally achieves marginal significance in model 12. Furthermore, Islam is also significant when regressed on CBRN weapons whereas ideology was not significant in model 5. Thus is further evidence that this secondary ideology measure adds value to this study in under- standing Islamic extremists odds of pursuing these weapons in absolute terms. Across these models, Islamic extremist groups had 3.6 and 7.3 times of the odds of pursuing CBRN weapons and BRN weapons, respectively. Whereas ethnic fractionalization was significant in both models 5 and 6, it was now only significant in model 11 due to the marginal significance obtained in model 12. This suggests that chemical weapons may be driving the findings of significance for ethnic diversity. Nevertheless, when regressed on CBRN weapons pursuit, countries with low levels of ethnic diversity had 4.4 times the odds of having a group pursue these weapons. Again, religious fractionalization was not signifi- cant. However, elevation was significant again, but only in the model regressed on CBRN weapons.

In the final model, polity achieved significance for the first time. Thus, each unit increase on the polity scale is associated with a small reduction in odds of having a group pursue BRN weapons. Put simply, democratic countries are less likely to have terrorist groups pursue more lethal weapons of

Table 6. Pursuit by All Groups

Model 9 Model 10

Variable B S.E. p Exp(B) B S.E. p Exp(B)

Constant −2.930 0.685 0.000 0.053 −4.312 1.115 0.000 0.013 Islam 0.871 0.348 0.012 2.388 1.445 0.474 0.002 4.242 Ethnic Fraq. 0.023 0.077

Intermediate 1.008 0.542 0.063 2.741 1.856 0.844 0.028 6.395 Low 1.333 0.488 0.006 3.793 1.662 0.789 0.035 5.272

Religious Fraq 0.029 0.589 Intermediate −1.129 0.460 0.014 0.323 −0.645 0.683 0.346 0.525 Low −1.101 0.467 0.018 0.333 −0.672 0.705 0.341 0.511

Elevation 0.000 0.000 0.273 1.000 0.000 0.001 0.724 1.000 Tropics −0.568 0.522 0.277 0.567 −0.938 0.793 0.237 0.392 Landlocked −1.366 0.970 0.159 0.255 0.469 1.004 0.641 1.598 Polity 0.017 0.029 0.553 1.018 −0.005 0.038 0.886 0.995 AUC 0.707 0.76 Classification 92.2 96 Nagelkerke R-square 0.1 0.121 Chi-square 27.022 22.069 2LL 317.022 188.106 df 9 9 N 628 628

TERRORISM AND POLITICAL VIOLENCE 605

mass destruction. Conversely, wealthier countries are more likely to experience CBRN weapons pursuit. Again, this suggests that chemical weapons are driving much of the findings here, as most democracies are in fact wealthier in this study. Transnational was significant once again, in that groups that had any level of transnational activity had 6.8 and 8.5 times the odds of pursuing CBRN weapons; although this finding is likely even more true of BRN weapons pursuit (see above). CPA is only marginally significant when regressed on CBRN weapons pursuit, but retains significance (as it did in model 6) when the outcome is BRN weapons pursuit. Thus, groups that are more lethal are more likely to pursue more lethal weapons.

Discussion

This study used twelve different binary logistic regression models constrained on different subsamples of the newly merged dataset to examine whether Islamic extremists are more likely to pursue CBRN weapons. After execution of the forgoing models, none of the five hypotheses found strong support, with three finding decent support depending on the model conditions, and two hypotheses finding no support. The exclusion of chemical weapons cases via use of the second outcome measure results in important differences in the direction and magnitude of various effects. While all BRN cases with a positive outcome (“1”) are CBRN cases with the same outcome value; the reverse is not true. For obvious reasons, the discussion below distinguishes between these two constructs in a manner that reflects this reality along with the need to study terrorists’ pursuit of non-chemical weapons and agents.

With regard to H1, transnational groups were only significantly more likely to pursue CBRN weapons—not BRN weapons—and only when the analysis was able to include the measure (models 5 and 11). There was no such data for individuals, and it was too perfect a predictor of BRN weapons

Table 7. Pursuit by All Groups with additional measures

Model 11 Model 12

Variable B S.E. p Exp(B) B S.E. p Exp(B)

Constant −13.102 4.018 0.001 0.000 −9.648 5.070 0.057 0.000 Islam 1.301 0.457 0.004 3.672 1.990 0.652 0.002 7.314 Ethnic Fraq. 0.030 0.051

Intermediate 1.028 0.636 0.106 2.794 1.815 0.914 0.047 6.139 Low 1.492 0.570 0.009 4.447 2.033 0.834 0.015 7.639

Religious Fraq 0.787 0.781 Intermediate −0.446 0.739 0.546 0.640 −0.613 0.971 0.528 0.542 Low −0.219 0.736 0.766 0.804 −0.634 0.932 0.496 0.530

Elevation 0.001 0.000 0.022 1.001 0.000 0.001 0.953 1.000 Tropics −0.508 0.664 0.444 0.601 −2.420 1.321 0.067 0.089 Landlocked −0.855 0.970 0.378 0.425 0.408 1.168 0.727 1.503 Polity −0.050 0.035 0.151 0.951 −0.090 0.045 0.043 0.914 GDP(log) 0.282 0.144 0.050 1.326 0.184 0.180 0.309 1.201 TCH 0.573 0.437 0.190 1.774 1.158 0.658 0.078 3.184 Transnational 0.000

Some 1.917 0.527 0.000 6.802 Significant 2.146 0.549 0.000 8.548

CPA 0.004 0.002 0.073 1.004 0.005 0.002 0.013 1.005 AUC 0.82 0.873 Classification 92.9 96.1 Nagelkerke R-square 0.255 0.248 Chi-square 57.319 38.932 2LL 215.273 138.775 df 14 12 N 524 542

606 W. S. MCCANN

pursuit in general, and for Islamic groups specifically, on both outcomes. As such, this hypothesis obtains only partial support because it could not be fully tested on BRN weapons pursuit, even though every indication pointed to it being the case as well.

The most notable finding across the models was regarding H2. Here, Islamic actors are more likely to pursue CBRN and BRN weapons when the models exclude individuals. The results collectively signify that Islamic actors have higher odds of pursuing any of these weapons when groups are considered alone, although they were marginally significant when individuals were included. Once individuals are filtered out, it is Islamic groups that are the most problematic as they are still significantly more likely to pursue BRN weapons; which are exponentially more dangerous. Thus, this hypothesis found no support, given that the prior literature had not established any significant differences between Islamic and non-Islamic inspired actors—except for the study by Guarrieri and Meisel85 which largely focused on individuals—and the mixed findings about religious actors in general.86 However, the findings do support recent scholarship by McCann that found that individuals and religious actors are in fact more likely to pursue CBRN weapons.87 These findings are not absolute, however, as they do not denote that Islamic actors are more likely to pursue these weapons in comparison to any other specific ideological grouping; just that they are more likely than non- Islamic motivated actors. Nevertheless, this finding is important and should be considered by future scholarship in modeling CBRN terrorism.

Another major takeaway concerning individual actors being the most likely to pursue both CBRN and BRN weapons (H3) when all cases were considered (models 1, 2, 7, & 8) reflects prior findings.88

This likely indicates that a disproportionate amount of individual actors within the merged dataset were not Islamic inspired and were in Western countries, whereas a much higher concentration of groups were Islamic inspired and located outside of Western countries. It is thus likely that a smaller share of these cases (e.g. individuals) are driving these results, and since groups are more equipped to pursue more complex weapons (BRN weapons), it is plausible that BRN cases are more likely pursued by groups. This reflects the findings from many of the models, whereby only groups are considered and even in those models, Islamic groups were still significantly more likely to pursue BRN weapons as compared to non-Islamic groups.

Taken together, Islamic groups are much more likely to pursue more advanced weapons, whereas individuals are more likely to pursue CBRN weapons in general. However, the results still demonstrate that, both individuals and Islamic actors are significantly more likely to pursue both CBRN and BRN weapons; it just matters how the models are constrained. Thus, there is a fair amount of overlap amongst individuals and groups in their propensity for pursuit of various weapons when the findings are extrapolated across the models. This provides support for prior scholarship by Ackerman and Pinson that argues individuals are more dangerous than groups in terms of CBRN weapons pursuit.89

Recent scholarship by McCann and others supports this assertion as well.90 It could be that individuals are more capable of skirting the checks and balances imposed by group hierarchies and culture. More scholarship should be done to discern whether the pathways and rationalizations for CBRN weapons pursuit and use are different for individuals versus groups.

Ethnic and religious fractionalization also turned out to be significant predictors of CBRN and BRN weapons pursuit as well. Prior scholarship had only focused on their relation to group survival and actor lethality.91 Religious fractionalization was only significant, however, in models constrained on groups without additional predictors. There were drastically different findings across these models, nevertheless. In the latter, countries with low and intermediate levels of fractionalization experienced reduced odds of having groups pursue CBRN weapons. In the later models, Islamic actors were still significantly more likely to pursue CBRN weapons. It must be stated, however, that with the addition of actor-orientation and lethality measures, religious fractionalization fell out of significance. Thus, this finding should be taken with caution, as it seems both chemical weapons and the exclusion of individual actors largely explain these findings.

TERRORISM AND POLITICAL VIOLENCE 607

On the other hand, ethnic fractionalization was consistently significant across the models. When looking only at groups, countries with low levels of fractionalization were more likely to experience CBRN weapons pursuit compared to countries with high levels; and was marginally significant for BRN weapons pursuit. Once lethality and group-orientation measures were included, these became signifi- cant in both models 5 and 6 as well as model 11. This signifies that that the additional measures mediate the effects of ethnic fractionalization and that disaggregating models to study the effects is important.

Collectively, these findings provide some support for H4 in line with the recent study by McCann, despite intermediate levels of ethnic fractionalization not being significantly different than high levels in explaining weapons pursuit.92 Thus, religious diversity within a society can play a role in under- standing CBRN weapons pursuit, but in drastically different ways, whereas ethnic diversity, or lack thereof, is significantly associated with increased odds of CBRN and BRN weapons pursuit when studying groups specifically. It could be the case that in extremely homogenous societies, out-group othering is easier to justify and substantiate. For example, the countries where groups are more likely to pursue CBRN weapons are Muslim-majority countries; many times ranging from 90–99 percent Muslim. A lack of interaction with other social groups coupled with strong religious and ethnic identities can make out-group othering more prominent alongside the large pool of eligible members for the in-group.93 But it is not clear as to why a lack of diversity would increase the propensity for CBRN weapons pursuit and use. More scholarship is needed on this topic. Regardless, specifying the type of weapon and type of actor provide the most context across models.

Polity was only significant in one of the twelve models. Thus, H5 received very little support, and the findings of this study largely conflict with prior scholarly work, namely the Ivanova and Sandler study, but also McCann’s study in part as well.94 In one model, the more democratic a country was, the less likely they were to experience BRN weapons pursuit by groups. Again, chemical weapons seem to be driving this finding, as the most dangerous weapons are seemingly pursued in more authoritarian societies. However, wealthy countries were also found to be more likely to experience CBRN weapons pursuit by groups. These two findings conflict to some extent, as there is a correlation between a country’s wealth and its polity, in that wealthier countries tend to be more democratic.

In fact, CPA and Transnational turned out to be significant predictors when included in several of the models. While these measures are only populated for groups, they mediated many of the factors included in the prior group constrained models. Groups that had carried out any level of transnational activity were significantly more likely to pursue CBRN weapons. In fact, this measure could not be included in the BRN weapons models because it was too strong of a predictor; as almost every case of BRN weapons pursuit was by a group with some transnational activity. This finding supports Ivanova and Sandler’s work, although the current study adds the differentiation by type of weapons pursuit and focus on groups versus individuals.95 CPA was a significant predictor of BRN weapons pursuit by groups, but was marginally significant when regressed on CBRN weapons pursuit. Basically, groups that are more lethal are more likely to pursue the more dangerous types of weapons.

Lastly, geographical measures added some value to the models as well. While landlocked did not reach significance across any of the models, tropics reached marginal significance in the models focused on BRN weapons pursuit by groups. Furthermore, a countries’ elevation was significant related to CBRN weapons pursuit (but not BRN pursuit) in the two models constrained on only groups with expanded explanatory measures. Thus, the more elevation a country has, the higher the odds that groups will pursue CBRN weapons. This partially supports assertions from Gaibulloev and Sandler that mountainous terrains provide more safe havens for terrorists, even though actors operating in landlocked nations were not significantly different in their behavior.96 These findings are somewhat tenuous given that they were only significant in very specific contexts. More exploration of these factors is needed; especially in regards to group success.

608 W. S. MCCANN

Overall, many of the results were expected. Results from a similar study by McCann had demon- strated that religious actors, individuals, and actors residing in countries with less ethnic diversity were more likely to pursue CBRN and BRN weapons.97 Again, a countries wealth, polity, and environ- mental factors played almost no role in explaining CBRN weapons pursuit, except in very specific contexts. Similarly, religious fractionalization only explained CBRN weapons pursuit when the models were constrained on groups. Thus, this study adds further support to prior scholarship that focused on religious terrorism more broadly, while specifying how Islamic extremists are potentially even more problematic within the vein of religious terrorism and CBRN weapons pursuit.

Limitations, recommendations, and concluding remarks

The biggest limitation of this study was the lack of demographic factors within the models. However, demographic factors would only be assessed on an individual level and the models included both individuals and groups. For reasons outlined in the Methods and Methodological Appendix, no such factors were available. Nevertheless, this limitation precludes further inference about the correlates of individual actors’ pursuit of CBRN weapons. Another related limitation was the inability to control for prior history and longevity. Since this study used aggregated data, it would be impossible and illogical to include temporal measures of actor longevity or prior use of CBRN weapons. Similarly, other specific forms of ideology (e.g., anti-government, far-right, far-left, etc.) could not be used within the models due to low cell counts. This speaks to the obvious fact that religion and nationalistic ideologies (not mutually exclusive) dominate the pursuit of CBRN weapons. However, given the data, single- issue actors were not coded for, specifically. This is because of the distinct coding schemes the EDTG and POICN use that preclude the merging of far-right and far-left ideologies (EDTG) with single-issue actors (POICN). Thus, coding Islam was significantly easier because it was simply a further specifica- tion of “religious” ideology which both databases code for in an obvious and parsimonious manner. Keep in mind, however, that the GTD Ideological Motivations of Terrorism in the U.S. database was used to discern whether individuals included from the GTD were motivated by an Islamic ideology.98

This database only contains cases occurring in the U.S. Thus, the non-CBRN terrorism cases outside of the U.S. were left as missing for this measure since no data can be located to confirm their respective ideological affiliation. This results in a more conservative estimate of the effect of an Islamic ideology on pursuit when individuals are considered. This measure was still a very strong predictor across all of the models, but when considered alongside individuals, the results should be seen again as under- estimations of the magnitude of these effects. Put simply, it is very likely that an Islamic ideology is more problematic in terms of CBRN pursuit than is stated in this study.

Another notable limitation is that this study relies on open-source data. To date, the best available data on terrorism is open-source, despite it being riddled with numerous issues concerning bias, selection effects, noise accumulation, and threats to construct validity.99 Despite this limitation, the immense effort undertaken to merge these datasets while culling out invalid or unreliable cases in a systematic manner (see Methodological Appendix) reduces most threats to construct validity and external validity. For the foregoing reasons, researchers should continue to model CBRN cases against non-CBRN cases to better assess the impact of actor, country, and environmental factors on pursuit and use.

This study’s findings also inform future scholarship. Scholars should continue to examine the relationship between religion and CBRN weapons pursuit. It seems this relationship is dominated largely by Islamic actors and chemical weapons however, so more nuance into specific religious beliefs and specific pursuits is warranted. While obtainment or production of BRN weapons may prove more difficult, it may also prove more deadly; serving as a low probability-high risk event that the world can properly plan for if certain policy steps are taken. It must also be noted that the CBRN and BRN distinctions were not demarcations of single forms of pursuit; many actors simultaneously pursued chemical and biological weapons, and some pursued all of them. Future studies should examine whether pursuit should be modeled on a continuum in terms of agent(s) total severity rather than in a simple dichotomous fashion. This may also yield more nuanced findings for actors to be classified

TERRORISM AND POLITICAL VIOLENCE 609

along a dangerousness index. Such an index could also prove extensive insights into the magnitude and direction of the risk and protective factors explored here. It may be the case that Islamic groups are more likely to pursue BRN weapons than other actors, but this may because these groups are in a better position to pursue them as they have more capital, resources, and a more robust infrastructure, especially in authoritarian societies or failed states. Furthermore, scholars should delve deeper into the available evidence on the rationale to pursue CBRN weapons by all actors, but especially Islamic groups in light of this study’s findings. Such analyses may yield strong insights into the factors that various actors considered when choosing to pursue or not pursue CBRN weapons. This is a salient topic given that most actors, even Islamic actors, do not pursue CBRN weapons ever.

It is obvious from the findings that biological weapons are driving many of the findings in the non- CBRN models since chemical weapons make up the vast amount of CBRN weapons pursuit. However, biological weapons make up the vast majority within the BRN category given that nuclear and radiological weapons have rarely been pursued by any actors regardless of ideology. Within the data used in this study, about half of all biological (not including toxins) pursuits are by groups and by Islamic actors. This alone is a different proportion from CBRN weapons pursuit, in general, and further signifies the importance of model specification and nuance. But there are too many unknowns here, and recent scholarship has called attention to the need to study bioterrorism more vigorously.100

A step in the right direction vis-à-vis policy would be updating the Biological Weapons Convention (BWC) which went into effect in 1975 to prevent a biological arms race around the world. This does not preclude researchers and governments from developing biological agents and toxins for prophylactic or defensive purposes but does so for offensive purposes. This treaty needs to be strengthened via a legally binding instrument, however, as there is no real enforcement mechanism for biological weapons development.101 This proposal is not new, as many scholars have advocated for an enforcement mechanism since the Chemical Weapons Convention of 1993 does have a verification process to certify state conformity and several international protections are in place for nuclear weapons oversight (e.g. International Atomic Energy Agency).102 Thus, biological weapons defense may be the frontier moving forward given how COVID-19 has drastically impacted societies around the world, and such a threat could become more problematic in light of advances in synthetic biology.103 Ackerman and Peterson have even postulated that the recent pandemic may have altered the risk-reward nexus for many non- state actors given the wide-ranging impacts a biological agent can have; especially for groups of the millenarian vein.104 Hopefully a binding enforcement instrument can be executed during the Ninth BWC Review Conference, but if not, scholars and practitioners should continue to advocate for such a mechanism in light of this study’s findings.

While the BWC is focused on the control of state programs, an update to the BWC would ensure that compliance in weak, transitioning, or failed states can be monitored to ensure that states do not transfer their agents to nefarious recipients. A UN Security Council meeting in 2017 even highlighted the urgent need for more international collaboration to prevent non-state actors from acquiring WMDs in the wake of growing technological development and use by groups such as ISIS.105 A chief concern of members in this meeting was that some states were supplying non-state actors with WMDs. Aside from a verification instrument, suggestions included more robust regulation of dual-use technologies, stricter enforcement of export-import controls, and penalties for non-compliance. A verification regime will not stop all threats of biological terrorism, but it can greatly reduce the risk given that state actors are the most likely proprietors of biological and chemical agents sought by non-state actors. This analysis only supports the urgency of this call given the findings that Islamic extremist groups are substantially more likely to pursue CBRN weapons in general, and BRN weapons specifically as the latter is almost entirely biological weapons. Prior research has shown that religious groups in general are more likely to pursue these types of weapons as well, so the call for a stronger BWC is not a hollow one.106

Overall, this study fills in several gaps evident within the CBRN terrorism literature using novel techniques and data. The biggest takeaways are that: (1) Islamic groups are more likely to pursue all forms of CBRN weapons; (2) Islamic extremism is largely driving prior findings about “religion” and CBRN weapons pursuit; (3) the inclusion of chemical weapons in the models undervalues the role of

610 W. S. MCCANN

ideology, especially when it is Islamic; and (4) transnational groups are extremely problematic in terms of CBRN weapons pursue; especially Islamic groups. More work is needed in these areas to discern why these actors are actually more likely to pursue these weapons, however.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from several third-party sources. This includes the Global Terrorism Database (GTD) at START, the Profiles of Incidents Involving CBRN and Non-State Actors (POICN) at START, and the Extended Data on Terrorist Groups, 1970 to 2016, which is publicly available at https://doi.org/10.1177/ 0022002719857145. Restrictions apply to the availability of some of these data, which were used under license for this study. POICN and the GTD are available to the public with the permission of START. The Methodological Appendix outlines how these datasets can be merged, once they have been acquired by researchers.

Notes on contributor

Wesley S. McCann is an Assistant Professor in the Department of Criminology, Law and Society at George Mason University. He holds a PhD in criminal justice and criminology from Washington State University and has published numerous articles dealing with terrorism. His current research focuses on issues concerning hate crime and terrorism legislation, counterterrorism policy, the immigrant–crime relationship, anti-immigrant hate crime, and CBRN terror- ism. His work has appeared in Studies in Conflict & Terrorism, Terrorism and Political Violence, Criminal Justice Policy Review, American Journal of Criminal Justice, International Journal of Offender Therapy and Comparative Criminology, Criminal Justice Review, Corrections, Criminal Law Bulletin, Journal of Criminal Justice Education, and various law reviews.

Notes

1. See Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons,” Journal of Applied Security Research (2020). doi.org/10.1080/19361610.2020.1860631

2. Ibid. 3. Paper under review, author identity concealed to protect anonymity. 4. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 5. This hypothesis is structured in this way since very little scholarship has been done on Islamic extremists’ pursuit

of CBRN weapons, and thus, the literature is not clear on whether they are more likely to actually pursue these weapons even though recent scholarship indicates that religious actors in general are. Thus, this hypothesis adopts a conservative approach in testing this relationship.

6. Gary A. Ackerman and Lauren E. Pinson, “An Army of One: Assessing CBRN Pursuit and Use by Lone Wolves and Autonomous Cells,” Terrorism and Political Violence 26, no. 1 (2014): 226–45; Victor H Asal, Gary A. Ackerman, and R K. Rethemeyer, “Connections Can Be Toxic: Terrorist Organizational Factors and the Pursuit of CBRN Weapons,” Studies in Conflict & Terrorism 35, no. 3 (2012): 229–54; Kate Ivanova and Todd Sandler, “CBRN Attack Perpetrators: An Empirical Study,” Foreign Policy Analysis 3, no. 4 (2007): 273–94.

7. David C. Rapoport, “Terrorism and Weapons of the Apocalypse,” National Security Studies Quarterly 5 (1999): 49–67; David Tucker, “What Is New about the New Terrorism and How Dangerous Is It?” Terrorism and Political Violence 13 (2001): 1–14.

8. Bruce Hoffman, “Holy Terror”: The Implications of Terrorism Motivated by a Religious Imperative (Santa Monica, CA: RAND Corporation, 1993). https://www.rand.org/pubs/papers/P7834.html. Also available in print form.

9. J. M. Post, “The Psychology of WMD Terrorism,” International Studies Review 7, no. 1 (2005): 148–51 10. Lewis Dunn, “Next Generation Weapons of Mass Destruction and Weapons of Mass Effects Terrorism,” Defense

Threat Reduction Agency, Advanced Systems and Concepts Office. Report Number ASCO 2008 001. (Washington, D.C, 2008).

11. David C. Rapoport, “Terrorism and Weapons of the Apocalypse,” National Security Studies Quarterly 5 (1999): 49–67; David Tucker, “What Is New about the New Terrorism and How Dangerous Is It?” Terrorism and Political Violence 13 (2001): 1–14.

TERRORISM AND POLITICAL VIOLENCE 611

12. Dipak Gupta and Kusum Mundra, “Suicide Bombing As a Strategic Weapon: an Empirical Investigation of Hamas and Islamic Jihad,” Terrorism and Political Violence 17 no. 4 (2005): 573–98; Compare with Pape, Robert A. Dying to Win: The Strategic Logic of Suicide Terrorism (New York: Random House Trade Paperbacks, 2006).

13. David Tucker, “What Is New about the New Terrorism and How Dangerous Is It?” 14. Victor H. Asal, Gary A. Ackerman, and R K. Rethemeyer, “Connections Can Be Toxic: Terrorist Organizational

Factors and the Pursuit of CBRN Weapons.” 15. The authors imply that religious groups are less likely to take credit for their actions in order to avoid detection,

citing Mark Juergensmeyer, Terror in the Mind of God: The Global Rise of Religious Violence (Berkeley: University of California Press, 2003).

16. Kate Ivanova and Todd Sandler, “CBRN Attack Perpetrators: An Empirical Study.” 17. J. Stern, “Prospect of Domestic Bioterrorism,” Emerging Infectious Diseases 5, no. 4 (1999): 517–22. 18. Paper under review. Citation withheld to preserve identity of author(s). 19. Thomas R. Guarrieri and Collin J. Meisel, “Extremists and Unconventional Weapons: Examining the Pursuit of

Chemical and Biological Agents,” Behavioral Sciences of Terrorism and Political Aggression (2019): 1–20. 20. Tyler A. Clark and Thomas R. Guarrieri, “Modeling Terrorist Attack Cycles as a Stochastic Process: Analyzing

Chemical, Biological, Radiological, and Nuclear (CBRN) Incidents,” Journal of Applied Security Research (2020). doi: 10.1080/19361610.2020.1761743.

21. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 22. Ibid. 23. This was based on a review of group names, location, motivation notes, and anecdotal evidence. 24. Thomas R. Guarrieri and Collin J. Meisel, “Extremists and Unconventional Weapons: Examining the Pursuit of

Chemical and Biological Agents.” 25. Jeffrey Bale, “Jihadist Ideology and Strategy and the Possible Employment of WMD,” in Jihadists and Weapons of

Mass Destruction, ed. Gary Ackerman and Jeremy Tamsett (Boca Raton: Crc Press, 2009), 3–60. 26. John P. Caves and W. Seth Carus, “The Future of Weapons of Mass Destruction: Their Nature and Role,”

National Defense University Press Occassional Paper No. 10 (2014). 27. Nasir bin Hamd al-Fahd, “A Treatise on the Legal Status of Using Weapons of Mass Destruction against Infidels,”

Available at https://nsarchive2.gwu.edu/nukevault/ebb270/07.pdf 28. “Ayatollah Khamenei Says Nuclear Weapons Are ‘Forbidden under Islamic Law,’” The World from PRX, May 30,

2019, https://www.pri.org/stories/2019-05-30/ayatollah-khamenei-says-nuclear-weapons-are-forbidden-under- islamic-sharia-law.

29. Reuven Paz, “Global Jihad and WMD: Between Martyrdom and Mass Destruction,” www.hudson.org. (Accessed September 12, 2005). https://www.hudson.org/research/9798-global-jihad-and-wmd-between-martyrdom-and- mass-destruction.

30. Ibid. 31. Sammy Salama and Lydia Hansell, “Does Intent Equal Capability? Al-Qaeda and Weapons of Mass Destruction,”

The Nonproliferation Review 12, no. 3 (November 2005): 615–53, https://doi.org/10.1080/10736700600601236. 32. Jeffrey Bale, “Jihadist Ideology and Strategy and the Possible Employment of WMD.” 33. Ibid. 34. Jeffrey Bale, “Jihadist Ideology and Strategy and the Possible Employment of WMD”; James J. F. Forest and

Sammy Salama, “Jihadist Tactics and Targeting.” 35. Jeffrey Bale, “Jihadist Ideology and Strategy and the Possible Employment of WMD,” 23 36. Mark Dechesne, “Obedience to Divinity? The Psychology of Jihadist WMD Terrorism,” in Jihadists and Weapons

of Mass Destruction, ed. Gary Ackerman and Jeremy Tamsett (Boca Raton: Crc Press, 2009), 61–82. 37. Reuven Paz, “Global Jihad and WMD: Between Martyrdom and Mass Destruction — by Reuven Paz,” www.

hudson.org (Accessed September 12, 2005), https://www.hudson.org/research/9798-global-jihad-and-wmd- between-martyrdom-and-mass-destruction.

38. Jeffrey Bale, “Jihadist Ideology and Strategy and the Possible Employment of WMD.” 39. James J. F. Forest and Sammy Salama, “Jihadist Tactics and Targeting,” in Jihadists and Weapons of Mass

Destruction, ed. Gary Ackerman and Jeremy Tamsett (Boca Raton: CRC Press, 2009), 83–100. 40. Ibid. 41. Gary Ackerman and Jeremy Tamsett, Jihadists and Weapons of Mass Destruction (Boca Raton: CRC Press, 2009);

Rolf Mowatt-Larssen, “Al Qaeda Weapons of Mass Destruction Threat: Hype or Reality?” Belfer Center for Science and International Affairs, 2010, https://www.belfercenter.org/publication/al-qaeda-weapons-mass- destruction-threat-hype-or-reality; John Stone, “Al Qaeda, Deterrence, and Weapons of Mass Destruction,” Studies in Conflict & Terrorism 32, no. 9 (September 4, 2009): 763–75, https://doi.org/10.1080/ 10576100903109693; Sammy Salama and Lydia Hansell, “Does Intent Equal Capability? Al-Qaeda and Weapons of Mass Destruction.”

42. Khusrav Gaibulloev and Todd Sandler, “What We Have Learned about Terrorism since 9/11,” Journal of Economic Literature 57, no. 2 (June 2019): 275–328, https://doi.org/10.1257/jel.20181444.

612 W. S. MCCANN

43. For a tertiary review, see the Global Terrorism Database located at START. 44. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 45. See Khusrav Gaibulloev and Todd Sandler, “Determinants of the Demise of Terrorist Organizations,” Southern

Economic Journal 79, no. 4 (2013): 774–92, https://doi.org/10.4284/0038-4038-2012.269. 46. S. Brock Blomberg, Khusrav Gaibulloev, and Todd Sandler, “Terrorist Group Survival: Ideology, Tactics, and

Base of Operations,” Public Choice 149, no. 3–4 (2011): 441–63, https://doi.org/10.1007/s11127-011-9837-4. 47. Citation redacted to protect author identity. Paper under review. 48. Dongfang Hou, Khusrav Gaibulloev, and Todd Sandler, “Introducing Extended Data on Terrorist Groups

(EDTG), 1970 to 2016.” 49. Khusrav Gaibulloev and Todd Sandler, “Determinants of the Demise of Terrorist Organizations.” 50. Victor H. Asal, Gary A. Ackerman, and R. Karl Rethemeyer, “Connections Can Be Toxic: Terrorist

Organizational Factors and the Pursuit of CBRN Weapons.” 51. Kate Ivanova and Todd Sandler, “CBRN Attack Perpetrators: An Empirical Study.” 52. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 53. Dongfang Hou, Khusrav Gaibulloev, and Todd Sandler, “Introducing Extended Data on Terrorist Groups

(EDTG), 1970 to 2016”; Kate Ivanova and Todd Sandler, “CBRN Attack Perpetrators: An Empirical Study.” Wealth as measured by income per capita did not explain the use of CBRN weapons by nationalist/separatist or fundamentalist/cult actors.

54. Khusrav Gaibulloev and Todd Sandler, “Determinants of the Demise of Terrorist Organizations.” 55. Victor H. Asal, Gary A. Ackerman, and R. Karl Rethemeyer, “Connections Can Be Toxic: Terrorist

Organizational Factors and the Pursuit of CBRN Weapons.” 56. Kate Ivanova and Todd Sandler, “CBRN Attack Perpetrators: An Empirical Study.” Wealth as measured by

income per capita did not explain the use of CBRN weapons by nationalist/separatist or fundamentalist/cult actors.

57. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 58. Under review, citation redacted to preserve author identity. 59. Thomas R. Guarrieri and Collin J. Meisel, “Extremists and Unconventional Weapons: Examining the Pursuit of

Chemical and Biological Agents.” 60. See S. Lewallen and P. Courtright, “Epidemiology in practice: case-control studies,” Community eye health 11,

no. 28 (1998): 57–58. 61. Markus K. Binder and Gary A. Ackerman, “Pick Your POICN: Introducing the Profiles of Incidents Involving

CBRN and Non-State Actors (POICN) Databases.” 62. Wesley S. McCann, “Who Said We Were Terrorists? Issues with Terrorism Data and Inclusion Criteria,” Studies

in Conflict & Terrorism, September 3, 2020, 1–21, https://doi.org/10.1080/1057610x.2020.1816676. 63. Dongfang Hou, K. Gaibulloev, and Todd Sandler, “Introducing Extended Data on Terrorist Groups (EDTG),

1976 to 2016,” Journal of Conflict Resolution 64, no.1 (2020): 199–225. 64. This was determined via a variable created in 2017 by START researchers called “individual.” This denotes

whether the actor was in factor “unaffiliated.” See CODEBOOK: INCLUSION CRITERIA AND VARIABLES, October 2019, 6.

65. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 66. Gary A. Ackerman and Lauren E. Pinson, “An Army of One: Assessing CBRN Pursuit and Use by Lone Wolves

and Autonomous Cells.” 67. D. Koehler and P. Popella, “Mapping Far-Right Chemical, Biological, Radiological, and Nuclear (CBRN)

Terrorism Efforts in the West: Characteristics of Plots and Perpetrators for Future Threat Assessment,” Terrorism and Political Violence (2018). https://doi.org/10.1080/09546553.2018.1500365.

68. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 69. Paper under review. Citation withheld to preserve identity of author(s). 70. This is the Ideology5 variable within the CTD. See Appendix. 71. Wesley S. McCann, “The Siege: Religious-Inspired Terrorism and CBRN Weapons.” 72. Gary Ackerman and Anastasia Kouloganes, “Single-Issue Terrorism.” 73. Ibid. 74. Dongfang Hou, Khusrav Gaibulloev, and Todd Sandler, “Introducing Extended Data on Terrorist Groups

(EDTG), 1970 to 2016,” for a review of using such a construct. 75. Ibid., for a review of using such a construct. 76. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 77. Monty Marshall and Ted Gurr, “Polity5 Project, Political Regime Characteristics and Transitions, 1800–2018

Database,” Center for Systemic Peace (2020). 78. Ibid., 16–17. 79. See Alberto F. Alesina et al., “Fractionalization,” SSRN Electronic Journal 8, no. 2 (2002), https://doi.org/10.2139/

ssrn.319762.

TERRORISM AND POLITICAL VIOLENCE 613

80. See Brock Blomberg, Khusrav Gaibulloev, and Todd Sandler, “Terrorist Group Survival: Ideology, Tactics, and Base of Operations”; Khusrav Gaibulloev and Todd Sandler, “Determinants of the Demise of Terrorist Organizations”; and Dongfang Hou, Khusrav Gaibulloev, and Todd Sandler, “Introducing Extended Data on Terrorist Groups (EDTG), 1970 to 2016.”

81. John Gallup, Andrew Mellinger, and Jeffrey Sachs, “Geography Datasets,” 2010. 82. See discussion in Khusrav Gaibulloev and Todd Sandler, “Determinants of the Demise of Terrorist

Organizations.” 83. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 84. Thomas R. Guarrieri and Collin J. Meisel, “Extremists and Unconventional Weapons: Examining the Pursuit of

Chemical and Biological Agents.” 85. See Gary A. Ackerman and Lauren E. Pinson, “An Army of One: Assessing CBRN Pursuit and Use by Lone

Wolves and Autonomous Cells”; Victor H. Asal, Gary A. Ackerman, and R. Karl Rethemeyer, “Connections Can Be Toxic: Terrorist Organizational Factors and the Pursuit of CBRN Weapons”; Thomas R. Guarrieri and Collin J. Meisel, “Extremists and Unconventional Weapons: Examining the Pursuit of Chemical and Biological Agents”; Kate Ivanova, and Todd Sandler, “CBRN Attack Perpetrators: An Empirical Study”; D. Koehler and P. Popella, “Mapping Far-Right Chemical, Biological, Radiological, and Nuclear (CBRN) Terrorism Efforts in the West: Characteristics of Plots and Perpetrators for Future Threat Assessment”; Wesley S. McCann, “The Siege: Religious-Inspired Terrorism and CBRN Weapons.”

86. Wesley S. McCann, “The Siege: Religious-Inspired Terrorism and CBRN Weapons.” 87. Paper under review, citation redacted to protect author identity. 88. Gary A. Ackerman and Lauren E. Pinson, “An Army of One: Assessing CBRN Pursuit and Use by Lone Wolves

and Autonomous Cells.” 89. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 90. D. Koehler and P. Popella, “Mapping Far-Right Chemical, Biological, Radiological, and Nuclear (CBRN)

Terrorism Efforts in the West: Characteristics of Plots and Perpetrators for Future Threat Assessment.” 91. Brock Blomberg, Khusrav Gaibulloev, and Todd Sandler, “Terrorist Group Survival: Ideology, Tactics, and Base

of Operations”; Khusrav Gaibulloev and Todd Sandler, “Determinants of the Demise of Terrorist Organizations”; Dongfang Hou, K. Gaibulloev, and Todd Sandler, “Introducing Extended Data on Terrorist Groups (EDTG), 1976 to 2016.”

92. Wesley S. McCann, “The Siege: Religious-Inspired Terrorism and CBRN Weapons.” 93. J. M. Berger, Extremism (Cambridge, MA: MIT Press, 2018) 94. Kate Ivanova and Todd Sandler, “CBRN Attack Perpetrators: An Empirical Study”; “Wesley S. McCann, “The

Siege: Religious-Inspired Terrorism and CBRN Weapons”; Article Under Review, Full Citation withheld to preserve author identity.

95. Kate Ivanova and Todd Sandler, “CBRN Attack Perpetrators: An Empirical Study”; “The Siege: Religious- Inspired Terrorism and CBRN Weapons.”

96. See discussion in Khusrav Gaibulloev and Todd Sandler, “Determinants of the Demise of Terrorist Organizations.”

97. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.” 98. Erin Miller, “Global Terrorism Database Ideological Motivations of Terrorism in the United States,” 2017,

https://doi.org/10.7910/DVN/SACQNK, Harvard Dataverse, V2. 99. See Wesley S. McCann, “Who Said We Were Terrorists? Issues with Terrorism Data and Inclusion Criteria.”

100. Under review, citation redacted to preserve author identity. 101. Anshula Sharma et al., “Next Generation Agents (Synthetic Agents): Emerging Threats and Challenges in

Detection, Protection, and Decontamination,” Handbook on Biological Warfare Preparedness, 2020, 217–56, https://doi.org/10.1016/b978-0-12-812026-2.00012–8, 248

102. Deepak K. Bhalla and David B. Warheit, “Biological Agents with Potential for Misuse: A Historical Perspective and Defensive Measures”; Steven Block, “The Growing Threat of Biological Weapons,” American Scientist 89, no. 1 (2001): 28, https://doi.org/10.1511/2001.1.28; S. Pullinger, “Fighting Biological Warfare: The Convention against Biological Weapons Badly Needs a Verification Protocol,” BMJ: British Medical Journal 320, no. 7242 (2000): 1089–90.

103. Steven Block, “The Growing Threat of Biological Weapons,” American Scientist 89, no. 1 (2001): Anshula Sharma et al., “Next Generation Agents (Synthetic Agents): Emerging Threats and Challenges in Detection, Protection, and Decontamination.”

104. Gary Ackerman and Hayley Peterson, “Terrorism and COVID-19: Actual and Potential Impacts,” Perspectives on Terrorism 14, no. 3 (2020).

105. United Nations Security Council, “States Must Step Up Efforts to Check Spread of Deadly Weapons as Non-State Actors Exploit Rapid Technological Advances, Speakers Tell Security Council,” SC/12888, 7985th Meeting (AM), June 28, 2017.

106. Wesley S. McCann, “The Siege: Religious-Inspired Actors and CBRN Weapons.”

614 W. S. MCCANN