Correlates of terror: Trends in types of terrorist groups and fatalities inflicted

Abstract A large body of recent work seeking to explain the strategies and causes of terrorism exists, with hypotheses ranging from frustration-aggression theories to strategic choice to psychological dysfunction or a number of other factors. Remarkably few studies examine the actual sources of terrorism, however. Despite the existence of some large databases that categorize terrorist incidents, such as the MIPT (Memorial Institute for the Prevention of Terrorism) Dataset, few published works analyze the correlation between terrorism and types of movements (nationalist, religious, anarchist, Marxist, environmentalist, etc.) and the number or scale of terrorist attacks. This research provides a simple, clear, evidence-based and relative measure of which kinds of groups resort to terrorism. Although databases such as that of MIPT suffer from a large number of terrorist incidents carried out by unknown perpetrators, as well as classification problems (groups can typically be both religious and nationalist, for instance), the evidence examined here indicates that nationalist groups and Islamist movements, in that order, were the most common kinds of non-state actors that resorted to terrorism between 1998 and 2007. Islamist groups do, however, appear responsible for more fatalities than any other kind of terrorist during this period. We conclude by offering tentative applications of these correlations to some of the most prevalent theoretical explanations for terrorism. We do this in only a preliminary sense, in order to see how a better notion of the actual ‘correlates of terrorism” might affect thinking and policymaking on the issue.


PUBLIC INTEREST STATEMENT
This research provides a simple, clear, evidencebased and relative measure of which kinds of groups resort to terrorism. Although databases such as the one relied upon here (MIPT) suffer from a large number of terrorist incidents carried out by unknown perpetrators, as well as classification problems (groups can typically be both religious and nationalist, for instance), the evidence examined here indicates that nationalist groups and Islamist movements, in that order, were the most common kinds of non-state actors that resorted to terrorism between 1998 and 2007. Islamist groups do, however, appear responsible for more fatalities than any other kind of terrorist during this period. We conclude by offering tentative applications of these correlations to some of the most prevalent theoretical explanations for terrorism. We do this in only a preliminary sense, in order to see how a better notion of the actual "correlates of terrorism" might affect thinking and policymaking on the issue.
correlations to some of the most prevalent theoretical explanations for terrorism. We do this in only a preliminary sense, in order to see how a better notion of the actual 'correlates of terrorism" might affect thinking and policymaking on the issue.  " (1990) and other works adopt different approaches to understanding terrorism, but implicitly or explicitly link Islamist movements with a disproportionate tendency to resort to terrorism. (Khan, 2006;Perry & Negrin, 2008;Cooley, 2002;Lewis) In his 2009 book, Old and New Terrorism, Peter Neumann states that "Those who claim to act in the name of Islam have killed more people in the last two decades than any other branch of religiously inspired terrorism (Neumann, 2009)." Like many, Neumann provides no citation or statistical evidence for this assertion. Even well before the 11 September 2001 attacks, scholars and laypeople alike seemed to accept an a priori notion that religious terrorism had become the most common form of terrorism. Magnus Ranstorp (1996), for instance, began his "Terrorism in the Name of Religion" article with the following observation: "Far afield from the traditionally violent Middle East, where religion and terrorism share a long history, a surge of religious fanaticism has manifested itself in spectacular acts of terrorism across the globe. This wave of violence is unprecedented, not only in its scope and the selection of targets, but also in its lethality and indiscriminate character (Ranstorp, 1996)." Even scholarly works that attempt to refute a correlation between Islam and terrorism, such as Esposito (1999Esposito ( , 2002), Sadowski (2006), Khan (2003) and Goody (2004) strengthen the notion of a correlation by trying to refute it via an examination of the "true message of Islam," without citing statistical evidence about which kinds of groups account for the most terrorist attacks and casualties (Esposito, 1999(Esposito, , 2002Khan (2003Khan ( , 2006Khan ( , 2011); Goody, 2004) Their implied or explicit message is that although a disproportionate amount of terrorism is committed in the name of Islam, Islam as a religion is not responsible for this.
We are not primarily concerned here with a theological treatment of Islam's position on terrorism. Rather, we find remarkable that there appears to be a real lack of scientific research establishing to what extent such a correlation exists a priori, despite the plethora of research programs and books which purport to explain "what went wrong" in the Muslim world, or why some Middle Eastern countries have proven such fertile ground for terrorist movements, or how today's "new" terrorism differs from yesterday's "old" terrorism. Even when taking a critical stance towards the notion of "new" and religious terrorism, Charles Townshend (2002) utilizes only minimal statistical references to explain the perception of a link between Islamism and terrorism (Townshend, 2002). Those conducting statistical analyses may even skip this starting point, such as Robison, Crenshaw, and Craig Jenkins (2006), who implicitly accepted that Islamist terrorism is the defining characteristic of recent times, and sought to ascertain a structural cause for this by comparing Islamist terrorist attacks with those of Leftist groups during the Cold War (Financial Times, 2005;Robison et al., 2006).
In some cases, the lack of statistical analysis has been used to strengthen what are essentially opinions regarding contemporary terrorism. Bruce Hoffman (2006) describes a handful of infamously deadly terrorist attacks, followed by questionable statistics that pretend such attacks are representative of terrorism in general (Hoffman, 2006). He claims, for example that "….while al Qaeda perpetrated only 0.1 percent of all terrorist attacks between 1998 and 2004, it was responsible for nearly 19 percent of total fatalities from terrorist attacks during that time period (Hoffman, 2006, p. 88)." He does not mention that these numbers are horrendously skewed by the inclusion of 9/11. While 9/11 is indeed an example of especially lethal terrorism committed by an Islamist group, it is also an immense outlier that caused approximately 15% of the fatalities in Hoffman's data. 1 A single incident is not a trend. That al Qaeda is representative of most Islamist terrorist organizations should not be taken for granted. While this does not preclude a conclusion that Islamist terrorism is responsible for a disproportionate number of attacks or casualties, it does not demonstrate it either. The assumption must be investigated further, with a larger time-frame if possible.
Effective science, including social science, needs to test starting assumptions, including those that appear intuitive or obvious. If the Muslim world turns out not to produce a disproportionate number of groups that resort to terrorism, then attention to and explanations for terrorism should be constructed in a way that applies to a broad variety of movements, rather than singling out Islamist terrorists. If, on the other hand, Islamist movements do represent a disproportionate share of the various kinds of groups that engage in terrorism, then establishing such a correlation allows us to proceed to subsequent choices of what kinds of terrorist groups to focus on, as well as explanations for terrorism and counter-terrorist policy recommendations, with a great deal more confidence.

Quantitative Problems with Illicit Phenomena
This article simply counts which kinds of groups resort to terrorism today, thereby assessing whether or not some kinds of groups do so much more than others. Although databases such as that of the Memorial Institute for the Prevention of Terrorism (MIPT) suffer from a large number of terrorist incidents carried out by unknown perpetrators, as well as classification problems (groups can typically be both religious and nationalist, for instance), our research nevertheless uncovers a number of significant and interesting correlations. These correlations are then compared to some of the most prevalent theoretical explanations for terrorism in a preliminary manner, to show why the associations are important.
Understanding the root causes of terrorism requires both a firm qualitative and quantitative approach. The illicit nature of terrorism makes quantitative approaches inherently difficult and flawed to various degrees. Nonetheless, databases exist which compile detailed empirical data on terrorist incidents. When we undertook this study, the MIPT Terrorism Knowledge Database (TKB) was the most comprehensive research tool available to the public for classifying terrorist events. The TKB allowed users to search terrorist incidents within a specific timeframe for known perpetrators, ideological affiliation, and number of casualties, among other statistics. Unfortunately, in March 2008 the MIPT took down the more comprehensive and assessable database and replaced it with a more limited one which, among other shortcomings, does not allow researchers to filter the data by ideological affiliation of terrorism perpetrators. Luckily, we had already accessed and filtered the data we needed for our study before the MIPT replaced its more comprehensive data set. Brian Houghton (2010) explained the downfall of the TKB: "The TKB's demise was simply brought about by the economic "free rider" principleeveryone loved using it, but nobody wanted to pay. Typically in situations like this, the government steps in and creates a method to fund the public good, but in the case of the TKB, the Department of Homeland Security office who provided previous funding did not see how a counter-terrorism knowledge base impacted their narrow focus, and those who utilized the knowledge base the most did not put enough pressure on DHS to continue the funding. After all the funding and effort to create such a useful tool, which truly was a living legacy for those who perished in past acts of terrorism, the TKB died from bureaucratic neglect (Houghton, 2010)." The loss of the more comprehensive TKB was especially disheartening as it was one of the few tools that allowed private researchers to study their chosen timeframes and groups. Reports released to the public (such as the National Counterterrorism Center: Reports on Terrorist Incidents) usually cover only one year, and provide selected data and trend patterns of the organization's choosing (Mora, 2012). 2 Even the more comprehensive START (National Consortium for the Study of Terrorism and Responses to Terrorism) Global Terrorism Database does not classify perpetrators of terrorist attacks according to ideology. Although a database such as START could be useful to cross-check the TKB database we use here (for incidents, casualties and perpetrators, for instance), START's failure to classify or label the groups engaged in terrorism means it cannot be used for the purposes described here.
Despite the dearth of real statistics regarding what kinds of terrorist groups are responsible for what proportion of worldwide attacks, several interesting empirical findings already existed when we began this study on the correlates of terrorism. Some studies have asked similar questions on the ideologies of terrorist groups, but all have a narrow focus, such as Miller, Smarick, and Simone (2011), who analyzed the ideologies of only those groups which attacked the U.S. homeland (Miller et al., 2011). They found that only 6 percent of the relevant groups held a religious-focused ideology, and within that category several types of movements were represented (Miller et al., 2011, p. 15).
Other studies have focused on the same question, of whether "religious" terrorists tend to inflict higher casualties than terrorist groups with other ideologies. A study by Piazza (Piazza, 2008) examined, with a critical eye, the specific question of whether Islamist terrorist groups inflict greater casualties than non-Islamists. Piazza found support for his hypothesis that Islamist terrorist groups are only disproportionately lethal when affiliated with the al Qaeda network. Dipak Gupta referenced the MIPT database for a related claim, and showed that for the decade of 1998-2007, religious terrorists caused 52.8 percent of total terrorism casualties (Gupta, 2008). The trend of perceiving modern terrorist attacks as causing more casualties is not new; Enders and Sandler (2000) studied fatalities during the Cold War and post-Cold War era to confirm that, without separating groups by ideology, terrorism became more lethal after the collapse of the Soviet Union (Enders & Sandler, 2000). On a related topic and in another long term study, Crenshaw () used the TKB database in the debate on "new" vs. "old" terrorism. "New" terrorism often refers to religiously motivated terrorism that use violence as the "aim rather than the means" (Crenshaw, ). In other words, "new" terrorists are not easily bargained with as they do not have rational political goals in the way a traditional nationalist terrorist would. This is a notion popularized by Hoffman (2006), "For others like al Qaeda, however, the religious motive is overriding and indeed, the religious imperative for terrorism is the most important defining characteristic of terrorist activity today (Hoffman, 2006, p. 82)." However, Crenshaw () shows that there is less of a difference between "new" and "old" terrorism than many researchers believe. For instance, she states that according to the TKB less than half of the most lethal groups are solely religious in nature (Crenshaw,p. 23). Many of the remaining religious groups are classified as hybrids, with a religious-nationalist combination being the most common of the hybrids.
The notion of 'new' terrorism is still an important one, however. It is a term often invoked, whether to critique, support, or redefine. It would be impossible to provide a specific, agreed upon definition of 'new' terrorism. One effective summary comes from Tucker (2001): "This terrorism is reputedly distinguished from the old by a new structure, a new kind of personnel, and a new attitude toward violence. The new structure is a network, facilitated by information technology, the new personnel are amateurs, who often come together in ad hoc or transitory groupings, and the new attitude an increased willingness to cause mass casualties, perhaps by using chemical, biological, nuclear or radiological (CBNR) weapons (Tucker, 2001)." Thus, to speak of "new" terrorism one might only refer to one aspect of groups resorting to terrorism, such as a more diffuse structure, or an apparently greater lethality. What is of interest to us is not the structure, but ideology. Scholars such as Hoffman (1993) focus on the role of religion in "new" terrorism, with statements such as "They [religious terrorists] execute their terrorist attacks for no audience but themselves," and "Thus religious terrorist violence becomes almost an end in itself-a morally justified, divinely instigated expedient toward the attainment of the terrorists' ultimate ends (Hoffman,p. 20)." This would imply a very different kind of terrorism than is normally seen. "New" terrorists would not have attainable goals; they would simply wish to kill in the name of their religion. This view has been criticized for a variety of reasons. From one perspective, Townshend (2002) muses as to whether such activity should be considered terrorism at all, if it exists as described by proponents of "new" terrorism (Hoffman,vol. 3,p. 12). In another, Duyvesteyn (2004) discusses two archetypical "new" terrorist attacks, Aum Shinryiko's sarin attack and Timothy McVeigh's Oklahoma City bombing: "Their willingness to take the lives of innocent bystanders was larger than their willingness to sacrifice their own lives in order to please their spiritual leader (Duyvesteyn, 2004) ." Some go farther and, rather than challenging the underlying assumptions of "new" terrorism, directly attack the use of "new" and "Islamic" terrorism. Richard Jackson (2007) states "Crucially, the above narratives imply that because 'Islamic terrorism' is fanatical, religiously motivated, murderous and irrational, there is no possibility of negotiation, compromise or appeasement; instead, eradication, deterrence and forceful counter-terrorism are the only reasonable responses (Jackson, 2007)." Before going too far into this debate, however, one should at least have a better sense of what proportion of terrorist attacks various groups-classified by their type of ideology-are responsible for.
Our study uses the same TKB data that Crenshaw and others relied upon, with a slightly longer timeframe, to test some of these assertions. We also examine the number and effectiveness of terrorist groups from each religion, instead of placing them together in one broad "religious" category. In this way we attempt to provide a clearer picture than analyses that fail to differentiate between kinds of religious groups, exemplified by the statement of Ellis (2003) that "The lethality of religious terrorism is also reflected by the fact that in 1995 religious terrorist groups committed 25 percent of the recorded international terrorist incidents yet resulted in 58 percent of terrorist related fatalities that year (Ellis, 2003)." If most religious terrorists are Muslim, as many observers implicitly or explicitly seem to believe, then the observation above would imply that Islamist terrorists inflict a disproportionate number of casualties compared to secular terrorist groups. As Ellis presents the data, however, we cannot know if this is the case.
The distinction between religious terrorism and Islamist terrorism is important as the latter constitutes a very specific subcategory of the former. Furthermore, the idea of rising religious terrorism in general needs careful scrutiny, given observations such as the following: "According to the RAND-St Andrews statistics for 1996 no less than 37 per cent of all international terrorist acts that could be definitely linked to an organization were committed by ethnic separatist groups" (Wilkinson, 2001). Similarly, the National Counter Terrorism Center (NCTC) stated that "only 19 per cent of the terrorism incidents in 2004 were attributable to Islamic extremists…25 percent were secular/political attacks" (Financial Times, 2005). Unfortunately, neither the NCTC nor the Financial Times made the data or terrorist group classifications upon which they based their estimates accessible, like the TKB data that was available until March 2008. Other recent and well received studies, such as Abadie (2004), Krueger (2007;, Burgoon (2006), Piazza (2008), Dreher and Gassebner (2008), Krieger and Meierrieks (2010), Basuchoudhary and Shughart (2010), Bird, Blomberg, and Hess (2008), Andreas Freytag, Kruger, and Schneider (2010), and Campos and Gassebner (2009) present statistical analyses of terrorism, but do not provide a breakdown of which kinds of groups account for how much terrorism (Abadie, 2004;Andreas Freytag et al., 2010;Basuchoudhary & Shughart, 2010;Bird et al., 2008;Burgoon, 2006;Campos & Gassebner, 2009;Dreher & Gassebner, 2008;Krieger & Meierrieks, 2010;Krueger, 2007Krueger, , 2008Piazza, 2008). This article thus hopes to contribute to the debate by evaluating terrorism trends and perpetrators over a timeframe of nearly ten years. The focus is on specific types of religious terrorism and its relative prominence and deadliness compared to secular groups.
Before proceeding on to a discussion of our methodology and findings, it seems necessary to add still more caveats about problems that seem to bedevil virtually all statistical analyses of terrorism. First, most datasets (including the one we relied upon here) use Western press accounts as their primary data source. Besides the aforementioned coding problems that occur, this means that while almost 100% of deaths in Iraq or Afghanistan are counted, very few of those in countries such as the Democratic Republic of the Congo get counted, since international media tend not to pay as much attention to states like the DRC. Secondly, such data also tends to prioritize direct kinetic attacks (bombings, shootings) over social and economic warfare (blockades, deaths due to forced displacement, etc.) and "structural violence" (Galtung & Hoivik, 1971), even when the latter are deliberate policies intended to cause civilian harm. Thirdly, "state terrorism" probably accounts for far more civilian deaths over time than any actions of non-state actors employing violence. Finally, if we had datasets which counted both domestic and international terrorism that stretched back to the African liberation and civil wars of the 1960s-80s, the trends would probably look dramatically different (with more relative weighting of secular nationalist and Leftist terrorist groups). The same could be said if the data set included the period from 2008 to 2018, during which the rise of the so-called "Islamic State" (ISIS) and its grisly civilian death toll in Syria and Iraq would have skewed results heavily towards Islamist religious groups. Nonetheless, quantitative data analyses of terrorism appear here to stay despite these serious shortcomings, representing an imperfect attempt to shed more light on a very difficult-to-quantify phenomenon.

Methodology
For the purpose of this study, terrorism was defined as the intentional violent targeting of civilians by nonstate political actors. The specific focus on civilian casualties was intended to exclude guerilla opposition movements that only target agents of the state, and therefore, present a different kind of strategic threat than terrorists. Using the MIPT Terrorism Knowledge Base (TKB), which relies on an equivalent definition of terrorism, information was extracted on terrorist incidents from 1 January 1998 to 6 August 2007 (the data before 1998 excluded "domestic" attacks within the United States and we wished to avoid combining incommensurable data sets or excluding terrorist attacks that did not cross international borders). While this time period does include the attacks of 11 September 2001 and the first few years following the overthrow of Saddam Hussein in Iraq 2003, it omits the rise of the so-called "Islamic State" and the civil war in Syria-when attacks on civilians multiplied so much that they would have seriously skewed the results towards religious groups engaged in terrorism.
Incidents were screened for known perpetrators, perpetrator classification, region, victim type, and number of fatalities. If a specific terrorist group was linked to an incident, the database classified the group based on its ideology. Several groups were classified under multiple ideologies (i.e. Islamic/ Nationalist, or Nationalist/Anti-Globalization/Anarchist/Racist). 3 These classifications varied in both specificity and accuracy according to available information on the groups' actions and stated motives. Groups without sufficient information were left unclassified. Similarly, attributing attacks to specific groups was inherently difficult as sometimes multiple groups claimed responsibility for an attack, and sometimes no groups claimed responsibility.
At the time of study, the TKB listed 560 groups that were currently active, or had been active during the specified timeframe. The TKB divided these groups into 50 different classifications or combination of classifications (i.e. unknown, anti-globalization, communist/anti-globalization/leftist, etc.). We made comparisons by calculating the number of groups in each category. The level of specificity was kept consistent with the TKB (i.e. all sects of Islam and Christianity were classified as simply Islamic or Christian, and all forms of nationalism and communism were classified as simply Nationalist or Communist).
In order to determine the effectiveness of the different groups, we collected data on the amount of fatalities each group had reportedly caused during the given time period. We used fatalities instead of casualties (which include non-lethal injuries) in the interest of fair comparisons. We thereby avoided equating groups that inflicted ten fatalities with groups that inflicted ten casualties, as relatively minor injuries are sometimes placed in the casualty category. We only included groups with five or more fatalities in this section of the study. Given the time frame, this number seemed a reasonable cut-off point to separate significant groups from insignificant ones. Once the data was collected, it was compiled back into the group classifications for a comparison of lethality.

Data
Data from the TKB (see Graph 1) showed that between 1 January 1998 and 6 August 2007, more solely nationalist terrorist groups, 18.3 percent, existed than any other kind of group. The next most common secular group was communist, which accounted for 9 percent of all groups. Solely Islamic groups accounted for 11.4 percent of all groups, while groups classified as both Islamic and nationalist accounted for 13.9 percent of all groups (Hamas, for instance, was classified as both "nationalist" and "Islamist" while the Kurdistan Workers' Party was classified as nationalist and communist, despite being overwhelmingly made up of Muslims). Around 27 percent of the groups thus had "Islamic" as at least one of their classifications, while nearly 40 percent of the groups had Nationalist as at least one of their classifications. All remaining non-Muslim religious groups combined accounted for slightly more than 1 percent of the total (again, this number only represents active terrorist groups during the specified time frame). Seventy of the groups were unclassified and thirty-seven of the groups were listed as "Other".
We compared this data to statistics on the percentage of the world population that adheres to a specific religion. Christians account for 33% of the world's population, Muslims account for 21%, 16% are "nonreligious,"14% are Hindu and .22% are Jewish (Adherents.com, 2009). With all other variables held equal, one would normally expect that the number of terrorist groups coming out of each religious category would be roughly proportional to these statistics. This, however, is not the case. Keeping in mind the "five death threshold" for inclusion in our study, Muslim groups account for 96% of our religiously motivated terrorist organizations, a much higher percentage than would be expected. Jewish groups, responsible for 1 percent, represent a larger proportion of deadly terrorist organizations than their percentage of religious adherents (although this is a far smaller discrepancy and of questionable statistic value given the small numbers involved). Christians, in our time period, account for only 3 percent of the groups and Hindus, atheists and agnostics for none-significantly lower percentages than should be expected (See Graph 3). The percentage of secular nationalist groups is roughly proportional to the percentage of non-adherents (although the secular nationalists did not attack explicitly in the name of non-adherence to a religion, this is still notable).

Graph 1.
The data suggest that during this time period, religious terrorism in general was not significant in a general sense, but rather, Islamist terrorism accounted for the lion's share of religious terrorism and nationalist terrorism remained strong as well. Within this contemporary time frame nationalist ideology was also more prevalent amongst terrorist groups than any religious ideology. If nationalists account for most active terrorist groups, then why do Islamist groups seem to draw greater attention?
The relative effectiveness of different terrorist groups, in terms of fatalities inflicted, probably provides part of the answer.

Gross Number of Fatalities per Religiously Identified Terrorist Group
Islamic Christian Jewish 96% 3% 1%

Religiously identified terrorist groups
Islamic Christian Jewish Graph 2. Romano et al., Cogent Social Sciences (2019) fatalities were attributed to solely Islamist terrorist groups between 1998 and 2007. The TKB attributed around 5,035 fatalities to groups that were both Islamist and nationalist. At the same time, only 1,375 fatalities were attributed to solely nationalist groups, 1,815 fatalities were attributed to solely communist groups, and 530 were attributed to solely Christian groups. Nationalist groups with no Islamic element were attributed 1,940 fatalities.
This data clarifies several important points. Solely Islamist groups greatly outweighed all secular and non-Islamic religious groups in terms of fatalities. In general, groups with an Islamist element appeared highly effective in producing fatalities. Islamist groups, and nationalist groups with an Islamist element, produced far more fatalities than solely nationalist groups or nationalist groups without an Islamist element (i.e. nationalist groups plus nationalist/religious groups of all non-Muslim religions, combined). This finding appears to lend weight to the aforementioned proponents of the concept of "new terrorism," wherein religious extremists may seek to maximize the body count of unbelievers. Concurrently or alternately, suicide bombings first popularized in the 1980s and 90s by the Tamil Tigers (a secular nationalist group) and Palestinian Hamas (and Islamist nationalist group), and then adopted as a tactic of choice by radical Islamist groups, account for the higher lethality of Islamist groups. 4 So whereas the Irish Republican Army or similar secular groups engaged in terrorism avoided suicide bombings aimed at civilians (with the exception of the Tamil Tigers), this became a standard practice of Islamist terrorist groups such as al Qaeda.
It is also very interesting to note the fatality numbers attributed to the few Christian-identified terrorist groups. While Islamist terrorist groups receive attention for high overall numbers, Christian groups, as it happens, are even more deadly when they set out to kill. Four Christian groups, in our time period, were responsible for 540 deaths, which works out to an average of 135 per group. Compare this to each Islamic group, which on average is responsible for 80 deaths. A similar emphasis on self-sacrifice and the afterlife is found in both religions. It seems that when Christian groups decide to use their religion as a vehicle for redressing grievances, they use themes in common with Islam to similar effect.

Correlation versus Causation
Clearly, Muslim groups are by far the most numerous of the terrorist groups with a religious aspect to them. Such a correlation does not mean that Islam somehow "causes terrorism" of course. Islam today may just serve as more of an idiom for political dissent and opposition than other

Religiously Justified Terrorist Attacks
Percentage of wold adherents Terrorist attacks as a percentage of religious attacks Graph 3.
religions do in their respective societies. If so, we should look for explanations of terrorism that examine the reasons for political action and dissent outside legal, institutionalized channels. In the Muslim world, such dissent and political action will often take a religious, Islamist form, and some Islamist dissidents will adopt terrorist rather than non-violent or guerrilla tactics. In other non-Muslim societies dissidence outside the legal political system will take a nationalist, communist or other form, but rarely a religious one. If an equivalent TKB database had been available for the 1960s and 1970s, it seems probable that Islamist terrorist groups would not have figured very prominently in any correlations emerging from the data. Indeed, data 5 on the regional frequency of terrorist incidents worldwide indicates that until the late 1980s, there were a good deal more terrorist incidents in Western Europe and Latin America than in the Muslim world (Dugan, Lafree, Cragin, & Kasupski, 2008). One need only think of groups such as the Red Brigades, the Weathermen, and the Baader-Meinhof Gang to recall what form many cases of political dissidence in the West used to take. Since 2007, this trend towards terrorism occurring in or emanating from strife-torn Muslim countries seems to have increased further: A noteworthy finding of the Global Terrorism Index report is that 80 percent of all terrorist attacks in 2013 occurred in only five countries: Iraq, Afghanistan, Pakistan, Nigeria, and Syria (START, Global Terrorism Index, 2014). Until the 1980s, however, in most of the Muslim world (and particularly in the Arab world), nationalism and communism served as the principle ideologies of dissent, with fewer political Islamist movements vying for political power. With the increasing decline in the appeal of pan-Arab ideologies after 1967, the Iranian revolution of 1979, and the collapse of the communist bloc in 1989, political Islam took its place as the principle currency of dissent in much of the Muslim world.
Theories about the causes of terrorism in general, such as frustration-aggression theories 6 or theories of rational calculus in conditions of asymmetric conflict (Byman, 1998;Pape, 2003), 367 may simply need to make sure that they add an intervening variable involving the forms of political dissent and ideologies common to each society. For instance, a causal chain of explanation might look like this: If being shut out from any means of legal political participation breeds frustration, and frustration constitutes a major cause of aggression, and weak aggressors often turn to terrorism as one of their few available weapons, then the disproportionately authoritarian regimes of the Muslim world may produce a more frustrated populace that since the 1990s has tended to turn to Islamist idioms of dissent. This could help account for the 27% of terrorist groups that are either primarily Islamist or partially Islamist from 1998 to 2007. Scholars such as Abadie (2004), 8 Burgoon (2006) and Kurrild-Klitgaard, Justesen, and Klemmensen (2006) have, in fact, found a correlation between levels of political and economic freedom and terrorism (Abadie, 2004;Burgoon, 2006;Kurrild-Klitgaard et al., 2006).
Of course, authoritarianism in combination with the specific trends and circumstances prevalent in particular societies would constitute only one of several contextually-dependent explanations for the emergence of terrorist groups. This study found that by far the most common kinds of groups resorting to terrorism have a nationalist component to their ideology, rather than an Islamist one. The PKK (Kurdistan Workers' Party) in Turkey offers an example of a secular nationalist group that emerged in a Muslim semi-democratic state, for example. Likewise, the Tamil Tigers in Sri Lanka were a secular nationalist group fighting the democratically elected state on a mostly Buddhist Sinhalese island. Both the PKK and Tamil Tigers employed terrorist attacks and specifically suicide bombings (the Tamil Tigers are in fact credited with perfecting the suicide bomber vest). 9 In these cases, frustration aggression theories would focus on relative deprivation and state discrimination affecting the Kurdish and Tamil minorities (even though the states in question are democratically elected), which breeds frustration and hence aggression. The plurality of terrorist groups with a nationalist component to their ideology should not, in fact, overly surprise us. In today's world of nation-states, the number of nations far exceeds the supply of states. Many national groups are thus in conflict with the regimes that rule over them, seeking to secede and form their own states, replace "foreign" ruling elites with their own national kin, or trying to force their central governments to grant them more political power or minority rights. Since such nationalist groups generally face much more powerful central governments, and terrorism is a weapon of the weak, the large number of terrorist groups with a nationalist element to them makes perfect sense. Analyses such as Robert Pape's (2003) thus typically view terrorism as the rational calculus of the weak in asymmetric "national liberation" struggles against democratic states, insisting that the resort to terrorism or the specific tactic of suicide bombings is not religiously inspired (Pape, 2003).
Of course, some ideologies may be more likely to sanction terrorist tactics. The correlations in the data examined here show that various nationalisms and some interpretations of Islam (in that order) account for the largest number of terrorist groups between 1998 and 2007, with far-Leftist groups coming in a more distant third. Explaining the correlation of nationalist and Islamist groups to terrorist tactics require analysts to go a step further than simply deducing frustration-aggression or a rational strategy in a context of asymmetric conflict. In this case, the ideologies themselves may matter, particularly the way that many Right-wing nationalisms and some interpretations of Islam build an "in-group" and "out-group" dialectic world in which the members of the in-group can resort to any means necessary to advance their "noble" cause, while outsiders are typically demonized and dehumanized. In the case of militant interpretations of Islam, Egyptian ideologue Sayyid Qutb's black and white worldview offers a clear example of this kind of ideology: "It is not the function of Islam to compromise with the concepts of Jahiliyyah [unbelief] which are current in the world or to coexist in the same land together with a jahili system. Islam's stand is very clear. It says that the truth is one and cannot be divided; if it is not the truth, then it must be falsehood. The mixing and co-existence of the truth and falsehood is impossible. Command belongs to God, or otherwise to Jahiliyyah; God's Shari'ah will prevail, or else people's desires (Qutb, undated)." Explanations of terrorism relying on socio-psychological theories of individual and group behavior would thus focus on the "we-they" dialectic that nationalism and religion often encourage (Burton, 1979;Horowitz, 1985;Tajfel, 1970). 10 Once an outside group becomes sufficiently vilified even heretofore repugnant tactics to deal with them gain acceptance.
Such general approaches would still face the difficulty of explaining the correlation linking the number of Islamist groups, but not so much Christian, Hindu or other religious groups, and terrorist tactics. On one level, such a difference could be explained by going back to the aforementioned observation that Islamism now serves as the principle vehicle of political dissent (violent or nonviolent) in the Muslim world. Alternately, sociological, psychological and cultural explanations focusing on group identities, status, and the self-validation that individuals derive from a satisfactory group identity and relative standing could be brought to bear. In this case, factors such as the Islamic world's relative decline vis-à-vis the West since the eighteenth century, a series of lost wars against Israel, oppression of the Palestinians and other Muslims by non-Muslims, the wars in Afghanistan and Iraq, colonialism and imperialism all form typical explanations for the resort to terrorism as a kind of "backlash" during the time period analyzed here. 11 The same sort of logic would function for terrorism conducted by secular nationalist movements of groups such the Tamils, Kurds, Palestinians, Sikhs, Punjabis, Basque or Irish.
Additionally, some observers may look to other cultural approaches for an explanation of the contemporary correlation between Islamist groups and terrorism, given the relatively rare phenomenon of non-Muslim religious terrorism. This kind of discourse would look to the theology, history and culture of the Muslim world for a causal explanation. The danger in this involves treating Islam as one unchanging homogenous belief system and culture, ignoring the dizzying amount of complexity and difference in Islam and the various societies in which it is practiced. The fact that Christian, Hindu and Jewish terrorist groups exist likewise implies that any theological system remains open to all manner of interpretations. Scholars such as Sami Zubaida (1993) therefore exhort us to avoid cultural essentialism, all the while remaining sensitive to the particular complex histories and cultures of different societies (Zubaida, 1993).
With Zubaida's warning in mind, one might still reasonably hypothesize that belief systems extolling and emphasizing self-sacrifice and an afterlife facilitate risky operations and tactics such as suicide bombing. Terrorism conducted by Islamist groups is deadlier, on average, than that of other groups. Even if we remove the attacks of September 11 from consideration, primarily Islamist and partially Islamist groups still account for a somewhat disproportionate share of terrorist fatalities between 1998 and 2007, larger than that of nationalist groups. This suggests that Muslim terrorist groups tend to adopt more effective tactics, such as suicide bombings, than other kinds of terrorist groups. Groups like the aforementioned secular Kurdish nationalist PKK only engaged in a handful of suicide bombings since it launched its insurgency in 1982 (primarily aimed at hard military targets rather than civilians), which might leave the non-Muslim Tamil Tigers (specialists in suicide attacks) as the exception that proves the rule. Secular terrorists may have trouble painting as motivating a picture of the afterlife for would-be martyrs, and terrorists unafraid of death appear much more effective at taking others with them. Advocates of a distinction between "old" and "new" terrorists also argue that the "new" terrorists (e.g. Islamist terrorists) seek to maximize casualties, while the "old" terrorists sought to maximize audiences. 12 Had our dataset's time period included the rise of ISIS, such an interpretation would probably gain even more weight: Indeed, ISIS seemed intent, with its videos of mass executions and other grisly actions, maximizing both casualties and audiences.
Alternately or additionally, the state sponsors of some Islamist and nationalist/Islamist terror groups may have provided more effective support than most non-Muslim groups enjoyed. Statesupplied plastic explosives in particular may help explain higher fatalities for Islamist groups. Iran, Syria and, until recently, Libya, stand out on U.S. watch lists as some of the most active supporters of terrorism. Private donors in some Gulf Arab countries may have likewise offered higher levels of material support for Islamist terrorist groups than most non-Islamist groups could hope to garner. The higher number of casualties caused by Islamist groups has in turn drawn a greater share of world attention than other terrorist campaigns, particularly since 11 September 2001.

Conclusion
Clearly, no singular explanation can clarify the myriad causes and manifestations of terrorism. Edited volumes such as Tore Bjorgo's The Root Causes of Terrorism (2005) gather together more than a dozen scholars on the issue, advancing three times as many explanations for terrorism in different contexts and at different times (Bjorgo, 2005). Social scientists must continue to look for both general and specific explanations of complex social phenomena such as terrorism, all the while aware of the problems relating to multi-causality in a very complex world. This paper nonetheless sought to address what the authors saw as a glaring gap in research on terrorism. As researchers attempt to understand terrorism, even basic assumptions should not be taken for granted. Qualitative literature on the "new terrorism" needs the support of quantitative data to ensure that common perceptions are grounded in fact. Although the difficulty of obtaining accurate quantitative data on terrorism can prove daunting and frustrating (especially when some of the better sources of data on terrorism suddenly get withdrawn), skipping this step often leads to unstated and unjustified assumptions.
The data that we were able to analyze suggest that in contemporary terrorism nationalist ideology is a more common trait than religious ideology. Of the 560 groups active within our roughly ten-year time frame, 40% of those groups had a nationalist element. Furthermore, there were more solely secular nationalist groups than solely religious groups. Despite the prevalence of secular nationalist groups, however, around 25% of all terrorist groups had an explicitly Islamist aspect to their ideology. Within the subset of non-secular (e.g. religious) terrorist groups, the vast majority were Islamist (96%).
The data also shows why more focus is currently placed on Islamist groups than nationalist groups. Comparatively, religious and most commonly Islamist terrorist groups produce more fatalities than other kinds of terrorists. Groups classified as solely Islamist are credited with more fatalities than all secular nationalist groups combined. Although the TKB data attributed Christian groups with a significant number of fatalities (more than 500), the total was small compared to nationalist groups and especially Islamist groups. This study thus challenges the view that contemporary terrorism is primarily religious in nature, but supports the assumption that Islamist terrorism, specifically, threatens higher body counts. This trend would have been even more pronounced had the dataset been able to include the post-2008 time period, during which ISIS made headlines around the world with its attacks in Syria, Iraq, Europe and elsewhere. Given this tendency towards particular lethality on the part of religious terrorist groups, counter-terrorism efforts focused on containing the spread of radical Islamist groups or other religious extremists do not seem out of place. deprivation, colonialism, neo-colonialism, imperialism, sexual frustration, rigid social structures, exclusion, discrimination, alienation in the face of modernization, lack of economic opportunities, or authoritarianism and a general "democratic deficit." 7. See, for example, Pape (2003) and Byman (1998). 8. According to Abadie, however, the correlation functions in a non-monotonic manner: "Countries with intermediate levels of political freedom are shown to be more prone to terrorism than countries with high levels of political freedom or countries with highly authoritarian regimes. This result suggests that…transitions from an authoritarian regime to a democracy may be accompanied by temporary increases in terrorism" (2004:3). 9. The PKK's suicide bombings, which mainly occurred in 1998 and 1999, appeared to target Turkish military and police forces rather than civilians. 10. Prominent examples of these kinds of approaches to explaining conflict in general include Tajfel (1970); Horowitz (1985); Burton (1979). 11. Given the startlingly high number of works that address terrorism from this kind of perspective, it strikes us as more prudent to avoid an attempted listing of some of the most representative or authoritative works of this genre. 12. See Crenshaw for a critique of such a distinction between "old" and "new" terrorism.