The Global Terrorism Database (GTD), which is managed by the University of Maryland’s federally funded National Consortium for the Study of Terrorism and Responses to Terrorism (START) recently released its 2015 user-friendly database to the public. The significance of such a robust catalogue of terrorism is that it is used to generate scholarly articles and to supply the Department of State, among other agencies, with a factual basis on which to author official reports. In turn, such reports may be used to buttress open-source foreign policy assessments and perhaps inform decision makers. It is therefore very important to understand the degree of accuracy associated with the GTD’s statistics, as well as the attendant policy implications that might be drawn from them. Although there are many areas in which to focus one’s review of the database, this article is primarily concerned with the inferences that might be mistakenly arrived at with regard to the situation in Israel. I have found a few trends, and some factual errors, which I will highlight and analyze below. My aim here is to alert future researchers and decision makers to the distortion of material facts, and the attendant conclusions, that may be discerned from this particular macro-level database.
The “Israel” Dataset
The GTD indicates that there were 58 incidents of terrorism in Israel (excluding the West Bank and Gaza) in 2015. Of that number, 8 were said to be committed by “Israeli Extremists” (7), or “Jewish Extremists” (1). In reviewing these instances, I found that one such event may not have occurred at all (201510030023). The single source cited could not be found in any online search (e.g., Lexis Nexis, Google Scholar and general Google searches yielded no results). Additionally, no attack type, no weapon type, and no named perpetrator were listed in the database. Nonetheless, the GTD summarized the incident in the following way, “Assailants attacked an Arab driver in Jerusalem, Israel. A pedestrian was wounded in the attack. No group claimed responsibility for the incident; however, sources attributed the attack to Israeli extremists” [emphasis added]. A single illusive source is hardly evidence of an event having transpired.
In researching another incident (201510130038), committed by “Israeli Extremists,” it initially appeared to me that the GTD missed an Arab terrorism event (in East Talpiot/Armon HaNatziv). That attack resulted in 2 Israeli deaths, one of whom was an American citizen, despite the fact that the GTD reported zero American deaths, coded as “nkillus,” for 2015 inside Israel. This came to light as I was reading the two source documents the GTD attributed to the “Israeli Extremists” attack. Both articles cited 4 other Palestinian attacks, of which the GTD only accounted for 3. I did discover, however, that the GTD coded the East Talpiot/Armon HaNatziv event as occurring in the “West Bank and Gaza Strip” dataset. The database specifically names the intersection of Olei HaGardom St. and Moshe Barazam Street [sic; Barzani is the correct spelling] as the location in which the attack occurred. Although the attack did take place on the first of the two named streets, there is no indication that it did so at the intersection of the second.
A reasonable question to ask is why such a minute detail matters for the integrity of the database. The answer is that if the attack were to truly have taken place where the GTD says it did, then it would support a narrative that the lethal attack on Israelis (and in this case an American) occurred in the so-called “occupied territories,” thus lending credence to the oft-cited attack rationale associated with “resistance to the occupation.” Indeed, the GTD gives the coordinates 31.749712/ 35.236954, which is a Jewish neighborhood in East Jerusalem. This seemingly minor detail also informs the analyst as to where the GTD has decided to draw its own lines.
The “West Bank and Gaza” Dataset
Beyond the “recognized” borders of the State of Israel, which the GTD has apparently defined for researchers (see my above analysis), they consider the “West Bank and Gaza” in a separate dataset. With regard to Jewish or Israeli perpetrators, they code them as either “Israeli Extremists,” (always listed as plural) or “Israeli Settler” (always listed as singular). Such a distinction appears to be arbitrary, but one matter worth consideration is that to call someone a “settler” undermines the legitimacy of their legal status as a citizen. In this respect, the GTD staff is discriminating in a way that appears to be other than neutral.
To summarize the statistics, the combined total number of attacks for 2015 in the “West Bank and Gaza” dataset numbered 247, with 120 non-perpetrator deaths (including 3 U.S. citizens), and 242 injuries (including 3 U.S. citizens). The number of attacks attributed to “Israeli Extremists” was 5 and “Israeli Settler” was 14, for a total of 19 out of the 247 attacks, or just under 8%, being committed by Jews/Israelis. It should be noted that all attacks in Gaza, which numbered 32, exclusively featured Arab attackers and/or Arab victims and resulted in no fatalities.
With regard to the “West Bank” portion of the dataset, looking closely at the alleged attacks by Israeli settlers/extremists, reveals some strange coding preferences that require further analysis. For instance, on October 9 in Kiryat Arba, an event (201510130063) occurred, for which the perpetrating party was coded as “Israeli Settler.” According to the GTD, the incident resulted in no injuries and no weapons were used. The single source upon which the GTD relied was from a Palestinian news agency, Ma’an, which produced pictures that didn’t necessarily indicate an act of violence per se. Rather, while carefully reviewing the photographic “evidence,” it appears that the Palestinian journalists entered a Jewish neighborhood to report on an incident occurring nearby, in which a Palestinian dressed as a journalist, stabbed an Israeli soldier, and were subsequently being forced to lower their cameras. This event took place during the most violent month of October and given the above-described specific circumstances was likely a defensive measure, rather than an act of “terrorism” that the GTD ascribes to “Israeli Settler[s].” In the database, there is room to indicate if there is doubt that an act is terroristic in nature. None was conveyed for the instant case.
Another incident is even more troubling, because it actually didn’t happen. While attempting to corroborate the single source cited for an alleged attack on a mosque in Mughayir, it became apparent that the GTD actually recycled an article from al Jazeera that dated from 2011. Indeed, all of the alleged “facts” in the GTD database were identical to the story that was promulgated by al Jazeera four years earlier.
For two “terrorism” incidents (201502250058/201502250059), which involved the alleged firing of automatic weapons at Palestinian homes, the GTD used a single, uncorroborated, source: The Muslim News. Judging from the language used in the article, there is reason to believe that it may not have occurred. Even still, there could at least have been an indicator as to the veracity of its alleged occurrence. The GTD’s staff doesn’t issue such assessments. That is the job of the analyst.
While investigating a vandalism attack (201503040051), which likely did occur and was categorized by Israeli authorities as “Jewish terror,” according to one of the two sources the GTD used, an event (201411120036) from the previous year was mentioned. In that event, it was alleged by local Palestinians that a mosque in Mughayir was set ablaze by Jews. As the article points out, however, the police determined that the cause of the fire was faulty wiring. To their credit, the GTD did state that it may not have been terrorism, although the incident remains in the database from 2014. The unfortunate result is that deficient Palestinian engineering is given the equivalent status of an act of Jewish terrorism in the database.
In another likely case of an electrical fire having been mistaken as a Jewish terror attack (201508240041), the GTD, again to its credit, did indicate doubt that terrorism was afoot. However, as in the previous unlikely incident, this too is counted in the aggregate as resulting from Jewish terror.
Severity of Jewish/Israeli Attacks
For all of the purported Israeli attacks inside Israel (not including the West Bank and Gaza), only 1 (5%) resulted in the death of a (non-Arab) civilian, whereas the total number of recorded deaths (not including those of the perpetrators) in Israel for 2015 was 20. In the instant case, the only one in which the GTD labeled the single assailant as “Jewish Extremists” [sic], the perpetrator attacked a Gay Pride parade for the second time and was consequently sentenced to life in prison. This event was very much an anomaly in that attacks against fellow Israelis for their sexual orientation is a rare occurrence. What is notable is that not only does the State of Israel apprehend, prosecute and punish Israelis who perpetrate violence against Arabs, but also against those who attack gays. Although I may be stating the obvious, for context’s sake it is well-known that Israeli tolerance of others’ identities is evidenced by the fact that it is by far the only gay-friendly country in the region (to include Arab-controlled territory in the West Bank and Gaza).
With regard to attacks by Jews/Israelis in the West Bank (recalling that Gaza had none), only 1 of the 19 was lethal, but it resulted in 3 deaths, accounting for about 2.5% of all deaths caused by terrorism in this particularly named area. It was the July 31 arson attack against a Palestinian home, for which a man and a minor were originally indicted, although it remains unclear if the man had acted alone. Nonetheless, Israeli authorities have been actively pursuing the suspect’s potential ring of similar offenders. Again, we have prima facie evidence that the Israeli government takes Jewish violence seriously. Of the 19 attacks carried out by Jews/Israelis, 12 (63%) were with an incendiary device, 4 by firearms (21%), 1 chemical [tear gas] (5%), 1 melee (5%) and 1 unknown (5%).
What a database can’t provide is context. This effort requires an analyst. For example, in the “Israel” dataset, 2 attacks, or 25% of all Israeli/Jewish attacks, were arsons against churches. In another example, an attack was alleged to have occurred in Netanya (201510090056), in which the GTD cites “sources attributed the attack to Israeli extremists.” Again, there was an uncorroborated single source, which, when checked, referenced a different article in which the account (in Hebrew) described a back and forth exchange between Jews and Arabs, the latter of whom were allegedly shouting, “Allahu Akbar.” At the time in which this incident took place, 6 Palestinian terrorist attacks had already occurred inside Israel that month and 9 in the “West Bank and Gaza.” Indeed, October was the most violent month in 2015, featuring 26 out of 58 attacks, or 45% for the “Israel” dataset that year. Of this figure, 5 of the 26 attacks were coded as Israeli/Jewish, 19% of the monthly total, representing 63% of the 8 that occurred that year in the “Israel” dataset.
In the “West Bank and Gaza” dataset, October was, similar to attacks in the “Israel” dataset, the most violent month of the year. It featured 72 attacks, or 29% of the total number of attacks in the West Bank and Gaza. For a comparative assessment, the previous 7 months featured a total of 74 attacks. The proportion of attacks by “Israeli Extremists,” or “Israeli Settlers,” when compared to attacks by others in that month was 6 out of 72, or 8% (in line with the previously addressed annual percentage).
Combining the two datasets (see Appendix A), the data shows that of the total 305 terrorist attacks the GTD recorded for 2015, 27 (9%) were committed by Jews/Israelis, which if adjusted for the aforementioned fact that at least one attack didn’t occur and several others could not be substantiated would reduce the percentage to something around 8%. Privileging the raw figures provided by the GTD, during the month of October, the total ratio of Jewish attacks to that of the Palestinians was 11:98, or 11%.
Analysts may be inclined to explore two distinct causal factors to explain the October violence. First, any actions of the Israeli military should be considered. Such actions do not meet the GTD’s definition of terrorism and would therefore not appear in the database. Secondly, any widespread Palestinian media campaigns that aimed to incite violence, which the GTD also doesn’t include, should be investigated. A single Israeli military action, or Palestinian broadcast, would likely not account for the level of violence that persisted at a high tempo beyond the month of October. Therefore, one should gauge their search results with an eye toward intensity, duration and chronological sequencing.
Conclusion and Implications
The complete picture of the data is represented in Table 1 (Appendix A), which can be summarized as follows. Jewish terrorism during 2015 represented approximately 10% of the total monthly attacks for the year, with 4 months having no attacks. The monthly percentage never exceeded 11%, even in the most violent month of October. The overall statement about non-Jewish terror is that it was quite the opposite. The average monthly attacks were 90% and there were no months in which non-Jewish (read: Arab) attacks numbered less than 5. There were some months, however, when Jewish attacks were 0 and the non-Jewish attacks were as follows: January (15), April (14), May (10) and September (13). During months when Jewish attacks were 1, non-Jewish attacks were as follows: March (5), June (8), and in two extraordinary months, November (49) and December (42).
In light of these statistics, one of the trends that appear to be obvious is that the Palestinian attack intensity increased in October, with no apparent connection to any Israeli attacks having occurred the previous month. Another trend appears to be that once Israeli terrorism was at its peak in October, it quickly diminished to 1/month for the next two months. This was despite the third apparent trend, which was that non-Israeli terrorism slowly diminished during the same period. One could therefore say that Arab terrorism ratcheted up quickly, but down slowly (and painfully). One additional insight is that Arab terrorism appears causally unconnected with Jewish terrorism.
Although the GTD does capture a fair number of incidents, its database is largely populated by students, who respond to directions provided to them by algorithms. The result is that one who is unaware of the context of particular conflicts, as well as some of the nuances that attend it, may not generate an accurate threat picture with regard to terrorism. For instance, the Israeli Security Agency (Shabak) composes its own, publicly available, annual reports on terrorism. Its 2015 report reflects statistics that differ markedly from those of the GTD database, but more scholars and policy analysts may be inclined to use the latter, rather than the former.
The reasons that the GTD’s aggregate data differs from that of the Shabak’s are threefold. First, there is a difference in the way attacks are calculated. The GTD doesn’t include many of the failed or foiled attacks, nor does it consider most rocket/missile/mortar attacks originating in Gaza. This is a bit odd, considering that the GTD’s “Data Collection Methodology” includes “an intentional act of violence, or threat of violence by a non-state actor” [emphasis in the original]. One might be inclined to assume that the GTD considers Hamas to be a state actor. However, this would be out of step with the fact that they did record a Hamas assassination attempt within Gaza (201502160091) in 2015. Additionally, the GTD recorded (in the Israel dataset) only one rocket fired from Gaza (201510100023), but indicated that the assailants were “unknown,” therefore not coding them as “Palestinian Extremists.” To wit, there are only Palestinians living in Gaza and the act of firing on a civilian population is patently extreme.
The other requirements that the data must meet also render the exclusion of Hamas/Gaza as suspicious. According to the GTD’s methodology, two of the following three criteria must be met:
- The violent act was aimed at attaining a political, economic, religious, or social goal;
- The violent act included evidence of an intention to coerce, intimidate, or convey some other message to a larger audience (or audiences) other than the immediate victims; and
- The violent act was outside the precepts of International Humanitarian Law.
In looking at projectile attacks emanating from Gaza, one can’t logically conclude that it fails to meet the benchmarks set forth by the GTD. In fact, Hamas’s attacks qualify on each of the 3 grounds. It cannot reasonably be argued otherwise, unless the data coders are in agreement that such actions are somehow legitimate. This would place the GTD firmly in the camp of those who assert the rights of Hamas/Gaza to attack Israel. Such an inference, if it were correct, would undermine the credibility of START as a nonpartisan think tank. After all, the center is a “Department of Homeland Security Center of Excellence.”
The second reason that the GTD’s data differs from that of the Shabak’s has to do with the way in which it distinguishes territorial lines. As was already mentioned, the GTD has two distinct datasets for “Israel” and the “West Bank and Gaza.” Scholars who miss this fact will arrive at some very skewed conclusions. Even when accounting for this fact, however, the statistics are still vastly different.
Thirdly, the GTD doesn’t appear to retrospectively analyze the data it collects, although its leading scholars use the data to produce peer-reviewed journal articles and books. Again, I have listed a few occasions in which a matter was either reported to be a terrorist incident when it turns out not to have been, or an incident was said to have taken place when it seems likely that it did not. I have also pointed out the many events that the database has missed. The unfortunate consequence is that the GTD’s aggregate data enables those who conduct macro-level studies to manipulate it in such a way as to draw imprecise conclusions. This is the precise reason why I have offered a micro-level assessment of the GTD’s database with regard to a specific independent variable, Israeli/Jewish terrorism.
Appendix A: Visual Data
Despite the various faults with the data, identified above, the following is a visual summary of the GTD’s combined datasets, “Israel” and the “West Bank and Gaza.”
Table 1: Average Monthly Attacks in “Israel” and the “West Bank and Gaza” 2015.
Graph 1: Numerical Expression of Average Monthly Attacks in “Israel” and the “West Bank and Gaza.”
Graph 2: Percentage Expression of Average Monthly Attacks in “Israel” and the “West Bank and Gaza.”
David Firester is the founder and CEO of TRAC Intelligence (Threat, Reporting, and Analysis Consultants), which is a premier threat analysis firm. TRAC Intelligence provides threat assessment in the private sector.