The challenge of establishing the impact of terrorist organisations: development of a database on ETA’s activities

ABSTRACT This article presents a new account of the size and scale of ETA’s activities. Highlighting a range of political considerations that shape existing records of impact, the article traces the deployment of a range of distinct methods of data collection and evidence verification in the creation of the most comprehensive database of ETA’s activities produced to date. The resulting database suggests that ETA’s impact is significantly greater than previously recognised, with a total of 1,047 attacks from 1959 to 2010. Within those attacks, 66 people were kidnapped (32 of whom were killed), 956 people were killed, there were 1949 injuries and 3 post-incident deaths via suicide and cancer that were attributed to trauma. Of the attacks, 503 were bombings, 456 were shootings and 66 were kidnappings. The remaining attacks used other forms of violence (Molotov cocktails, arson, electrocution or being thrown from moving vehicles). This comprehensive account suggests that previous records underestimate the total number of people killed by over 100, with injuries often not even recorded formally. As more murders are detected and attributed to ETA, the methods deployed within this new database will ensure that the impact continues to be tracked. Those methods enable scholars of terrorism to track the impact of organisations more broadly. As such, this article serves as a means of fostering real discussion on methods more broadly, particularly in terms of criteria adopted for a range of political reasons.


Introduction
Establishing the scale and scope of a terrorist organisation's activities is essential to understanding the social, economic and humanitarian impact on the organisation's target, and more broadly, on society. The impact of the Basque terrorist group, Euskadi Ta Askatasuna (ETA), both within and beyond the Basque Country is simultaneously well documented and hotly contested. First, there is disagreement on the number of victims that ETA's campaign claimed. Second, while some studies have used the number of attacks as their unit of measurement, this does not allow differentiation in terms of impact related to the scale of the event. Indeed, ETA's attacks range from those targeting individuals to bombings that killed or injured over 100 people. The difficulty arises in defining a "victim", assuming that this is more than the binary measurement of mortal victims.
To capture broader detail of the impact of terrorism, a range of different methods have been deployed. Enders and Sandler (1996), for example, set out the sources and extent of the economic impact. However, arguably the most important variable is a political one: the definition of victimhood depends upon the contextual interests and motivations of the actor collating the data. As such, there are widely divergent accounts of the number of ETA's mortal victims according to the Spanish Government, victims organisations, and Non-Government Organisations (NGOs). In addition to this, the European Parliament conceded in 2022 that over 300 murders in Spain that could still be attributed to ETA remain unsolved (Petitions 2022).
Academics at the University of the Basque Country (Universidad del País Vasco/Euskal Herriko Unibertsitatea) highlight a number of deficits in holistic recording of ETA's impacts, going so far as to state that ETA's victims have not received sufficient attention in literature to date (Llera and Leonisio 2015). The consequence is that the full scale of ETA's activities is subject to ongoing debate, with real consequences for those who require recognition as victims in order to secure official compensation and/or support. Should this be true of the activities of a Western European terrorist organisation, it is likely to be even more true of other organisations in different parts of the world with less efficiently organised records.
This article describes the process of compiling a database of ETA's impacts for nonpolitical purposes. The findings are the result of interrogation and synthesis of multiple existing databases and datasets. It sets out a series of challenges that are likely to affect any such effort and concludes by suggesting that the total number of victims of ETA is higher than currently recorded by any single existing database. This offers pathways for colleagues attempting to conduct work on ETA and other terrorist organisations. In detailing this new account of ETA's impact, we first outline the reasons for deficits in existing databases, arguing that political considerations have influenced the methods used.

Measuring terrorism's impact
Numerous attempts have been made to capture the full impact of terrorism through a range of indicators and units of measurement (Frey, Luechinger, and Stutzer 2007). In recent times, these efforts have focused on economic impact, which is seen as a difficultto-measure but significant impact of terrorism. According to Enders and Sandler (Enders and Sandler 1996), the economic impact of terrorism has at least five sources: First, tourism revenue is affected by terrorism's discouraging of visitors (Liu and Pratt 2017), while governments can also dissuade tourism through official guidance and bans (FCDO 2022). Second, Foreign Direct Investment (FDI) is heavily influenced by an area's security, but not always in the linear direction presumed. A study by Ari and Ibrahim (Ari and Ibrahim 2020) concludes that, while FDI is impacted negatively in the immediate and short-term aftermath of a terrorist event, there is a potential long-term positive FDI impact, especially where domestic investment through development follows (Muhammad, Wen, and Haseeb 2019). Third, there is damage to, and disruption of, infrastructure, either through repairing damage, installing terrorist-proof infrastructure or restricting business in the vicinity of an attack. Climate change, natural disasters, political unrest and cyber-attacks all mean that separating direct terrorist impacts from other impacts is problematic (Ebrahimnejad and Khanbaba 2021). Fourth, investment in counterterrorist (CT) activities may be state reactions (or overreactions) that increase economic costs (Chowdhury and Fitzsimmons 2013). Fifth, and finally, there is investment in direct concessions with terrorist organisations (blackmail and ransom), which is seldom monitored or reported accurately given government proscriptions of ransom payments.
These efforts have highlighted the complexity of measuring the impacts of terrorism, introducing a series of conceptual and methodological difficulties to research. Most importantly, however, the focus on economic impacts has shifted attention away from the direct physical impacts of terrorism in terms of violence and coercion, which is presented here as a description for kidnapping. In part, this is because there is an assumption that counting those who have died or been harmed is less problematic. Indeed, in many cases, the numbers of fatalities are publicly available and verifiable via multiple sources. However, there are serious challenges, meaning that records of victims differ widely in their scope and scale. There is, for example, debate about what counts as a direct cause of fatality or injury, with some sources including those who suffer heart attacks from the biopsychosocial impacts of events. This extends to the mental health impacts of terrorism, including suicides of terrorist attack survivors. There is also debate on what constitutes a recordable injury, which differs according to definition of severity. The definitions deployed have a political basis, since governments and authorities bear responsibility for recording impacts and, in many cases, compensating or providing ongoing support to victims. It is this political consideration, and the overlapping and competing interests of governments or authorities and victims, that underpins ongoing debate over the scale of impacts of terrorism in a wide range of contexts.

ETA
ETA was active from 1959 until the group disbanded and disarmed in 2018, with its last victim claimed in France in 2010. During their active years, most measures assert that ETA was responsible for carrying out over 3,300 attacks within and beyond the Spanish borders, predominantly targeting Spanish law enforcement officers and their families, politicians and their families, journalists, and physical sites of importance to the Spanish or Basque governments. The exact number of failed attacks is not known, as ETA has thus far rejected all requests to disclose all historical elements of their campaign fully and publicly. However, the precise recorded number and nature of incidents and their impacts differ significantly across multiple databases and datasets, with differing formats, methods, and purposes. Put simply, there is no definitive or commonly accepted number of victims attributed to ETA. The total number of individuals killed by ETA is heavily disputed within and beyond the Basque Country, but generally ranges in number between 800 and 952, with the number of injured presented as exceeding 2,000 people and an additional unknown number of inhabitants displaced from the targeted areas. According to the Spanish and Basque governments, major global news agencies, and organisations representing the victims of ETA, the total number of victims is accepted as 829 (FVDT 2017). The Basque Victims Association, however, states 843 as the number of victims (Alonso, Domínguez, and García Rey 2010).
In order to understand the relationship between attacks and a range of other phenomena, and public opinion in particular, there is prima facie reason to believe that all of ETA's attacks may carry importance. Given that terrorism is directed at shifting opinion through invocation of fear, data on attacks resulting in fatalities, kidnappings, injuries and material damage are all relevant. De la Calle and Sánchez-Cuenca explain that instilling fear is usually one of the primary goals of terrorists (De la Calle and Sánchez-Cuenca 2011). Terrorist campaigns are aimed at leveraging fear in achieving political settlements with state organisations. In order to understand the relationship between types of attack, public opinion and political outcomes, it is necessary to produce an exhaustive, effectively-sourced catalogue of attacks.

Divergence within databases
There are several key sources of data on ETA's impact, ranging from those collated globally to those recorded at local level. Each provide significantly different accounts of the extent of impact in terms of violence and coercion. The following list provides details of the problems encountered with each source when attempting to use it as a single source of truth.

The Global Terror Database (GTB)
The GTB is managed by the Global Consortium for the Study of Terrorism Responses to Terrorism (START) and includes more than 200,000 terrorist attacks (START 2022). It is, as its name indicates, a global database and therefore includes data related to, but not exclusively focused on, ETA. It is hosted and maintained by the University of Maryland, an Emeritus Centre of Excellence of the U.S. Department of Homeland Security. While it contains data from the global stage, it has a bias towards events that have an impact on the U.S. The first iteration of the database was completed in 2006, with the authors explaining the need for the database to address the lack of cutting-edge empirical data, with the aim to "code and verify a previously unavailable dataset composed of 67,165 terrorist events recorded for the entire world from 1970 to 1997" (Lafree et al. 2006, 6).
While this database has a broad set of data, there are two main issues that render it problematic in terms of data on ETA. First, the GTB only records terrorist events from 1970 onwards, leaving 11 years of ETA activity untracked. Second, its data labelling and metadata are complex and ambiguous with regards to conflict in the Basque region, which precludes verification. An example of how this manifests itself is in searching for attacks where ETA is regarded as responsible. The list of attacks in the region include 3,761 instances in which the perpetrator is known and a further 89,231 attacks in which the perpetrator is unknown. The database listed as possible perpetrators of violence in the name of Basque independence nine different actors: a) Basque Country Autonomous Self-Defence Group; b) Basque Extremists; c) Basque Fatherland and Freedom (ETA); d) Basque Guerillas; e) Basque Justice; f) Basque Rectitudes; g) Basque Refugee Support Group; h) Basque Separatists; i) Basque Terrorists. While ETA appears on this list as a standalone group, there is a lack of clarity on whether attacks attributed to the groups in the other categories might also have been carried out by ETA or whether, as is likely, the categories overlap. This renders attribution problematic.

The Fundación Juan March (FJM) dataset
FJM (March 2022) is described as a list of information about fatalities caused by Basque nationalist terrorist organisations during the period 1960-2006. The Foundation, based in Madrid, promotes scientific investigation with a view to furthering Spanish culture and community. The creators state that "the dataset is more precise and offers more information than other such existing lists and datasets". While the data labelling and metadata are simpler than the GTB, there are two key deficits. First, the sole unit of observation is fatality, meaning that there is no record of non-fatal violent or coercive acts. Second, as with the GTB, the database does not cover the entire period during which ETA was active; the FJM dataset omits data from the first year and final twelve years of ETA's activities.

The Spanish government
The Spanish Government keeps an updated list of the fatalities linked to terrorist attacks (España 2022a). Under Spanish law, the Spanish State is legally obliged to pay damages and/or compensation, as well as providing ongoing support to victims of terrorism and their families (España 2022b). The principal purpose of the Spanish government's list, therefore, is to keep an accurate and updated count of deaths in order to ensure it meets this legal obligation. While regular updating prevents issues of data currency, it does not catalogue non-fatal injuries or kidnappings, or fatalities linked to the indirect impact of terrorist attacks.

The Universidad del País Vasco/Euskal Herriko Unibertsitatea (UPV-EHU) kidnappings database
The UPV-EHU kidnappings database (Llera and Leonisio 2015) presents itself as the "first systematic database, based on documentary information, about the circumstances around the kidnappings committed in Spain by ETA and its related organisations". The database was compiled by academics at the University of the Basque Country (Universidad del País Vasco/Euskal Herriko Unibertsitatea). While this database is an important source of information regarding kidnappings (whether or not they resulted in fatalities), it does not cover events after 1997, with the final recorded ETA kidnapping according to broader sources occurring in 2007 (Figaro 2007). It also only deals with kidnappings, and therefore does not record the majority of deaths related to ETA's campaign.

The Asociación Catalana de Víctimas de Organizaciones Terroristas (Catalan Association of Terrorist Victims) (ACVOT)
ACVOT updates and publishes a chronology of all Catalans killed in terrorist attacks around the world since 1971. The main purpose of this list is to ensure that Catalans receive recognition under the Spanish/Catalan Government's compensation and support schemes. While this list has detailed metadata regarding the type of attack, the perpetrators and details of the victims, it is limited in three important ways: 1) it only catalogues victims who are Catalan (regardless of where the attack took place and who the perpetrators were); 2) it only lists fatalities; 3) it does not cover the years prior to 1971.

The 'Mapa del Terror' of the Colectivo de Víctimas del Terrorismo (the Spanish terrorist victims' organisation) (COVITE)
The "Mapa del Terror" (the Terror Map) (COVITE, Mapa del Terror 2022b) aims to help "develop and promote investigations that help document and contextualise the history of terrorism for future generations" (org 2022a). While the dataset is comprehensive and up to date, the data is presented and accessed via a map, and the search criteria only allow search by the following: a) Location of the attack; b) Occupation of the victim; c) Age of the victim; d) Gender of the victim; e) Group/perpetrator responsible for the attack; f) Weapon used; g) Country (France or Spain); h) Province, i) Town. While there is information in the dataset about kidnappings, the dataset focuses on fatalities and uses this as its unit of observation, and it is not possible to search on attacks where there were no fatalities. Injuries are recorded when they are part of an attack where there was a fatality, and the underlying data is not available in raw format.

The AROVITE archive (Archivo Online sobre la Violencia Terrorista en Euskadi)
AROVITE is part of a site dedicated to terrorist events in the Basque Country (AROVITE 2022). The organisation states that it brings together "texts concerning the phenomenon of terrorism in the Basque Country and Navarra, published in digital format based on research". The AROVITE site consists of journal articles, books, reports, book chapters and working papers. The works come from different disciplines: history, sociology, political science, psychology, health sciences, economics, ethics, philosophy, anthropology, law or science communication. The archive lists fatalities due to terrorist events in the Basque Country and catalogues them based on the different factions of ETA responsible for each attack. The limitations of this dataset are threefold: 1) the data starts in 1968; 2) it only counts fatalities (no injuries or kidnappings) and 3) there are no additional, contextual details around the individual incidents (simply a number count for each year).

El Correo newspaper
In 2010, the Basque newspaper, El Correo, created and published a list of over 850 deaths attributed to ETA (Correo 2022). Once again, this dataset uses the fatality as its unit of observation. In addition to this, the number that it catalogues is not consistent with other sources and the underlying data is not available, making it difficult to search, analyse and categorise.
As analysis of each of these widely cited records demonstrates that there is no clear, single, comprehensive source of record for ETA's terrorist attacks and their impact. While there are multiple lists of fatalities, there is no single source that simultaneously covers the entire duration of ETA's activity and also records different kinds of impacts in terms of fatalities, injuries and kidnappings. Indeed, the databases are definitively not exhaustive: many major news organisations, including those in Spain, have published lists of terrorist attacks that identify activities not recorded in those accounts above.
An example of this is the work carried out by the Spanish newspaper, El Mundo, in which it aggregated its own coverage of ETA (elmundo.es, 2009). Economic crimes (blackmail, extortion or robbery) that did not cause physical human harm have not yet been extensively catalogued due to resource constraints and availability of data, while victims of ETA's "terrorism tax", which includes large enterprises, are understandably reticent to come forward and give an account of their interactions with the terrorist group. Given the need to understand the nuances of impact of ETA's violent and coercive activities there is need for a comprehensive dataset that covers the whole duration and is not subject to politically driven criteria.

Methods
In order to create a comprehensive account of ETA's activities across the whole of its 1959-2018 lifespan, 10 key datapoints necessary to enable examination of impact on public opinion were established: (1) Date of the attack (day/month/year) (2) Type of attack (kidnap, murder, injury) (3) Total number of people impacted by each attack and the type of impact (deaths, injuries, kidnappings) (4) Intended target and actual victims (for example, in targeting a police station, civilians may have been killed and/or injured) (5) Method/weapon(s) used in the attack (car bomb, shooting, parcel bomb etc.) (6) Additional contextual information (relating to the victim(s) and/or the attack) (7) Location of the attack (8) Source(s) used to verify information about the attack (9) Recorded public reaction to the attack (where relevant and available) (10) Evidence of prior warning given by ETA (Y/N) Recording these datapoints necessarily creates a broader account than those discussed in the list of sources above and enables detailed analysis of causality.
In order to develop reliable data in an era of "fake" or politically motivate news, multiple methods of verification were identified. First, Fritch and Cromwell's (Fritch and Cromwell 2001) means of addressing technological aspects of verification were adopted. This means that stories were traced back to the originating author(s) and institutions to verify affiliations and confirm credibility. Second, all data analysis was conducted using Caulfield's (2017) Stop, Investigate the source, Find better coverage, Trace (SIFT) approach, which built on the earlier Currency, Relevancy, Accuracy, Authority, Purpose (CRAAP) test. SIFT and CRAAP are both widely used across academic institutions in the US in training researchers. SIFT requires the researcher to complete a similar process of source interrogation in order to verify the expertise and agenda of the originator. This requires researching secondary sources regarding both the source and its author(s), including the use of fact-checking databases. In light of the approach, the categories of applicable data sources were reduced to: (1) Academic sources from within and beyond the Basque Country that are peer reviewed (2) Official government sources (EU, US, Basque, Spanish, Catalan, French) (3) Existing victim databases (from the list in the previous section) (4) Terrorism-related NGOs (EU, US, Spain, France) (5) News and newspaper archives (Basque, Spanish, French, English) A search strategy for each of the categories was adopted. This proceeded under the following parameters for each of the sources: (1) Academic sources: Article headings and titles from contents pages and indexes as well as citations related to ETA and victims. This search was carried out in Spanish, French and English.
(2) Official government sources: Databases owned by Spanish, Catalan, Basque and French governments with reference to terrorism or ETA.
(3) Existing victim databases: Key search terms and keywords (ETA attack, terrorism, victim, fatality, kidnap). These terms and keywords were used in English, Spanish, Basque and French to ensure coverage across international organisations. (4) Terrorism-related NGOs: Key search terms and keywords (ETA attack, terrorism, victim, fatality, kidnap). These terms and keywords were used in English, Spanish, Basque and French to ensure coverage across international organisations. (5) News and newspaper archives: Key search terms and keywords (ETA attack, terrorism, victim, fatality, kidnap). These terms and keywords were used in English, Spanish, Basque and French to ensure coverage across international organisations.
Once data on each attack were identified, they were cross-referenced against sources from each of the five data source categories. Each data source corroborating an attack was recorded. Wherever possible, sources were used in their original published language (Spanish, French or English), but for other languages (e.g. Basque) official translations were used. Where electronic sources were available, the source web address (url) was recorded. While all sources are publicly available at the time of research, the practice of erecting paywalls around data is commonplace. A subscription may therefore be required at a later date in order to access some sources. The database has been made available to others via the Open Science Framework (https://osf.io/) and has been shared with the University of the Basque Country (Universidad del País Vasco) and the Centre for Basque Studies.

Results
In examining the database, we see a total of 1047 ETA attacks from 1959 to 2010. In terms of the number of individuals impacted, we have recorded: (1) 66 people kidnapped (32 of whom were killed) (2) 956 deaths (3) 1949 injuries (4) 3 post-incident deaths (suicide and trauma-induced cancer) Of the attacks, 503 were bombings, 456 were shootings and 66 were kidnappings. The remaining attacks used other forms of violence (Molotov cocktails, arson, electrocution or being thrown from moving vehicles).
In looking simply at the number of deaths, this number differs significantly from other sources. The Spanish government cites the number as 829 (España 2022a), the Associated Press uses 853 (APNews 2020) and Reuters uses 850 (Reuters 2018). The Fundación Juan March lists 834 deaths (March 2022) and AROVITE lists 843 (AROVITE 2022). In all instances, our dataset records a higher number of deaths. As more murders are resolved and attributed to ETA, we expect this number (and the dataset) to grow.

Discussion
The database produced is a superset of data gathered and stored elsewhere in discrete datasets. In attempting to collect data on all 10 criteria stated previously (date, type, magnitude, method/weapon, context, location, source, public reaction and warning), it is not possible to consistently gather this from a single source. While many sources record the data, type, method/weapon and location alongside details of victims, it is unusual to find contextual information and data regarding public reaction in factual datasets that are interested in the cataloguing of victims. The most data-rich source for contextual and impact information has been news and newspaper archives from local, national and international news outlets. This has required extensive research across local, national and international news sources. The established SIFT method and base dataset enables integration and expansion of the set as new data emerges. This is particularly important in the context of ETA where additional information becomes available over time. ETA's influence in and beyond the Basque region continues to this day, with over 100 Etarras (ETA terrorists and their supporters) still in prison in Spain (Monde 2022). The fear of repercussions is still prevalent, despite ETA's dissolution, but its reduction is enabling new accounts to emerge. These can now, for the first time, be integrated into a single dataset.

Conclusion
In gathering and publishing this data, a more complete record of the human impact of ETA's campaign is now available. The new dataset also presents itself as a framework that may be revised as more data surfaces, or as amendments are made to existing data. As stated earlier in this article, the European Parliament estimates that there are still more than 300 unsolved murders that could be attributed to ETA. Any detailed and comprehensive study regarding ETA's campaign requires a complete view of their attacks; the new dataset enables this.
In terms of new findings, the dataset has shown a significant increase in the deaths attributed to ETA during the course of its campaign. These include victims whose deaths may be indirectly attributed to their attacks as a result of suicide or trauma-related cancer. This has potential for use in obtaining official recognition for victims, which has implications for funding and support from governments in Spain. In having a complete set of data around each attack, it will now be possible to align attacks to other events in Spain/ the Basque Country at the time. For example, a study into changes in public opinion as a result of attacks will be facilitated due to the understanding the scope and scale of each attack. In turn, it will be possible to plot the success of ETA's campaign against their attacks, taking into consideration leverage they had with the Spanish governments or the international community in terms of negotiations or recognition.
When looking to understand terrorist organisations more broadly, the approach taken to compile this database could be replicated to create complete records in instances where there are gaps or incomplete data. In carrying out more detailed studies into the fortunes of ETA, this data could inform counterterrorist policy, media coverage and negotiations for other governments and organisations developing a strategy to deal with terrorism.

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

Notes on contributors
Kathryn Loosemore is a Doctoral Candidate in International Relations at Northumbria University and Chief Operating Officer at Kantar Worldpanel. Her research focuses on the relationship between ETA's activities and public opinion.
Matthew Johnson is Professor of Politics at Northumbria University. His research focuses on the relationship between public policy and health and social outcomes.