Abstract: Two recent attacks in London, in November 2019 and February 2020, by two convicted terrorists released from prison resulted in a surge of concern about terrorist recidivism. However, statistics and the academic literature suggest that, on the contrary, terrorists are unlikely to relapse into violent extremism. A review of the judiciary files of 557 jihadi terrorist convicts in Belgium, spanning the three decades from 1990, confirms that less than five percent reengaged in terrorist activities. These findings bear significant implications for counterterrorism policies broadly and for sentencing and post-penitentiary measures more specifically.
The fear that terrorist offenders could go back to their ‘old ways’ after their release from prison is widely shared among security services and the public. Two recent attacks have just added to this fear. On November 29, 2019, Usman Khan stabbed two people to death near London Bridge, around a year after his release from prison.1 He had been sentenced in 2012 for planning terrorist activities.2 On February 2, 2020, Sudesh Amman was shot dead by police shortly after he started stabbing passersby in Streatham, South London, only 10 days after his release from prison.3 He had been convicted in November 2018 to 40 months in jail for possession and dissemination of terrorist material, in connection with the Islamic State.4
For more than two years, European security services have been raising concerns about the planned release of hundreds of jihadis from prison. This is a “worrying threat that we are taking very seriously,” a European official told Agence France-Presse in early 2018.5 In its annual report published in 2018, the Belgian intelligence service, VSSE, warned of a potential new wave of terrorism resulting from a “recidivism surge” among released extremists.6 The report observed that “many” terrorists convicted in Belgium between 2001-2011 had reoffended, while highlighting a “current and persistent trend of recidivism” among terrorist offenders.7 Meanwhile, European prison and probation officers have discussed this issue on several occasions, in the context of the E.U.-wide Radicalisation Awareness Network (RAN).a
Even before the recent London stabbing attacks, the management of released terrorist offenders had been identified as a political priority. The 2018 final report of the European Union’s High-Level Commission Expert Group on Radicalisation (HLCEG-R)b warned that prison “may only delay” the threat posed by extremists.8 In June 2019, the Council of the European Union, gathering all E.U. ministers of justice, adopted conclusions on “dealing with terrorist and violent extremist offenders after release.”9 In February 2020, the Under-Secretary-General of the United Nations Counter-Terrorism Office, Vladimir Voronkov, stated that reducing the risk of recidivism for returning foreign fighters and terrorist offenders was a key priority for the United Nations.10
Clearly, the dominant perception is that a significant number of terrorists could potentially recidivate. However, in contrast with this perception, this article argues that though terrorist recidivism hits the headlines when it occurs, it is a very rare phenomenon.
This article starts with a review of the debate and data on terrorist recidivism. After defining (terrorist) recidivism, it highlights the main findings from the existing literature, which points to low rates of terrorist recidivism. Subsequently, this article introduces a new dataset on jihadi offenders in Belgium, including more than 500 terrorist convicts between January 1, 1990, and the end of 2019. The key finding is that there is a low rate of terrorist recidivism in Belgium, a similar conclusion to other studies in the literature.
Part One: The Debate and the Data So Far on Terrorist Recidivism
Terrorist recidivism is not a new phenomenon. A famous precedent was Cherif Kouachi, one of the co-perpetrators of the terrorist attack against the staff of the French satire magazine Charlie Hebdo in January 2015, killing 12.11 Kouachi had been arrested in 2005 and sentenced in 2008 for his role in a network that had sent jihadi volunteers to fight in Iraq. However, he walked free of the trial as he had already served 18 months between 2005-2006.12 Many other cases of terrorist reoffending could be cited.c Long before Amman, Khan, or Kouachi, one could, for instance, mention Abu Musab al-Zarqawi (jailed in 1993),13 founder of al-Qa`ida in Iraq; or Ayman al-Zawahiri (jailed in 1981),14 current leader of al-Qa`ida.
Given these remarkable precedents, the rising concerns about terrorist recidivism must be understood in light of the unprecedented magnitude of the jihadi threat, particularly in prison. There are currently more than 4,000 inmates in Western Europe that are either returning foreign fighters, convicted terrorists, radicalized inmates, or inmates “vulnerable to radicalization.”d The planned release of a significant number of these inmates in the coming two years is a considerable source of concern. (For example, approximately 90 percent of the 1,700 terrorist convicts and radicalized inmates in France will be released by 2025.15)
The fear of recidivism is further reinforced by concerns about radicalization in prisons, which have been commonly described as “breeding ground for radicalization” or “universities of jihad.” As Alain Grignard,16 long a leading figure in Belgian counterterrorism, has argued “rarely do people come out of prison better than when they went in … they can come out even more motivated than before.”17 In the United Kingdom, an independent review of Islamist extremism in prisons concluded in 2016 that radicalization in prison was a “growing problem” that was poorly handled.18 In France, new research claims that the jihadi movement is exploiting prisons to regroup.19 According to this view, terrorist offenders tend to withhold or strengthen their extremist views in prison, or radicalize others, making them even more dangerous upon release.
Such assessments suggest there will be more Usman Khans and Sudesh Ammans to come. The problem, however, is that there is little more than anecdotal evidence to support these gloomy evaluations. In fact, the academic literature challenges such assumptions. As noted by Andrew Silke, radicalization in prison remains a marginal phenomenon, and the fear of potential radicalization is often higher than the actual radicalization.20 The same can be said about recidivism, as this article will highlight.
Yet, policy discussions over radicalization in prison and terrorist recidivism are too often shaped by unsupported assumptions or misperceptions. Many analysts argue or assume, without evidence, that terrorists are likely to recidivate. As a result, policies tend to be more driven by fear than evidence.e
Recent discussions in the United Kingdom illustrate this. An emergency bill to restrict the release of terrorist offenders was passed by Parliament, following recent attacks.21 This was approved despite evidence that Khan and Amman are more likely to be eye-catching outliers than a harbinger of things to come. Indeed, according to recently released figures, only six terrorist offenders have been reconvicted of a further terrorist offense in England and Wales, out of 196 offenders released between January 2013 and December 2019 (3%).22 While the concern about terrorist recidivism is understandable, particularly from the point of view of security services, it is unclear how longer prison sentences would reduce recidivism, or how this would address the correlated risk of radicalization in prison. Yet, fear and emotions seem to have dominated, and it is hard to see anything else than a knee-jerk (over)reaction recently in the United Kingdom.
What is Recidivism?
Before looking into the literature on terrorist recidivism, it is necessary to first clarify what is meant by recidivism. Indeed, there can be different understandings leading to very different results. In its traditional, narrow sense, recidivism can be conceived as two separate convictions, for distinct offenses. An even narrower definition focuses on individuals who recidivated during their period of probation or reprieve (“legal recidivism”). In these classical understandings, recidivism rates incorporate reconvictions for any type of offense (for instance, murder and tax fraud). These rates exclude, however, re-arrest that did not lead to reconviction.
Along these lines, terrorist recidivism can be conceived in two different manners. In one broad conception, it can be a person who is convicted (at least) twice, including at least once for a terrorism-related offense. This would include, for instance, a former terrorist offender reconvicted for a criminal offense. In a narrower conception, it can be defined as two distinct convictions for terrorism-related offenses.
Arguably, the latter definition encapsulates the main fear of security services and society, that is a released terrorist who would go back to terrorist activities and possibly seek to commit an attack. In contrast, the former definition encapsulates terrorists with a criminal past, or future, covering the so-called “crime-terror nexus,” which is indeed a growing focus in the academic literature as well as in policy discussions.23 This would include a larger group of individuals who are not the primary concern of intelligence services, such as former terrorists who returned to criminal activities. This broad definition is more likely to approach “ordinary” rates of recidivism, which equally do not distinguish between the type of offense.
Some researchers argue convincingly that what matters is not reconviction rates (recidivism), but whether an individual reengages in terrorist activities or not. After all, this is indeed the main security concern. Terrorist reengagement has been conceptualized as “a return to terrorism after a period of disengagement, regardless of whether the disengagement was involuntary or voluntary.”24 Involuntary disengagement can refer to imprisonment, not necessarily sanctioned by a judiciary decision, whereas voluntary disengagement occurs when a person distances itself from violence on its own initiative. This additional conceptualization is helpful and complementary with terrorist recidivism. It accounts for individuals who reengaged without being convicted.f For instance, Usman Khan and Sudesh Amman would not—under a definition looking at convictions—be counted as recidivists, since they were killed, and therefore not prosecuted, although they definitely reengaged in terrorism. Logically, according to such a definition, there are more cases of reengagement than recidivism.
Another important distinction in the academic literature is between studies that have looked into terrorist recidivism in a specific country over an extended period of time, generally relying on data provided by the authorities, and studies that have evaluated deradicalization or rehabilitation programs, focusing on the clients of these programs only. In the second case, the objective is not to measure recidivism rates, but rather the success (or not) of the program. Often, these evaluations are conducted in a very opaque manner by the same agencies that implement these programs, hence raising legitimate questions about their findings.25
Finally, it should be mentioned that the study of (terrorist) recidivism is fraught with difficulties. Here is a short, non-exhaustive list of the challenges. First, terrorism is a marginal phenomenon in most parts of the world, and there is therefore only a limited amount of data available. Most studies on terrorist recidivism are based on relatively small samples. As a result, small changes in the figures may have significant consequences on the calculated rates. Second, measuring recidivism ideally involves the access to judiciary and/or penitentiary data, which is not always readily available or easily accessible. Some studies have therefore relied on open source data, which may include a series of bias, lacunae, imperfections, or mistakes.
Third, since terrorism is a transnational phenomenon, there are many cases of individuals being convicted for one terrorist offense in one country, then for a different terrorist offense in another country, which are cases of recidivism at the international level.g Such cases are extremely difficult to account for since it requires access to data from several countries. Fourth, various studies use different methodologies. Some account for recidivism broadly defined, others for terrorist recidivism or reengagement, but they do not always clearly state what they are measuring precisely (or how they are measuring it), hence making comparison between studies very difficult.
A final difficulty is the need to collect data over a sufficiently long period of time, to be able to actually measure recidivism, particularly since prison sentences can be relatively long in some countries and recidivism may only occur several years after release.
Terrorist Recidivism Rates: The Data So Far
Most countries do not hold records of recidivism rates, and studies on terrorist recidivism are fairly limited. However, the few studies that exist consistently indicate a very low rate of terrorist recidivism, certainly compared with the average rates of criminal recidivism (generally between 40 and 60 percent, worldwide),26 hence challenging the common perception that terrorists are likely recidivists.
The first estimates of terrorist recidivism rates appeared in early studies that analyzed the profile of jihadi terrorists across a range of criteria. Although the objective of these studies was different and the data collection not systematic, they found low rates of terrorist recidivism in their samples. In 2006, Edwin Bakker found that six out of 242 European jihadis (2.5%) had a prior record for terrorism-related offenses.27 Marc Sageman found no case of terrorist recidivism in his sample of 172 jihadis in 2004.28 Similarly, Frank Bovenkerk found in 2011 that 24 Moluccan terroristsh who had been convicted in the Netherlands in the 1970s and had served long prison sentences, had not relapsed since.29 The latter study, albeit based on anecdotal evidence, was interesting for its longer time frame.
A number of academic studies have recently looked into the issue of terrorist recidivism in a more systematic manner. Omi Hodwitz compiled a dataset of 561 individuals convicted of terrorism-related offenses in the United States between 2001 and 2018.30 Only nine of them recidivated (1.6%), five of whom did so in prison. However, only three cases were linked to terrorism (radicalization of other inmates), thus bringing the actual rate of terrorist recidivism (in the narrow sense) down to 0.5%. Christopher Wright compiled another dataset of 189 individuals who were involved in jihadi plots against targets in the United States between 1990 and 2019.31 Only two of them, he found, were involved in jihadi activities after their release (1%).i Fernando Reinares, Carola Garcia-Calvo, and Alvaro Vicente compiled a dataset of 199 jihadi terrorists convicted or killed in Spain between 2004 and 2018.32 They found that 14 of them recidivated in terrorism (7%), five of which in prison. In their detailed evaluation of a Dutch reintegration program for terrorist convicts, Liesbeth van der Heide and Bart Schuurman found that eight of the 189 individuals who participated in the program between 2012-2018 recidivated in relation to terrorism (4.2%).33
Figures released by some countries fall in similar ranges. Malaysia claims that only 13 terrorist convicts arrested between 2001-2011 “relapsed,” out of 240 individuals (5.4%).34 j In 2013, the Indonesian National Counterterrorism Agency claimed that 25 out of 300 terrorist convicts released from prison recidivated (8.3%).k Saudi Arabia and Pakistan have claimed very low rates of terrorist reoffending for participants in their deradicalization programs, below 3%, although some analysts argue it is likely higher.35
In the United Kingdom, terrorist recidivism figures for 2013-2019 were already mentioned above, at 3%. These figures are in continuity with previous decades. Between 2001 and 2008, there were 196 terrorist-related convictions in England and Wales, and more than 100 convicts had been released by early 2009. According to Andrew Silke, in a book published in 2014, none of them had been rearrested or reconvicted for a terrorist offense.36 In Northern Ireland, 453 paramilitary prisoners were released in 1998 as part of the Good Friday Peace Agreement. By 2011, only 10 of them had been recalled in custody for involvement in terrorist activities (2.2%).37
In the United States, a report from the Office of the Director of National Intelligence (DNI) indicates that as of January 2019, 124 of 729 former Guantanamo detainees had reengaged in terrorist activities (17%), while an additional 100 former detainees (13.7%) were suspected of reengagement.38 This rate of 30.7 percent is significantly higher than other rates discussed above. This could, of course, be linked to the particular profile of Guantanamo detainees. However, it should be clarified here that this rate is absolutely not comparable to other rates discussed above. Indeed, the DNI figures are about cases of reengagement, thus individuals who returned or were suspected to have returned to terrorist activities but who may not have (yet) been prosecuted for it. Moreover, most Guantanamo detainees have never been convicted in the first place, and sometimes the evidence of their involvement in terrorism was thin.39
Two recent studies might appear to contradict the consistently low rates of terrorist recidivism cited above. A study by Israeli researchers concluded on a five-year recidivism rate of 60.2% for 1,557 offenders between 2004-2017, mostly affiliated with Palestinian terrorist groups.40 However, these findings cannot be compared to other studies discussed above. First, it calculates “recidivism” on the basis on (re)incarceration, not (re)conviction, which is, of course, a much lower threshold. Furthermore, the study seems unable to isolate terrorist offenders, who are included in a broader set of “security offenders” and “criminal offenders,” based on data provided by the Israel Prison Service. As such, the study measures reengagement and not recidivism, and it possibly includes non-terrorist offenders and non-terrorist-related offenses.
The second study that might seem to contradict findings on recidivism is a review of 87 autobiographical accounts of terrorists, published between 1912 and 2011.41 Based on their sample, the researchers calculated a rate of reengagement of 61% (52 out of 85 individuals). However, these results are to be treated with caution. First, the authors looked at “re-engagement,” rather than “recidivism,” which is a much broader category as discussed above. Furthermore, the study is based on autobiographies, which are more likely to overrepresent reoffenders’ profiles. Terrorist leaders are more likely to write their memoirs than average members, whereas people with a short terrorist career are “less likely to pen a memoir and re-engage,” as the authors acknowledge themselves. As a result, it can be confidently said that this study does not challenge the findings from other studies.
Finally, to conclude this literature review, it seems appropriate to briefly look into the key findings of the literature on foreign fighters, given that current fears about terrorist recidivism are largely linked to returning foreign fighters from Syria and Iraq. A few studies have looked at the ratio of returning fighters involved in terrorist plots or attacks, which could be seen as one of the worst forms of recidivism or reengagement from a public security point of view. Although there are clear limitations in the ability to collect reliable data from open sources, the “blowback rate,” as it has been called, has been estimated to be certainly well below 10 percent for all jihadi returnees worldwide since the 1980s, and possibly even below one percent.42
To be fair, attack plotting is only one pathway to reengagement that foreign fighters can follow. Foreign fighters can present a threat in different manners. They can be involved in terrorist plots, training, recruitment, or mere logistical support for terrorist networks, at home or abroad.43 Some estimate, for instance, that 10 percent of the foreign fighters who traveled to Syria had a prior experience with jihad, while a majority of foreign fighters in Iraq in the early 2000s were allegedly recruited by veteran fighters.44 Overall, some foreign fighters can therefore remain engaged in terrorism for life and become ‘career terrorists.’ While security services are clearly concerned about this ‘veteran effect,’ the proportion of returning fighters who pursue their engagement over years appears limited,45 which is therefore unlikely to fundamentally challenge findings on recidivism.
A study by David Malet and Rachel Hayes calculated the time-lag between the return of jihadi foreign fighters in Western countries and their involvement in a terrorist plot.46 They compiled a dataset of 230 returnees-turned-plotters between 1980 and 2016, and were able to measure the time-lag for 134 of them. The average lag time was 10 months, but the majority of returnees-turned-plotters either conducted an attack or were arrested within the five months following their return. This would suggest that many foreign fighters returned with the intention of plotting an attack, as it was the case for the attacks in Paris (2015) and Brussels (2016), which cannot really be considered a form of recidivism or reengagement, but rather the continuation of their terrorist engagement. The “blowback rate” of returnees is therefore an imperfect measure for recidivism. The study of Malet and Hayes furthermore highlights that the threat of returning foreign fighters may not only decrease over time, but that it actually decreases very quickly.47
Part Two: Terrorist Recidivism in Belgium: A New Dataset
Given there is a disconnect between findings from the literature on terrorist recidivism and fears over the threat posed by this phenomenon, more research appears necessary. Belgium stands out as an interesting case study. The country has been confronted with Islamist terrorism since the early 1990s, with networks linked to the Algerian GIA or the Moroccan GICM.48 In recent years, Belgium was faced with an unprecedented mobilization for the jihad in Syria and Iraq. Four hundred thirty-three individuals traveled to join jihadi groups in the region,m meaning that Belgium had the highest ratio of jihadi fighters per capita in the European Union.49
Like most countries, Belgium does not maintain metrics on terrorist recidivism.n Therefore, it was necessary to build a dedicated dataset, the “Jihadi Terrorist Offenders in Belgium” (JTOB), based on all convictions for terrorism-related offenses pronounced by Belgian courts. In this regard, it should be noted that terrorist offenses were only included in the Belgian criminal code in 2003. For previous convictions, terrorism-related trials were identified on the basis of the nature of the incriminated offenses.50 All terrorism-related judgments from January 1, 1990,o until the end of 2019 were obtained with the help of the Federal Prosecutor’s Office, which is in charge of terrorism cases, and the Coordination Unit for the Threat Analysis (CUTA), which is Belgium’s counterterrorism fusion center.
Judgments are not readily available in Belgium but can be obtained upon justification, particularly for research purposes. They include a description of the events linked to the offense, as well as some basic information about the offender(s), and the sentencing decision itself. The length of these documents varies from about 10 pages to over 100 pages for more complex cases. Depending on the Court, documents are in French or in Dutch. Judgments from Courts of First Instance, Appeal, and Cassation were obtained. All these documents were read thoroughly and coded in a spreadsheet.
In the context of this project, following the conceptual work introduced by Mary Beth Altier, Emma Boyle, and John Horgan,51 the author decided to look at terrorist recidivism narrowly defined, as well as at terrorist reengagement more broadly. The author defines terrorist recidivism as two convictions for distinct terrorist offenses over distinct periods of time, separated by a clear period of disengagement (commonly imprisonment). Thus, two convictions covering a similar period of terrorist engagement, not clearly separated in time, would not be counted as a form of recidivism. Terrorist reengagement is defined as two periods of engagement in terrorism, separated by a period of voluntary or involuntary disengagement, but not necessarily sanctioned by a judiciary decision.
This distinction enables a more comprehensive overview of the individuals that are considered to be a lasting problem by the Belgian security services. It also addresses some of the most obvious pitfalls of the data on recidivism identified above, as illustrated by the cases of Usman Khan and Sudesh Amman, who were killed rather than convicted again for terrorism and would therefore be categorized as examples of terrorist reengagement rather than terrorist recidivism under the author’s methodology.
The methodology contains some important caveats. The JTOB dataset includes only individuals that were convicted at least once for terrorism in Belgium. As a result, all cases of terrorist recidivism are included. However, cases of Belgian citizens or residents who were involved in terrorist activities, but never convicted in Belgium, are not included, even if they were tried for terrorism-related offenses in another country.p Some cases of international recidivism following two convictions for terrorism, one in Belgium and one in another country, are also likely missing.
Furthermore, since the dataset is based on terrorist convictions primarily, data on reengagement is not exhaustive. It is limited to individuals who were prosecuted at least once for terrorism-related offenses in Belgium, but had engaged in terrorism before or reengaged after their conviction. For instance, some judgments referred to previous terrorist activities (in Belgium or abroad) that had not been formally sanctioned, such as imprisonment in Guantanamo. Individuals that may have reengaged in terrorism over time but were never prosecuted or convicted in Belgium do not appear in the JTOB dataset.
There was also a challenge linked to the fact that many Belgian foreign fighters are still in Syria, possibly including some recidivists. The JTOB data was therefore crossed with the list of frozen assets,52 which basically includes all Belgian foreign fighters still in Syria. This list is established by the Belgian National Security Council, upon recommendations of CUTA; it includes about 280 individuals suspected of terrorist activities. This crossing of data permitted the identification of a few additional cases of reengagement, which will likely become cases of recidivism in the near future, after prosecution in Belgium is completed.q
Finally, it is worth highlighting that this project is clearly a work in progress.r Terrorism-related trials occur almost on a weekly basis in Belgium these days, which means that the data is bound to evolve.s
The JTOB dataset includes 557 individuals, convicted in the context of jihadi terrorism between January 1, 1990, and December 31, 2019—a span of three decades. Compared with the previous studies reviewed above, the author’s JTOB dataset is therefore significantly larger than all but one study (Hodwitz),t and the time span longer than all but one as well (Wright). This makes the JTOB dataset comparatively and statistically robust.
This project was limited to jihadi terrorism. The original dataset included non-jihadi offenders, such as far-right terrorists, but they were eventually removed from the dataset for consistency purposes. They accounted for more than 20 individuals in total.
The JTOB dataset includes information about names, gender, date of birth, indications of previous criminal record,u date of (the start of) offense, date of judgment,v type of offense,w and length of sentence.x For cases of recidivism, information is repeated and the date of prison release was added, when available.
Given the unprecedented mobilization for jihad in Belgium since 2012, the majority of the convictions are unsurprisingly linked to Syria.y However, 110 individuals were convicted for a first terrorism offense that started between 1990 and 2010, thus before the Syrian mobilization.z
The dataset includes 472 men (85%) and 85 women (15%). While this men/women ratio is higher than for most other studies, the influence of the Syrian mobilization is evident: prior to 2011, only four women in the dataset (3.6%) had been convicted of a first terrorist offense compared to 106 men (96.4%).
The average age at the time of the first terrorist offense leading to conviction is 26.8 years.aa Some studies have noted that jihadi militants in Europe, and in Belgium specifically, were younger during the Syrian mobilization compared to previous mobilizations.53 The JTOB dataset confirms this observation. The average age at the time of the first offense that resulted in conviction was 28.7 prior to 2011 and 26.3 after 2011.
When available, the author looked for evidence of prior criminal records. The hypothesis has been formulated in the criminology literature that a longer criminal experience increases the risk of recidivism.54 The hypothesis has also been formulated that criminal activities could lead to jihadi militancy, along the new crime-terror nexus.55 The JTOB includes information on criminal records for 205 individuals. One hundred six of them had a prior criminal record before their first terrorist conviction (51.7%), while 99 did not. The data does not allow to infer for the entire dataset, but this ratio is in line with the statement of Belgium’s Federal Prosecutor who said that half of Belgium’s foreign fighters had a prior criminal experience.56 ab There is no significant difference in the data with regard to prior criminal offenses between the pre-2011 and the post-2011 offenders.
With regard to the types of offense, Figure 3 shows that foreign fighting (61.4%) and financial or logistical support (36.3%) are the most common terrorist offenses among first-time terrorist convicts. However, when looking at the break-down between the pre-2011 and post-2011 offenders, the influence of the Syrian mobilization can clearly be seen, with a jump in foreign fighting. Quite interestingly, the ratio of involvement in terrorist attacks or terrorist plots has remained stable over time, just above two percent—although it should be noted that some individuals were killed or died in Belgium in the context of terrorist attacks since 2015. Still, the overwhelming majority of terrorist convicts were simply not charged with participation in planning or carrying out a terrorist attack.
Finally, with regard to sentences for the first conviction for terrorism, 482 individuals were sentenced to time in jail, 28 to a suspended sentence or community service, and 47 were discharged.ad For the first conviction, the average length of actual sentences was 60.5 months (excluding discharges, suspensions, and community service), while the median was 60 months (five years). In fact, a majority of the individuals in the JTOB dataset spent five years or less in jail for their first conviction, as the effective time in prison is generally shorter than the sentence. They had therefore plenty of opportunities within the time horizon of the data collection to reengage or recidivate.ae It is worth noting that the average sentence has remained relatively stable over time, albeit increasing slightly. It was 55 months for individuals whose first terrorist offense leading to a conviction in Belgium started before 2011, and 61.7 months after 2011.
The Findings on Terrorist Recidivism in Belgium
In the JTOB dataset, the author identified 13 cases of terrorist recidivism (2.3%), which is in line with previous studies discussed above. In addition, the author identified 10 cases of reengagement, and four cases where reengagement was subject to discussion.af Even when including these four questionable cases, the rate of recidivism and reengagement when combined was only 4.8%.
To be complete, it should be added that 13 other individuals were convicted twice for a terrorist offense in two distinct trials, but covering the same context and period of engagement, thus without separation (i.e., disengagement) between the two offenses. This includes mostly individuals who were convicted twice in relation to Syria (e.g., once for participation in a terrorist organization, then for threats against Belgium, often expressed from Syria). Furthermore, 36 individuals were convicted for terrorism while in a situation of reprieve or probation for criminal activities (“legal recidivism”).ag
Among the 27 cases of terrorist recidivism and (suspected) terrorist reengagement, there are four women (14.8%) and 23 men (85.2%), which is exactly the same female-male ratio as in the whole dataset. Eighteen out of the 27 were already involved in terrorist activities before 2011, and 11 of them before 2005, which suggests a continuity of recidivist behavior over time. Two of them were detained in Guantanamo.ah Twenty-four out of the 27 cases (88.9%) had at least one offense linked to Syria, which strongly suggests that Syria offered an unprecedented opportunity for recidivism to previous offenders. Furthermore, seven individuals had two successive convictions linked to Syria, which suggests that the time span of the JTOB dataset is sufficiently large to start to detect such a form of recidivism.
Criminal records information was available for 15 of the 27 terrorist reoffenders, among which eight had prior criminal records and seven did not, which is also almost the same as the ratio in the dataset as a whole. Contrary to some of the hypotheses formulated in the literature, the author cannot therefore draw a connection between criminal experience and risk of terrorist recidivism/reengagement.
With regard to age, the average on the first offense among the 27 reoffenders was 28.9 years, which is higher than the average for the JTOB dataset as a whole (26.8 years). This finding is interesting because it seemingly contradicts a hypothesis formulated in the literature that a younger start in criminal or terrorist activities increases the likelihood of recidivism.57
The time span between the date of release from prison and the beginning of the second terrorist offense or engagement was also calculated. This was not possible for some individuals, as they did not go to prison for their first offense: they were either discharged or they traveled to Syria before their trial or imprisonment (but after a police arrest that marked the end of their first offense). One individual convicted in 2016 of terrorism in Belgium had previously been transferred from Guantanamo to Algeria in November 2008 and left to Syria in early 2013, but no precise date for his release could be identified as his situation in Algeria was unclear.ai For the 17 individuals for which full information was available, the average time-lag between prison release after the first terrorist offense and the beginning of the second terrorist offense was 23 months, and the median was nine months. In fact, the majority reoffended within nine months, and only three individuals recidivated or reengaged after more than three years. The precise dispersion can be seen in Figure 4.aj
While conclusions should be drawn carefully in light of the small size of the sample, this could suggest at least two things. Firstly, the first months after release see the higher risk of reoffending, contrary to a previous assumption in the literature that potential recidivists need years to show their ‘true colors.’58 Secondly, Syria provided a powerful pull for recidivism, as most reoffenders had at least one offense linked to the Syrian mobilization. Given the age of some of these reoffenders, above 45 or even 50 years old at the time of reoffending, and the relatively long time-lag between offenses for some reoffenders, it can be asked whether some of these individuals would have recidivated if it was not for the Syrian jihad.
Finally, with regard to the type of offenses, there are some slight differences between the 27 reoffenders and the full dataset. The ratio of individuals involved in a terrorist plot as a first offense is higher for reoffenders than in the dataset as a whole (7.4% vs 2.3%). This is also true for individuals involved in financing or logistical support (44.4% vs 36.3%). The ratio for the other types of first offenses, however, is very similar to the entire dataset (59.2% vs 61.4% FTF; 14.8% vs 15% Recruitment/Leadership; 14.8% vs 12.4% Propaganda/Apology).
In their second offense, reoffenders were much more likely to be involved in FTF activities (21 out of 27; 77.8%), than in their first offense, which is unsurprising since most of them were reoffending in connection to Syria. There was also a certain consistency of behavior for some, as 14 out of the 16 individuals engaged in FTF activities in their first period of engagement reoffended as FTF a second time. Furthermore, six individuals were involved in recruitment or leadership activities in their second offense (22.2%), four of which were former FTFs, which could reinforce the idea suggested above that veteran jihadis can become key recruiters or “entrepreneurs” for future networks.59 In contrast, no reoffender was involved in a terrorist plot for their second offense, which is in itself a significant finding.ak Six individuals were convicted for financing or logistical activities for their follow-up offense, and three for propaganda or apology.
The findings from this literature review and from the analysis of the JTOB dataset suggest that the threat of terrorist recidivism and reengagement is limited. Although every case of terrorist reoffending is one case too many, the fear of recidivism appears to be disproportionate compared to its actual occurrence. It is interesting to note that in existing studies, low recidivism rates were encountered across different forms of terrorism and in different contexts. This would suggest that low recidivism rates are not strongly correlated with specific sentencing measures, counterterrorism regimes, or political contexts.
A predictable critique of these findings is that recidivism may need more time to be observed or that major cases of reoffending will remain unaccounted for, including notably: those with prior terrorist convictions killed carrying out terrorist attacks, foreign fighters taking the fight to other conflict zones, or recruiters remaining within the boundary of the legal system. This is indeed a valid point. The JTOB sought to address this partly by including cases of reengagement, beyond a narrow definition of recidivism, but a number of terrorist reoffenders are most certainly missing. For instance, a key figure such as Oussama Atar does not appear in the JTOB dataset because he was never convicted in Belgium, although he is a prominent case of terrorist reengagement. Atar, who French intelligence services reportedly presume to be dead,60 has been presented as a leading member of the Islamic State in Syria and the mastermind of the attacks in Paris (2015) and Brussels (2016).61 In 2005, he had been arrested and detained in Iraq, notably in the camps of Abu Ghraib and Camp Bucca.62 He came back to Belgium in 2012 and left for Syria in December 2013.63
While acknowledging some missing cases, it is far from clear whether this has a significant impact on the rate of reengagement (and by definition, it does not influence the rate of recidivism, in any case). Indeed, there seems to be only a limited number of individuals known or suspected of reengagement by the Belgian authorities who do not appear in the JTOB dataset.am Moreover, traditional statistics on criminal recidivism are subject to similar flaws. (Similarly to terrorists, criminals can escape abroad, remain undetected, or die before prosecution).
In fact, one of the key reasons explaining low rates of terrorist recidivism could very simply be methodological. Terrorist recidivism calculates only a very narrow form of recidivism into a same type of offense (terrorism), which is itself a marginal phenomenon, whereas ordinary recidivism calculates recidivism into any type of offense. As a result, it is not surprising to encounter lower rates. As discussed above, when looking at the criminal records of terrorist offenders, in a broader understanding of terrorist recidivism, there is around the same rate as for ordinary criminals, with around 50 percent of jihadis convicted in Belgium having a prior criminal conviction.
Two recent evaluations from the Belgian counterterrorism fusion center (CUTA) further reinforce the findings of this article. They concluded that 84% of the male returnees from Syria (37/44) and 95% of women returnees (19/20),an and 73% of the “failed men and women travelers” (47/64)ao have been showing signs of disengagement.64 These figures only refer to individuals who were free, after serving time in prison or not, and thus exclude individuals still in prison or detained abroad. These evaluations, together with the literature on terrorist recidivism, tend to suggest that most terrorist convicts simply do not seek to return to their ‘old habits,’ contrary to the dominant perception.
In contrast, the two CUTA evaluations found that a minority of released offenders remain of high concern. This suggests that a small number of die-hards will remain active across successive waves of jihadi militancy, and remain a key concern for police and intelligence services. To some extent, the low rates of recidivism found by this author may be partly a sign of the success of Belgian security services. Yet, the gap between Belgian (and other Western security services’) (over)representation of the threat of recidivism as “surging,” as expressed by intelligence and police analysts,65 on the one hand, and the findings of this study, on the other hand, is puzzling and worth reflecting upon. It is conceivable that security services may be overly influenced by the magnifying effect of a few eye-catching, problematic cases, which take up a lot of their resources. As a result, these services might overlook the fact that most convicts simply leave terrorism behind after prison.
To conclude, it is interesting to note how much the fear of recidivism is part of the (post-)penal process, while so little is known about it. For instance, in the United States, the likelihood of reoffending has long been factored into judges’ sentencing decisions.66 Similarly, in Belgium, the risk of recidivism is often part of the motivation of a judge’s sentencing decision.ap In the United Kingdom as well, preventing recidivism is at the core of the management of terrorist cases.67 Yet, it is far from clear how judges (or probation services, later) assess the risk of recidivism of individuals. Similarly, the reduction of recidivism is one of the core goals of rehabilitation (disengagement or deradicalization) programs, and often the main or only criterion for evaluating success.68 However, in light of the low rates of terrorist recidivism identified in this article, it can be questioned whether this is indeed the right criteria for evaluation. Although it is not the aim of this article to criticize these rehabilitation programs, their ambition and added value should perhaps be reconsidered if terrorists are found to disengage on their own.
As a wave of terrorist offenders are about to be released from prison in the coming months, the conclusions of this article should be pondered carefully by policy makers and counterterrorism practitioners. The take-away should not be to discard the threat of recidivism altogether. The monitoring of released terrorist offenders and preventing their relapse is a clear and necessary task of security services, as even a small number of recidivists can still constitute a most serious threat in the short- to longer-term. However, this article clearly supports the need for evidence-based policies, aiming to increase public security, while mitigating the potentially counterproductive effects of indiscriminate, fear-based responses. As to scholars, more research should be conducted on this issue, notably with a focus on the causes of (the low degree of) recidivism. CTC
* The section of this article subtitled “Data Overview” was updated shortly after publication to provide added technical specificity on some of the findings.
Thomas Renard is a Senior Research Fellow at the Egmont Institute and Adjunct Professor at the Vesalius College in Brussels. The author thanks Cassandra Bockstael for her assistance in the coding of recidivism data and is also very grateful to the Belgian Coordination Unit for the Threat Analysis (CUTA), which provided tremendous support to this project, and to the Federal Prosecutor’s Office for granting access to terrorism-related judgments. Finally, the author thanks his colleague Rik Coolsaet and Andrew Silke, for their helpful comments on early drafts. Follow @tom_renard
[a] The RAN was established in 2011 and is funded by the European Commission. It brings together over 2,400 practitioners from across Europe on a regular basis to discuss specific issues and share good practices. There is notably a working group on Prison and Probation and another one on Exit programs.
[b] HLCEG-R was established in 2017 and convened relevant representatives from the E.U. institutions and E.U. member states.
[c] A recent review of 87 autobiographies of terrorists, published between 1912 and 2011, covering 70 different terrorist organizations, identified 52 individuals who reoffended multiple times. See Mary Beth Altier, Emma L. Boyle, and John G. Horgan, “Returning to the Fight: An Empirical Analysis of Terrorist Reengagement and Recidivism,” Terrorism and Political Violence, online November 18, 2019.
[d] This includes notably about 1,700 in France, 700 in the United Kingdom, over 700 in Spain and Italy, over 100 in Scandinavia, and over 200 in Belgium. This estimate is based on the author’s own compilation, relying on open sources and interviews with analysts from all relevant countries, in 2018-2019. A forthcoming publication by Rajan Basra and Peter Neumann of the ICSR in London has more figures on this.
[e] To be fair, the evidence is sometimes missing due to a paucity of rigorous academic research in these topics, providing empirical or quantitative evidence on the risk of recidivism, as highlighted in this article.
[f] Reengagement rates can include individuals who were convicted only once for terrorism-related offenses, but also individuals who were never convicted. There are some jihadi militants in Europe whose names appear in a number of cases, but have never been convicted or prosecuted.
[g] To the author’s knowledge, there is no study reviewing cases of international terrorists convicted in different countries. However, the author has come across several cases, just in Belgium. One such case is Farid Melouk, convicted in France in 1998 (in absentia) to seven years in prison for recruitment, and convicted to nine years in Belgium in 1999. Another case is Abdelkader Hakimi, convicted twice in Belgium for terrorism (in 2007 and 2018), and previously convicted to death in Morocco in 1985, for his participation to a youth jihadi movement.
[h] Moluccans refer to the indigeneous people from Maluku Islands (currently part of Indonesia). In the 1970s, some Moluccans living in the Netherlands, the former colonial occupying power, were responsible for a spate of attacks.
[i] Christopher Wright calculates a recidivism rate of 6.5%, on the basis of 31 released terrorist offenders. However, since reoffending is possible in prison, the rate of recidivism compared with Wright’s entire dataset is presented here. This is also more in line with the methodology of similar studies (Hodwitz, Reinares et al.) and with this study.
[j] In a number of studies or figures shared by governments, it is not entirely clear what is covered by recidivism figures. In the case of Malaysia, it is not entirely clear what is meant by “relapsing” and how it is calculated, which could be, for instance, based on intelligence information, re-arrest, or re-conviction.
[k] The rate of terrorist recidivism in Indonesia might be as high as 15%, according to researchers who counted additional cases of recidivism between 2013-2016. However, these researchers seem to use the same figure (300 in 2013) for released offenders as the Indonesian National Counterterrorism Agency did in calculating the rate, as if no terrorist offender were released between 2013 and 2016, which raises important methodological questions. See Noor Huda Ismail and Susan Tim, “From prison to carnage in Jakarta: Predicting terrorist recidivism in Indonesia’s prisons (Part 2),” Brookings, January 28, 2016.
[l] While this table brings together a number of studies on terrorist recidivism, it should be acknowledged that these studies are not entirely comparable for they use different methodologies, in terms of selection criteria, timespan, or definition of recidivism. For instance, Silke and Hodwitz use strictly two convictions as a criteria for recidivism, whereas Wright also include terrorist convicts or jihadi plotters who subsequently died in a terrorist attack, even if not convicted (of which he finds only one case). As discussed above, what is actually covered by official figures released by some governments (here Indonesia and Malaysia) is often unclear as well. As a result of these caveats, the ‘average’ presented in this table is purely indicative, and should be read with caution.
[m] Information provided by the Coordination Unit for the Threat Analysis (CUTA), as of December 2019.
[n] This project started with a question that the author raised to the relevant services in Belgium, asking for rates of terrorist recidivism. While indicating that such information was not readily available to them, they expressed a keen interest to investigate this matter further.
[o] The first recorded trial in the dataset is the appeal decision of September 1995, regarding members of the GIA arrested in March 1994 and first sentenced in October 1994.
[p] Some such cases are well known from open sources, but given the impossibility to collect such information in a systematic manner, the author decided to ignore it altogether. However, for individuals subject to at least one conviction in Belgium, information for reengagement was included when available, including terrorism-related convictions abroad.
[q] In Belgium, trials can be conducted in absentia, and all foreign fighters still in Syria (even if presumed dead) have been or will be prosecuted.
[r] Several meetings were organized with CUTA at various stages of the research, between October 2019 and March 2020, to limit the risk of missing cases in the dataset or to discuss specific cases for which the coding was not straightforward.
[s] As of December 2019, about 160 individuals in the JTOB dataset were still believed to be in Syria, while about 60 were in prison. However, it should be noted here that recidivism, or at least reengagement, is possible for most of these individuals. This is certainly the case for those in prison, as illustrated by some of the terrorist attacks that have taken place in France and the United Kingdom in recent years.
[t] The DNI figures and Israeli study also have larger samples, but as discussed, they are not comparable in their methodology.
[u] Judgments sometimes refer to the criminal records of the offender, although not systematically. This is therefore a category of the dataset for which there is no information for a number of offenders.
[v] In cases of appeal, the coders used the date of the judgment on appeal.
[w] The coders distinguished between five types of offense: recruitment or leadership of a terrorist network; terrorist attack or attempt; financing or logistical support to terrorism; propaganda or apology of terrorism; foreign fighting (or attempted foreign fighting). These offenses do not correspond to specific articles in the penal code, which means that judgment calls had to be made during the coding process. However, these categories correspond to clearly distinct terrorist behaviors.
[x] It is the length of sentence in the final judgment (e.g., appeal) that is coded.
[y] This includes foreign fighters, traveling or attempting to travel from Belgium to join a jihadi group in Syria, but also individuals convicted for recruitment, plotting an attack, financing, or propaganda in connection to a jihadi group active in Syria.
[z] A small number of individuals among these 110 were convicted in relation to the Syrian mobilization (post-2011) as members of recruitment networks notably, but their engagement in militant jihadi activities started before 2010, however, and thus in a different context.
[aa] Age was calculated on the basis of the beginning of the terrorist offense, as indicated in judgments. As a result, the true engagement with terrorism might be longer in a number of cases.
[ab] Similar ratios have been observed in France, Germany, the Netherlands, and the United Kingdom, notably, according to Basra and Neumann. See Rajan Basra and Peter R. Neumann, “Crime as Jihad: Developments in the Crime-Terror Nexus in Europe,” CTC Sentinel 10:9 (2017).
[ac] The sum of each column in Figure 3 is higher than 100% because individuals can be convicted for different types of offenses.
[ad] The author made the decision to maintain individuals that were discharged in the dataset, on the assumption (sometimes stated in the judgment or proven by later evidence) that this does not mean that these individuals had not engaged in terrorism-related activities. In any case, removing these cases from the dataset did not result in significant statistical changes.
[ae] See footnote S.
[af] Some cases of reengagement were uncertain because information available from judgments and from open sources could not confirm with certainty whether these individuals had really reengaged or not. These cases include, for instance, an individual who traveled first to Syria as a minor with her family (and thus not clearly on her own volition) and then again as an adult; and an individual who is suspected to have joined al-Shabaab in Somalia before the Islamic State (inside Syria), but evidence of this prior engagement was not confirmed by the Court.
[ag] This included, for instance, individuals convicted for terrorism who had been convicted previously for violence or traffic misdemeanors.
[ah] It should be noted here that one of these two individuals is possibly the most debatable case of reengagement, as the person was reconvicted in context of a burglary, with only indirect connections with terrorism.
[ai] He was prosecuted in Algeria and reportedly convicted for terrorism, but no information on his detention and release was found. See “Quatre ex-détenus de Guantanamo prochainement jugés devant le tribunal criminel d’Alger,” Associated Press, November 30, 2011.
[aj] If a date for the end of the first offense for the remaining offenders is approximated using the date of the first trial, the average time-lag until the beginning of the second terrorist offense, as stated by the judge in his/her judgment, or second engagement, for the entire group of 27 reoffenders becomes 22.5 months.
[ak] Christopher Wright came to a similar finding in his dataset of jihadi plotters in the United States, noting that in his “dataset, there is not a single individual in the United States who was jailed in a case related to jihadi plotting, was released, and then became involved in jihadi attack plotting again.” Christopher Wright, “An Examination of Jihadi Recidivism Rates in the United States,” CTC Sentinel 12:10 (2019).
[al] Time-lag was calculated automatically as the lapse between the date of effective release from prison and the beginning of the second offense, as stated by the judge in his/her judgment, or second engagement. As such, the value ‘zero’ indicates that the judge considered that the individual reengaged in terrorist activities on the day of his release from prison.
[am] This is based on known cases from open sources and from conversations with analysts at CUTA.
[an] In a previous version of this evaluation (2018), this figure was lower, at 75%, pointing to the positive evolution of most returnees after their release over a longer period of time.
[ao] For ‘failed travelers,’ the disengagement rate of men and women is similar, according to the evaluation.
[ap] In most judgments reviewed by the author for the JTOB, there is the following standard motivation: “Only a very severe and dissuasive prison sentence, commensurate to the offence, will ensure the objective of the legal proceedings and, as much as possible, will limit the risk of recidivism.” [emphasis added]
 “La radicalisation dans les prisons ne faiblit pas,” Annual Report of the State Security (VSSE) 2017-2018, Belgium, November 2018, p. 17.
 “Council Conclusions on preventing and combating radicalisation in prisons and on dealing with terrorist and violent extremist offenders after release,” conclusions of the Council of the European Union, June 6, 2019.
 Vladimir Voronkov emphasized this point in his introductory remarks and conclusions to the Joint Regional High-level Conference on “Foreign Terrorist Fighters – Addressing Current Challenges,” convened by the OSCE, UNOCT, and Switzerland, in cooperation with the Albanian OSCE Chairmanship (February 11-12, 2020, Vienna).
 “Twenty-third report of the Analytical Support and Sanctions Monitoring Team submitted pursuant to resolution 2368 (2017) concerning ISIL (Da’esh), Al-Qaida and associated individuals and entities,” United Nations Security Council, S/2019/50, January 15, 2019, p. 14.
 Alain Grignard was interviewed by this publication in 2015. See Paul Cruickshank, “A View from the CT Foxhole: An Interview with Alain Grignard, Brussels Federal Police,” CTC Sentinel 8:8 (2015).
 Andrew Silke, “Physical facilitating environments – prisons and madrassas: mechanisms and vehicles of violent radicalization,” in Anthony Richards et al. eds., Jihadist Terror: New Threats, New Responses (London: IB Tauris, 2019). See also, among others, Mark Hamm, The Spectacular Few: Prisoner Radicalization and Terrorism in the post-9/11 Era (New York: NYU Press, 2013); Clarke Jones, “Are prisons really schools for terrorism? Challenging rhetoric on prisoner radicalization,” Punishment & Society 16 (2014), pp. 74-103; Andrew Silke and Tinka Veldhuis, “Countering violent extremism in prisons: A review of key recent research and critical research gaps,” Perspectives on Terrorism 11:5 (2017), pp. 2-11; Thomas Renard, “Countering radicalisation in prison: The case of Belgium,” (London: ICSR, forthcoming).
 “Terrorism: Prisoners’ release: written question – HL782, Answer by Lord Keen of Elie,” U.K. House of Lords, February 11, 2020.
 Rajan Basra, Peter R. Neumann, and Claudia Brunner, Criminal Pasts, Terrorist Futures: European Jihadists and the New Crime-Terror Nexus (London: ICSR, 2016); “Security Council Expresses Concern over Links between International Terrorism, Organized Crime, Unanimously Adopting Resolution 2482 (2019),” UNSC Meetings Coverage, July 19, 2019.
 Mary Beth Altier, Emma L. Boyle, and John G. Horgan, “Returning to the Fight: An Empirical Analysis of Terrorist Reengagement and Recidivism,” Terrorism and Political Violence, online November 18, 2019.
 Marc Sageman, Understanding Terrorist Networks (Philadelphia: University of Pennsylvania Press, 2004).
 Andrew Silke, “Risk assessment of terrorist and extremist prisoners,” in Andrew Silke ed., Prisons, terrorism and extremism: Critical issues in management, radicalisation and reform (London: Routledge, 2014), pp. 108-121.
 “Summary of the Reengagement of Detainees Formerly Held at Guantanamo Bay, Cuba,” Office of the U.S. Director of National Intelligence, June 3, 2019.
 Mark Denbeaux, Joshua Denbeaux, and John Walter Gregorek, “Report on Guantanamo Detainees: A Profile of 517 Detainees Through Analysis of Department of Defense Data,” Seton Hall Public Law Research Paper 46 (February 2006); Scott Shane and Benjamin Weiser, “Judging Detainees’ Risk, Often With Flawed Evidence,” New York Times, April 24, 2011; Margot Williams, “At Guantánamo Bay, torture apologists take refuge in empty code words and euphemisms,” Intercept, January 29, 2020.
 Badi Hasisi, Tomer Carmel, David Weisburd, and Michael Wolfowicz, “Crime and terror: Examining criminal risk factors for terrorist recidivism,” Journal of Quantitative Criminology, online June 26, 2019.
 Thomas Hegghammer, “Should I stay or should I go? Explaining variation in Western Jihadists choice between domestic and foreign fighting,” American Political Science Review 107:1 (2013); Thomas Hegghammer and Petter Nesser, “Assessing the Islamic State’s commitment to attacking the West,” Perspectives on Terrorism 9:4 (2015); Daniel Byman, Road Warriors: Foreign Fighters in the Armies of Jihad (Oxford: OUP, 2019).
 Daniel Milton, “Lessons from September 11 about the post-conflict threat posed by foreign terrorist fighters,” Journal of Policing, Intelligence and Counter Terrorism, online February 6, 2020; Petter Nesser, Islamist Terrorism in Europe: A History (Oxford: OUP, 2015); Byman.
 Clinton Watts, “Foreign fighters: how are they being recruited? Two imperfect recruitment models,” Small Wars Journal, June 22, 2008; Brian Dodwell, Daniel Milton, and Don Rassler, The Caliphate’s Global Workforce: An Inside Look at the Islamic State’s Foreign Fighter Paper Trail (West Point, NY: Combating Terrorism Center, 2016).
 Alain Grignard, “The Islamist networks in Belgium,” in Rik Coolsaet ed., Jihadi Terrorism and the Radicalisation Challenge in Europe (Aldershot, England: Ashgate, 2008), pp. 85-93.
 Thomas Renard and Rik Coolsaet, “From the Kingdom to the Caliphate and Back: Returnees in Belgium,” in Thomas Renard and Rik Coolsaet eds. Returnees: Who Are They, Why Are They (Not) Coming Back and How Should We Deal with Them? Assessing Policies on Returning Foreign Terrorist Fighters in Belgium, Germany and the Netherlands, Egmont Paper 101 (Brussels: Egmont Institute, February 2018), pp. 19-40.
 The reference book on jihadi terrorism in Belgium, by Paul Ponsaers, was also useful in this regard. See Paul Ponsaers, Jihadi’s in België: de route naar Zaventem en Maalbeek (Antwerp: Maklu, 2017).
 “Liste nationale consolidée des personnes et entités dont les avoirs ou les ressources économiques sont gelées dans le cadre de la lutte contre le financement du terrorisme,” Belgian Ministry of Finances, November 8, 2019.
 Paul Gendreau, Tracy Little, and Claire Goggin, “A meta-analysis of the predictors of adult offender recidivism: what works!” Criminology 34:4 (1996): pp. 575-608; Mark Berg and Beth Huebner, “Reentry and the Ties that Bind: An Examination of Social Ties, Employment, and Recidivism,” Justice Quarterly 28:2 (2011): pp. 382-410.
 Altier, Boyle, and Horgan; Basra, Neumann, and Brunner.
 See, for instance, Hasisi, Carmel, Weisburd, and Wolfowicz.
 Matthieu Suc, Les Espions de la Terreur (Paris: HarperCollins, 2018).
 Author interview, CUTA official, Brussels, October 2018; author correspondence, CUTA official, November 2019.
 See, for example, “La radicalisation dans les prisons ne faiblit pas.”
 Kelly Berkell, “Risk reduction in terrorism cases: sentencing and the post-conviction environment,” Journal for Deradicalization 13 (Winter 2017-18): pp. 276-341.