Violent Extremism and Offending: Forensic Risk, Threat, and Psychological Evidence

4 days ago

19 min read

Author: Ukrainian Psychological Hub · Published: October 11, 2026 · Editorial Policy

Ukrainian Psychological Hub | Published October 11, 2026

Violent extremism and offending concern the transition from an ideological, political, religious, or social grievance to unlawful violence or other criminal conduct undertaken in support of a cause. Forensic psychology examines that transition through documented behavior, situational opportunities, social relationships, threat evidence, and the limits of psychological assessment. The central question is how a particular person became involved in violence, what evidence bears on a current threat, and what conclusions the available research can support in an individual legal case.

The most important finding is that violent extremists have no established universal psychological profile. Research identifies recurrent contexts and processes, but these do not function as a checklist proving future dangerousness. A 2025 Campbell systematic review found that studies of specialized extremism-risk instruments had not yet established prospective predictive validity. Threat-management decisions therefore require independent evidence, attention to uncertainty, and meaningful safeguards against treating ideology, identity, diagnosis, or a risk score as proof of crime.

What is violent extremism in psychology and law?

Violent extremism is a research and policy category covering the use, preparation, or support of violence motivated at least partly by ideological goals. Its boundaries differ across studies and justice systems. Some researchers study terrorist attacks; others include hate-motivated violence, violent extremist organization participation, attempted attacks, material support, or criminal acts embedded in an extremist social network. Those outcomes differ in severity, intention, and legal definition, so findings about one cannot be silently generalized to the others.

Extremist belief, radical political opposition, nonviolent activism, and criminal offending are different phenomena. Radicalization is often used to describe movement toward increasingly extreme beliefs or practices, whereas mobilization to violence denotes observable conduct associated with preparation or attempted action. Many people hold radical or socially rejected beliefs without harming anyone. Forensic inquiry focuses on behavior and legally material facts, rather than treating ideological intensity as a clinical symptom. The U.S. National Institute of Justice research overview emphasizes how difficult it is to explain the varied pathways to terrorist involvement.

The legal definition is jurisdiction-specific

In the United States, 18 U.S.C. § 2331 defines domestic terrorism for specified federal statutory purposes by reference to acts dangerous to human life that violate criminal law, apparent coercive or intimidatory purposes, and primarily domestic occurrence. The existence of a statutory definition does not make every politically motivated offense a terrorism conviction or create a universal psychology definition. Other federal statutes, state criminal laws, evidentiary rules, and sentencing provisions govern particular charges and findings. A court determines legal culpability; an evaluator can explain behavior and mental functioning within a defined psycholegal referral question.

U.S. constitutional protections also matter. The Department of Justice explicitly distinguishes criminal conduct from beliefs, associations, peaceful protest, and protected expression. European and other legal systems use different classifications, restrictions, and public-safety frameworks. Comparative research must state where a definition came from, which acts were counted, and whether an outcome was an arrest, allegation, charge, conviction, completed attack, or subsequent offense.

What does the psychological evidence actually explain?

The evidence supports a multilevel account in which personal history, identity, grievances, social relationships, ideological narratives, opportunity, and immediate circumstances may interact. None is a necessary or sufficient cause across cases. Long-term involvement in an extremist milieu, movement toward violence, attack preparation, and repeat offending require different analyses. An explanation that fits a completed offender retrospectively may do little to identify, among thousands of nonviolent individuals, who will commit a future attack.

Grievance, identity, and meaning

Perceived injustice, humiliation, exclusion, moral outrage, or a loss of status can become meaningful within narratives that identify an enemy and justify coercion. The psychological relevance lies in how grievances are interpreted and linked to targets and methods. Grievance is widespread in society and has low specificity for violent offending. A forensic opinion must establish the particular person's words, conduct, relationships, and temporal sequence before claiming that a grievance contributed to a criminal act. The NIJ research synthesis describes identity, ideology, social networks, and group dynamics as recurring research domains.

Social networks and group processes

Participation may develop through friendships, romantic relationships, family ties, mentors, recruitment contacts, or online communities. A group can provide belonging and rewards for conformity, while narratives normalize threats or portray violence as obligation. The mechanism is neither automatic nor uniform: networks can also discourage escalation, offer exits, and sustain nonviolent dissent. The case-relevant question is which relationships facilitated specific choices, resources, and opportunities, with documented dates and corroboration.

Moral justification and the normalization of harm

Perpetrators sometimes frame violence as defense, revenge, sacred duty, or an exceptional means of achieving an alleged higher good. Researchers study these accounts through moral disengagement, threat perception, selective empathy, and justification narratives. Such mechanisms can describe how a person reconciled violence with a self-image, but self-reported motives may also be strategic, retrospective, inconsistent, or shaped by interrogation and publicity. Narrative reconstruction must distinguish contemporary evidence from explanations offered after arrest.

Opportunity, capability, and changing circumstances

Even strongly held violent intentions may not lead to offending without access, opportunity, operational knowledge, resources, and an opportunity to act. Conversely, an acute change in access or preparation can make a threat more immediate despite a stable ideological position. This is why threat assessment attends to behavior and circumstances across time, including credible target selection, concrete preparations, and escalation or interruption, while preserving the distinction between observation and a proven criminal element.

Risk and protective factors: what studies can establish

A risk factor is a characteristic or circumstance statistically associated with an outcome in a studied population. A protective factor is associated with reduced likelihood, interrupted pathways, or better outcomes under particular conditions. Neither establishes causation by itself. Factors can be static, such as documented prior offending, or dynamic, such as current threatening conduct, increased access, destabilizing crises, or supportive engagement. Some factors are context-specific and others appear across violence research.

The NIJ-supported case-control study by Chermak and colleagues compared violent extremist offenders, nonviolent extremist offenders, nonoffending extremists, and other violent offenders. Such designs are particularly useful because they test whether a suspected characteristic distinguishes people who offended from those with similar beliefs who did not. Interpretation still depends on how samples were recruited, what counts as extremism, differences in exposure opportunity, measurement quality, and which confounders were controlled.

Factors that may warrant closer contextual assessment

Prior violent behavior, documented threats, identification with perpetrators, relationships facilitating attack preparation, explicit endorsement of violence in a particular context, acquisition of means, planning, and recent observable mobilization can be relevant. Their meaning depends on timing, authenticity, intended target, role in a group, baseline behavior, and alternative explanations. An online statement alone may be satire, fiction, rhetoric, quotation, bravado, or a meaningful threat; context and corroboration determine the weight of evidence.

Loss, trauma exposure, social isolation, unemployment, perceived discrimination, or identity uncertainty may appear in individual histories, but they are common among nonoffenders and are unsuitable as stand-alone indicators of extremist violence. The NIJ discussion of trauma and mental health stresses variability across cases and the small minority of exposed people who become involved in violent extremism. Psychological assessment should avoid converting population vulnerabilities into personal accusations.

Protective contexts and interruption

Reliable supportive relationships, opportunities for lawful participation, disengagement from violent associates, practical stability, an ability to question violent narratives, constructive conflict management, and access to appropriate support may facilitate safer trajectories. A factor that looks protective at one stage may have little effect once a person has reached an advanced operational stage, and evidence for causal prevention remains limited. Assessors should identify concrete protective conditions, their durability, and whether they are accessible to the individual.

The difference between prediction and explanation

An association observed among known offenders is not an individual prediction. If a study samples only convicted perpetrators, it may identify frequent historical experiences while missing how often the same experiences occur among millions of people who never offend. Selection bias, retrospective reconstruction, inconsistent definitions, and unequal surveillance further complicate generalization. A forensic report should specify the comparison population and outcome period, not silently convert a historical pattern into a probability.

Mental illness, personality, and violent extremism

Violent extremism is not a psychiatric diagnosis in DSM or ICD classification, and a clinical diagnosis does not establish violent intent, ideological adherence, criminal responsibility, or future offending. Psychiatric symptoms, substance use, personality traits, trauma, and cognitive difficulties can affect an individual case, but their contribution must be demonstrated rather than assumed. An assessment of mental state at an alleged offense also differs from assessing present decision-making capacity or future threat.

A 2022 Campbell systematic review and meta-analysis synthesized mental-health findings from terrorist samples. It estimated lifetime diagnosed mental disorder prevalence of 17.4% across 18 studies, but the authors found high risk of bias and did not support the broad claim that terrorist samples are characterized by mental-health difficulties above population expectations. The review could not produce a single pooled comparative risk effect because comparator groups differed substantially. These results are not a diagnostic statement about any specific person.

Claims that psychosis, autism, personality disorder, or trauma inevitably lead to extremist violence are scientifically unsound. If mental functioning is legally relevant, a qualified clinician examines contemporaneous symptoms, functional consequences, collateral evidence, and plausible alternatives. The APA Specialty Guidelines for Forensic Psychology emphasize impartiality, appropriate methods, competing hypotheses, and corroboration. The evaluator's opinion addresses psychological functioning; a court applies the relevant legal standard.

Online radicalization and digital evidence

Online platforms can facilitate ideological exposure, recruitment, social reinforcement, identity experimentation, contacts with organizers, and the circulation of justifications for violence. They can also supply evidence of changed communication and participation over time. A peer-reviewed review of online extremism describes the internet as an environment used for socialization, communication, and organizing, while the evidence leaves major questions about direct causality and differences between online involvement and physical-world violence.

Exposure to extremist content is not equivalent to endorsement; endorsement is not equivalent to intention; and intention is not equivalent to preparation or a completed offense. Algorithms, offline peers, existing motivation, local conflict, and selective searching can all shape what a researcher observes. Digital traces are vulnerable to missing context, shared accounts, impersonation, manipulated screenshots, platform changes, and incomplete retention. A forensic interpretation should document how material was collected and authenticated and avoid overstating what passive viewing can show.

Digital evidence can be relevant when it establishes threats, concrete planning, communications, chronology, or links to actions supported by other records. Researchers also distinguish being radicalized entirely online from using the internet as one part of a wider social trajectory. A claim that a specific platform caused violence requires a stronger design than observing that an offender used that platform. The broader scientific question about digital persuasion belongs to social and media psychology; this article focuses on legally relevant offending and threat evidence.

Threat assessment versus forensic risk assessment

Threat assessment is typically a problem-solving process concerned with whether an identifiable situation is moving toward violence and which proportionate interventions could reduce danger. Postconviction risk assessment often addresses supervision, management, rehabilitation, or release decisions. The referral question and available evidence differ: a community concern may arise before any charge, whereas correctional evaluators often have established legal records and a defined population. Neither process establishes guilt.

Observable mobilization and behavioral evidence

The 2025 U.S. Violent Extremism Mobilization Indicators resource from the FBI and partner agencies directs readers to observable behaviors and situational context. It notes that some people planning violence issue direct or indirect threats, whereas others conceal their plans. It also explicitly protects constitutionally safeguarded activities and legally protected characteristics from being treated as a sole basis for action. Indicators are investigative aids; they are not diagnoses, proof of intention, or validated stand-alone probability estimates.

A credible assessment integrates time-ordered and corroborated accounts, recent changes from baseline, direct communications, access and preparation, known barriers, protective contacts, and competing explanations. The analyst distinguishes verified acts from secondhand reports and separates imminent actionable concerns from chronic or nonspecific vulnerabilities. Where available evidence points to a specific, immediate threat, appropriate institutional safety protocols take precedence over speculative profiling. Interpretation must remain within the professional's role and applicable law.

Behavioral threat management

Threat management asks what could reasonably reduce a developing risk: resolving access and safety issues, preserving lawful evidence, coordinating with appropriate services, addressing acute crises, or increasing proportionate supervision when justified by an authorized decision maker. Decisions are revisited as conditions change. Effective management depends on process quality, communication, documentation, and rights safeguards as well as psychological understanding; declaring somebody dangerous is not itself an intervention.

Risk tools: VERA-2R, ERG22+, TRAP-18, and related approaches

Structured professional judgment frameworks organize what an assessor should examine, prompt consideration of aggravating and protective conditions, and support a transparent rationale. Examples used in research or practice include VERA, ERG22+, TRAP-18, MLG-V2, and other locally applicable systems. Their contexts and intended tasks differ. A tool selected for people convicted of extremist offenses should not automatically be applied to an uncharged school student, an unrelated ideological community, or a population for which it lacks validation.

What the systematic review actually found

The 2025 systematic review by Brouillette-Alarie and colleagues identified 20 eligible studies involving six instruments and 1,106 participants, predominantly adult men. Agreement between raters was sometimes encouraging in research conditions, but results were less consistent in field settings. Findings on construct structure were limited; studies often used small or retrospective open-source samples. Critically, no included study tested true prospective predictive validity. The review did not establish that any one instrument was preferable for a particular population or operational context.

A reported postdictive association for TRAP-18 was obtained by coding cases after their outcomes were known and showed substantial between-study heterogeneity. Such a result does not mean the instrument accurately forecasts attacks in a real population. The review also identified limited evidence about convergence between tools and shortcomings in some subscale consistency estimates. A practitioner who mentions a validation coefficient should say what sample, outcome, design, and measurement procedure generated it.

ERG22+ in its original correctional context

The England and Wales Ministry of Justice psychometric investigation examined the internal characteristics and reliability of ERG22+, an assessment used to inform operational judgments about people with extremism-related convictions. Its findings are relevant to structured assessment practice in that jurisdiction. They do not show that the tool can prospectively identify an otherwise unknown future attacker in the general population. A 2024 rapid evidence assessment further mapped the research behind ERG factors and highlighted areas where evidence remains sparse or has developed.

A checklist is not a probability meter

Structured professional judgment produces a reasoned, context-dependent formulation; an ordered category or narrative opinion cannot be translated into a specific chance of offending without validated probability estimates for a defined population and time horizon. The choice of instrument must follow the referral question, cultural and linguistic context, training, and evidence of transportability. Scores alone cannot serve as proof of guilt, responsibility, or intent. They also cannot replace case evidence when decisions may restrict liberty.

Predictive validity, calibration, and false classifications

Predictive validity asks whether a prospectively administered assessment distinguishes future outcomes. Discrimination describes relative ordering across outcome groups; calibration asks whether predicted probabilities correspond to observed rates; reliability concerns consistency of ratings. These are distinct psychometric properties. A measure can show agreement between users or fit known attack histories while remaining uncalibrated for an individual in a community setting. Without prospective follow-up, the accuracy of warnings and reassurances is unknown.

Why low base rates matter

For illustration only, imagine a hypothetical screening procedure applied to 10,000 people where 10 would commit the specified outcome in the next year. If the procedure had 80% sensitivity and 90% specificity, it would identify about eight true positives, miss two cases, and falsely flag around 999 people. Under those invented assumptions, fewer than one in 100 positive screens would represent an actual future event. This is a mathematical illustration of the base-rate problem, not a performance claim about any named extremism tool.

False positives can lead to stigma, unnecessary surveillance, disproportionate coercion, strained community relations, and restrictions imposed on people who would never offend. False negatives can leave credible threats insufficiently addressed. An assessment protocol must consider the costs of each error, the legally permitted response, opportunities for correction, and the difference between offering voluntary help and authorizing coercive action. Error rates observed in one setting cannot be transferred into another without validation.

Transportability and cultural fairness

Research samples frequently overrepresent investigated, imprisoned, male, or publicly documented individuals; they can underrepresent women, youth, minority religious communities, politically different ideologies, and people who disengage without intervention. Apparent associations may partly reflect enforcement priorities and who came to official attention. Fairness evaluation requires transparent population definitions, relevant subgroup analysis, interpreters when needed, and attention to equal error burden, rather than assuming an English-language instrument is equally valid worldwide.

Changes in ideology, communication technology, policing, public awareness, and case definitions can cause performance drift. Reporting only an overall area-under-the-curve value, if one is available, conceals calibration and subgroup questions. Absence of subgroup evidence should be reported as a limitation rather than interpreted as evidence of equal accuracy. The Campbell risk-tool review identifies serious gaps in generalizability and prospective validation.

How forensic psychologists evaluate an individual case

Forensic assessment begins with the precise psycholegal question. Is the evaluator asked about alleged motive, mental state, competency, risk management, offense-linked functioning, treatment needs, or the reliability of a particular behavioral inference? Each question requires different evidence and may be controlled by different laws. The clinician documents who retained the evaluator, whose interests are affected, the limits of confidentiality, the evidence received, and the intended audience.

A defensible formulation draws on interviews where appropriate, collateral witnesses, contemporaneous messages, verified timelines, behavioral records, accessible correctional and clinical documentation, relevant testing used only for supported purposes, and competing explanations. The evaluator states which assertions are verified, disputed, inferred, or unknown. Interviews alone can be affected by impression management, fear, poor recall, language differences, and the legal stakes of the matter.

Psychological opinion should distinguish ideology, cognitive processes, emotional state, observable conduct, risk-relevant circumstances, and legal responsibility. Clinical instruments cannot determine whether a defendant satisfied the statutory elements of a particular crime, whether a statement constitutes a true threat under controlling law, or whether a jury should convict. Competency, criminal responsibility, and admissibility are separate questions with jurisdiction-specific standards. The APA forensic guidelines recommend testing rival hypotheses and communicating limitations.

Historical case comparisons sometimes help generate hypotheses, but they can create hindsight bias and false familiarity. Media accounts and public manifestos are often incomplete and curated. Unless sufficient records are available, responsible case analysis stops at documenting what is known; it does not remotely diagnose an offender or construct a confident personality profile from a dramatic event. For a broader discussion of the discipline, see Forensic Psychology: What It Is and What Forensic Psychologists Do.

Threat, motivation, intent, and criminal responsibility are different questions

A forensic threat opinion concerns future-oriented risk or current situational concern. A motive account describes an apparent reason for conduct. Intent concerns an individual's legally relevant mental state at a particular time. Criminal responsibility is the legal consequence of findings under applicable rules. These categories may be related but cannot be substituted for one another. A documented extremist belief does not itself resolve intent for a specific offense; a threat assessment cannot establish beyond a reasonable doubt that an accused person committed it.

Similarly, an extremist-affiliated criminal act may involve personal disputes, financial incentives, status seeking, coercion, group pressure, or mixed motives. Investigators and experts should evaluate competing hypotheses and the strength of contemporaneous proof. Post hoc attribution of every act to ideology may obscure the evidence. The pertinent statute and charging documents establish which motive or intent elements the prosecution must prove, if any.

Disengagement, rehabilitation, and prevention

Disengagement usually concerns leaving violent activity, organizational participation, or operational roles; deradicalization concerns changes in beliefs; desistance concerns sustained cessation of offending. A person may stop violence while retaining extreme beliefs, and a change in stated ideology may not demonstrate durable behavioral change. This distinction is central to correctional planning, release assessments, and research outcomes. A cross-field review by Raets compares these processes with the wider evidence on criminal desistance.

The 2024 NIJ synthesis of terrorism prevention-program evaluation emphasizes participant and community engagement, context-specific relevance, appropriate comparison groups, attrition, and difficulty inferring effectiveness from weak evaluation designs. Program completion, attendance, or apparent improved attitudes are intermediate results rather than demonstrated prevention of future attacks. Evaluations should describe which outcomes were monitored, for how long, and against which comparison.

A 2025 scoping review of tertiary psychological interventions identified programs across multiple countries and documented considerable variation in intervention delivery. Its findings map existing practice but do not by themselves establish which psychotherapy or rehabilitation package prevents recidivism. Ethical services should focus on assessed needs, voluntary engagement where possible, lawful case management, meaningful social reintegration, and measurable outcomes without assuming a standard treatment fits every ideological background.

Preventive work can include responses to tangible threats, access to educational and employment supports, nonviolent identity reconstruction, family engagement when safe and appropriate, and opportunities to leave coercive networks. Effective support requires individualized assessment and an awareness of unintended harms. The degree of evidence is generally stronger for identifying implementation challenges than for estimating a reliable reduction in rare violent-extremist events.

Evidence quality and the questions research still cannot answer

The field faces unavoidable obstacles: rare outcomes, classification disagreements, classified intelligence, unequal detection, low numbers of matched controls, politically variable labels, publication bias, and the ethical difficulty of experimenting on high-consequence risk. Completed-attack case series provide valuable descriptive information but cannot estimate how likely a person with similar experiences is to offend. Conviction-based cohorts omit undetected acts and people who never passed into criminal justice systems.

A systematic review has stronger synthesis methods than a single case report, but its conclusions remain bounded by included study designs and dates. The 2025 Campbell review was published in 2025 while its searches covered studies available through December 2021. This matters for any claim that a newly released tool has been comprehensively evaluated by that review. Later validation and current operational practices require separate examination. Official agency guidance describes professional practices, not automatically independent scientific validation.

Meaningful future research should include preregistered prospective cohorts, transparent definitions of intended outcomes, external validation across jurisdictions and ideologies, calibration plots, reliable false-positive estimates, subgroup analysis, protected-rights auditing, and long follow-up for desistance. Ethical research must also test whether operational decisions made because of a risk signal change the observed outcome, making evaluation more complex.

Common myths and misconceptions

Myth: extremists share one diagnosable personality

No demonstrated psychological profile identifies all violent extremists. The research concerns heterogeneous people whose affiliations, motives, contexts, resources, and behavior vary. Psychological characteristics sometimes help explain individual cases when there is adequate corroboration, but they cannot be promoted to a universal offender template.

Myth: radical beliefs predict an attack

Extremist or offensive expression can matter as part of documented context, but belief alone is not proof of preparation or future violence. The legal significance of expression differs across jurisdictions. In the United States, investigative and criminal-justice decisions must respect constitutionally protected activity. The FBI's 2025 guidance makes contextual evaluation and rights protection explicit.

Myth: a high risk score proves someone will be violent

A score is an assessment output, not an event or legal finding. In specialized violent-extremism assessment, the strongest available synthesis found important gaps in prospective validity. Any court or public-safety decision should specify the evidence, instrument limits, base rates, alternatives, and legal authority rather than translating an instrument result into certainty.

Myth: internet exposure causes extremist offending

Online encounters can shape socialization and facilitate organization, but correlation with content consumption does not identify a causal mechanism for a particular case. A credible account must examine previous beliefs, offline circumstances, chronology, peer connections, and actual conduct, while acknowledging missing digital records and platform bias.

Frequently asked questions

Is violent extremism a mental disorder?

No. It is a behavioral, social, policy, and sometimes legal category. A person can have psychiatric symptoms and engage in extremist offending, but neither fact establishes the other. Mental-health assessment follows clinical standards and requires individualized evaluation.

What is the difference between radicalization and violent mobilization?

Radicalization refers broadly to a process involving changes in ideas, allegiance, or orientation toward extreme positions. Violent mobilization concerns preparation and movement toward action. A person may be ideologically radical without being on a pathway to violence. Operational definitions vary, and researchers should state them clearly.

Can psychologists predict a terrorist attack?

Current specialist instruments do not provide a validated, precise probability of a specific individual's future terrorist attack in general populations. Assessors can identify contextual concerns, document observable changes, test hypotheses, and contribute to threat management, while clearly communicating uncertainty and error risk.

Are lone actors always socially isolated?

No. The label usually refers to the operational configuration of an attack rather than the person's complete social history. Some people who act without direct operational partners still maintain social or online ties. Our related published article on Lone-Actor Violence and Threat Assessment examines those distinctions and the limits of threat inference.

How do VERA-2R and TRAP-18 differ from ordinary clinical tests?

They are specialist risk or threat frameworks used to structure assessment of particular processes and evidence. They are not general psychiatric diagnostic tests and should not be interpreted as such. Their applicability and empirical support vary by context and population.

What does a false positive mean in this field?

A false positive occurs when an assessment indicates a specified future event or condition that does not occur. Its meaning depends on a defined outcome and observation period. Rare events can yield many false positives even when a hypothetical screening system has high sensitivity and specificity; the tool's actual performance must be validated, not assumed.

Does a previous criminal conviction prove extremist motivation?

No. Conviction establishes only what a court legally found under the relevant law. Extremist motive may be part of an offense, an aggravating fact, a disputed narrative, or irrelevant to the particular charge. A psychologist should not infer it from generic criminal history.

Can someone disengage without changing every belief?

Yes. Behavioral disengagement and ideological change are separate outcome dimensions. Prevention and reintegration programs should monitor conduct, support conditions, and the durability of change while avoiding the assumption that verbal belief change proves rehabilitation.

What should a high-quality forensic report contain?

It should identify the referral question, jurisdiction, information sources, relevant timeline, collateral verification, assessment methods, competing explanations, limitations, population validity of any tool, cultural considerations, and conclusions framed at the level that the evidence supports. It should separate observed acts from hypotheses, risk judgments from probabilities, and clinical opinions from legal findings.

How does this topic relate to ordinary violence-risk psychology?

There is overlap in methods, ethics, and risk-management principles, but ideologically motivated offending has additional social, legal, and sampling complexities. The parent evidence on generic future violence is explored in our Violence Risk Assessment article, while Why Do People Commit Crimes? owns the broader general-criminal-behavior question.

References

American Psychological Association. (2013). Specialty guidelines for forensic psychology. American Psychologist, 68, 7–19. APA official guideline.

Aryaeinejad, K., & Scherer, T. L. (2024). Evaluating and assessing terrorism prevention programs: What research sponsored by the National Institute of Justice tells us. National Institute of Justice. Official report.

Brouillette-Alarie, S., et al. (2025). Reliability and validity of risk assessment tools for violent extremism: A systematic review. Campbell Systematic Reviews, 21(4), e70080. DOI: 10.1002/cl2.70080.

Chermak, S., Freilich, J., Ross, A., & Turner, N. (2023). Examining radicalization's risk and protective factors: A case-control study. National Institute of Justice. Research record.

Federal Bureau of Investigation, National Counterterrorism Center, & Department of Homeland Security. (2025). U.S. Violent Extremism Mobilization Indicators, 2025 edition. Official guidance.

Ministry of Justice. (2023). Extremism Risk Guidance 22+: An exploratory psychometric analysis. Government of the United Kingdom. Official report.

National Institute of Justice. (2024). Domestic radicalization and violent extremism: An overview of NIJ's research portfolio. NIJ article.

Raets, S. (2024). Desistance, disengagement, and deradicalization: A cross-field comparison. International Journal of Offender Therapy and Comparative Criminology, 68(4), 389–426. DOI: 10.1177/0306624X221102802.

Sarma, K. M., Carthy, S. L., & Cox, K. M. (2022). Mental disorder, psychological problems and terrorist behaviour: A systematic review and meta-analysis. Campbell Systematic Reviews, 18(3), e1268. DOI: 10.1002/cl2.1268.

Savoia, E. (2025). The use of psychological interventions in tertiary prevention programs for individuals engaged in violent extremism: A scoping review and interviews. National Institute of Justice. Research record.

Silke, A. (2024). Factors in terrorist risk assessment: A rapid evidence assessment of the extremism risk guidance (ERG22+) factors. Journal of Criminal Psychology. DOI: 10.1108/JCP-04-2024-0035.

United States Congress. (Current codification). 18 U.S.C. § 2331, Definitions. Office of the Law Revision Counsel. U.S. Code.

United States Department of Justice. (2022). Assistant Attorney General Matthew G. Olsen delivers keynote address at the George Washington University Program on Extremism symposium. Official statement.

Winter, C., Neumann, P., Meleagrou-Hitchens, A., Ranstorp, M., Vidino, L., & Fürst, J. (2020). Online extremism: Research trends in internet activism, radicalization, and counter-strategies. International Journal for Conflict and Violence. DOI: 10.4119/ijcv-3809.

Psychological Encyclopedia