Criminology is the systematic study of crime, delinquency, victimisation, their causes and consequences, and the ways societies prevent and respond to them. It asks why some harmful or prohibited behaviours occur, how crime is measured, why risk varies across places and times, what happens to victims and communities, which institutions respond, and which interventions actually reduce harm.
The American Society of Criminology describes its field as pursuing scholarly, scientific and professional knowledge concerning the measurement, etiology, consequences, prevention, control and treatment of crime and delinquency. Source: American Society of Criminology.
Criminology should be distinguished from criminal law and criminal justice. Criminal law defines offences and legal liability within a jurisdiction. Criminal justice concerns institutions such as police, prosecution, courts, corrections and legal aid. Criminology studies those systems while also asking broader social, behavioural, environmental and statistical questions about crime and harm.
Reading routes: begin with the child-friendly explanation; follow crime measurement; explore major explanatory theories, crime prevention, criminal justice, reoffending and reintegration, ethics and research methods; then use the learning workshop.
This is an academic guide. It does not provide tactics for committing or evading detection of crime, and it does not support profiling people by race, ethnicity, religion, gender or other protected characteristics. Criminological explanation should remain proportional to evidence and respectful of victims, accused persons, convicted persons and communities.
Explain criminology to a child: count carefully, ask why, then test what helps
Imagine a school has ten missing bicycles. One person might say, “There is a bicycle crime problem.” A criminologist asks more questions.
Were ten bicycles stolen in one week or ten years? Were they all taken from one place? How many bicycles were parked there? Did people report every theft? Did a new bike rack reduce losses later?
The first lesson is that a number needs a denominator, a time period and a definition.
The second lesson is that explaining a pattern and preventing it are different jobs. A good theory should help us make predictions that evidence can test.
1. Crime is partly a legal category and partly a measured social phenomenon
What counts as a criminal offence depends on law. Laws vary across jurisdictions and change through time.
Criminologists therefore distinguish legal definitions from broader categories of harmful behaviour and victimisation.
An offence count is not a pure measure of underlying behaviour. It can also reflect reporting, police recording, legal definitions and enforcement priorities.
Good criminological analysis begins by defining the event being counted and the institutional process through which it becomes data.
2. Police-recorded crime captures only part of the picture
Administrative crime data are valuable because they record offences known to authorities and can provide detailed incident information.
But not every victim reports an offence, and not every report is recorded in the same way.
Victimisation surveys provide another route by asking samples of people about experiences whether or not they were reported to police.
The US Bureau of Justice Statistics explicitly compares police-recorded and victimisation-survey measures because they use different methods and cover overlapping but non-identical phenomena. Source: BJS, The Nation’s Two Crime Measures, March 2026.
3. The “dark figure” of crime is the part not visible in official records
Some offences have high reporting rates. Others may be hidden because victims fear retaliation, distrust authorities, do not recognise the event as criminal, or believe reporting will not help.
Underreporting can vary by offence, population and context.
This means a lower recorded crime rate can sometimes reflect less crime, less reporting, different recording or several mechanisms together.
Criminologists therefore triangulate sources instead of treating one dataset as complete reality.
4. A crime count and a crime rate answer different questions
Original teaching example: Area A records 200 burglaries among 100,000 residents. Area B records 150 burglaries among 50,000 residents.
Area A has more burglaries in absolute count. Its population rate is 200/100,000 × 100,000 = 200 per 100,000 residents.
Area B’s rate is 150/50,000 × 100,000 = 300 per 100,000 residents.
Area B therefore has the lower count but higher rate under this simplified population denominator.
The example also has a limitation: residential population may not be the best exposure denominator for every offence. A transport hub, nightlife district or commercial area can contain many non-residents.
5. Rates need an appropriate population at risk
Vehicle theft should be related to vehicles or vehicle exposure as well as people. Shoplifting may depend on retail activity. Workplace offences depend on workers and operating conditions.
A denominator that is convenient but unrelated to exposure can distort comparison.
Criminologists therefore ask which population had a meaningful opportunity to experience the event.
The same discipline appears in epidemiology and public health: a rate is only interpretable when numerator, denominator, time and population belong together.
6. Repeated victimisation changes how harm is distributed
Crime is not always evenly spread. Some people, places or facilities experience repeated incidents.
This matters because the number of incidents and number of victims answer different questions.
Ten offences can involve ten different victims or one victim harmed ten times.
Prevention may therefore need to protect highly exposed locations or support repeat victims rather than assuming the next incident is equally likely everywhere.
7. Hot spots are concentrations, not automatic explanations
Crime can cluster at particular streets, facilities or micro-places.
A hot spot identifies where incidents concentrate. It does not by itself explain why.
Possible mechanisms include foot traffic, guardianship, land use, transport, alcohol availability, target concentration, reporting patterns or enforcement.
A place-based intervention should therefore test the local mechanism instead of treating the map as the cause.
8. Criminological theories explain different levels of behaviour
There is no single theory of crime that explains every offence.
Some theories focus on individual development. Others focus on peer learning, social bonds, neighbourhood structure, opportunity, inequality, institutions or life-course transitions.
A good theory identifies variables and mechanisms that produce testable expectations.
Theories should be compared by explanatory scope and evidence rather than treated as ideological labels.
9. Routine activity theory focuses on opportunity
Routine activity theory examines how crime can occur when a motivated offender, suitable target and absence of capable guardianship converge in time and space.
The theory directs attention to everyday patterns of movement and opportunity.
It does not claim that victims are responsible for offences committed against them.
Its value is analytical: changing guardianship, target accessibility or routine activity can change opportunity even if offender motivation remains unchanged.
10. Situational crime prevention changes immediate opportunity
Situational prevention seeks to make specific crimes harder, riskier or less rewarding while reducing provocations or excuses where appropriate.
Examples can include improved access control, lighting, secure storage, transaction design or product changes.
The relevant question is whether the intervention reduces the intended harm without merely displacing it to another location, time, target or offence.
Displacement is possible but not inevitable; diffusion of benefits can also occur when prevention effects extend beyond the treated location.
11. Social learning theories examine how behaviour is learned
People can learn definitions, skills, rewards and norms through social interaction.
Peer groups can therefore influence behaviour, but association does not automatically prove causal influence.
People also select friends with similar interests and backgrounds.
Research needs designs capable of distinguishing influence from selection as far as possible.
12. Social control and social bond theories ask why people conform
Instead of asking only why crime occurs, control theories ask why most people refrain from many prohibited acts.
Attachment, commitment, involvement and belief can connect people to conventional institutions and expectations.
Theories of informal social control also examine how families, communities and organisations regulate behaviour outside formal punishment.
These ideas should not be reduced to blaming communities for crime. Structural conditions can shape the resources available for informal control.
13. Strain theories examine pressures and blocked goals
Strain theories propose that certain pressures, losses or blocked opportunities can increase the likelihood of deviant coping for some people.
Theories differ in which forms of strain matter and how emotions, coping resources and social context influence outcomes.
Most people experiencing hardship do not commit crime.
Theories therefore describe probabilistic mechanisms rather than deterministic predictions about individuals.
14. Social disorganisation theories examine neighbourhood structure
Some neighbourhood conditions can weaken collective capacity to maintain informal social control or provide stable support.
Residential instability, concentrated disadvantage and institutional weakness have been studied as possible mechanisms.
These are area-level relationships and should not be converted into stereotypes about residents.
Communities can also display resilience, networks and collective efficacy that reduce harm despite structural disadvantage.
15. Life-course criminology studies change through age and transitions
Criminal behaviour often changes across the life course. Many people who offend in adolescence do not continue indefinitely.
Research examines onset, persistence, desistance and the influence of relationships, work, parenthood, identity and institutions.
Age patterns are population tendencies, not destiny for an individual.
This perspective matters because prevention and rehabilitation strategies should fit developmental stage and personal circumstances rather than assuming fixed criminal identity.
16. Developmental criminology studies early risk and protective factors
Childhood adversity, school experiences, family conditions, peer environments and individual traits can correlate with later offending.
Risk factors are not labels that determine a child’s future.
Good developmental research examines combinations of risk and protective factors and avoids treating statistical association as moral character.
Early intervention should support development and reduce harm, not stigmatise children as future criminals.
17. Biosocial research requires especially careful interpretation
Criminology can examine biological, neurological and genetic correlates of behaviour alongside social and environmental influences.
These effects are complex and usually probabilistic.
Genetic or biological association must not be used to justify racial essentialism or deterministic profiling.
Responsible biosocial criminology separates individual variation from socially constructed population categories and recognises strong environmental interaction.
18. White-collar and corporate crime expand the field beyond street offences
Criminology also studies fraud, corruption, regulatory offences, occupational misconduct and harmful organisational behaviour.
These offences can be difficult to measure because victims may be diffuse, transactions complex and detection dependent on audits or regulators.
The social harm can be large even when the offence lacks a dramatic physical event.
This reminds us that visible street crime is only one part of the criminological landscape.
19. Cybercrime changes scale, jurisdiction and victimisation patterns
Digital offences can cross borders rapidly and affect many victims at once.
Measurement is difficult because one incident can involve many systems and victims across jurisdictions.
Criminology studies victimisation, offender organisation, reporting and prevention without needing to reproduce operational intrusion methods.
The field’s core questions remain: what is the harm, how is it measured, which mechanisms produce opportunity, and which interventions reduce harm?
20. Crime prevention can act before, during and after offending
Prevention can reduce the conditions producing crime, protect likely targets, interrupt immediate opportunities, support victims or reduce reoffending.
UN crime-prevention guidelines define prevention broadly as strategies and measures that reduce the risk of crimes and their harmful effects by influencing multiple causes. Source: UN Guidelines for the Prevention of Crime.
This is broader than enforcement. Education, health, urban design, family support, treatment and community institutions can all have preventive roles when evidence supports them.
The correct intervention depends on the mechanism and population, not on a generic desire to “get tough” or “be soft.”
21. Prevention should be evaluated against displacement and unintended effects
Suppose improved bicycle security reduces thefts at one station. Did theft fall overall or move to the next station?
Did users stop cycling because security procedures became too burdensome?
Did enforcement disproportionately affect legitimate users?
A complete evaluation therefore measures intended outcomes and important side effects.
22. Worked prevention example: absolute and relative change
Original model: a site records 100 thefts in one comparable period and 70 after an intervention.
Absolute reduction is 30 thefts. Relative reduction is 30/100 = 30 per cent.
Now suppose a comparison site fell from 100 to 85 during the same period without the intervention. The treated site’s additional difference is smaller than the raw 30-per-cent figure suggests.
This is why criminologists use comparison groups, time-series designs or other causal methods when possible.
23. Criminal justice is a chain of institutions and decisions
The justice system can include police, prosecutors, defence, courts, corrections, legal aid, victim services and alternatives to imprisonment.
UNODC describes effective criminal justice as needing to be fair, humane and accountable and linked to rule of law and human rights. Source: UNODC Crime Prevention and Criminal Justice.
Each institution has different authority and evidence standards.
Criminology studies how decisions accumulate across the chain, including possible disparities and unintended consequences.
24. Police data reflect both offending and institutional activity
Arrests depend on offences, reporting, detection, police deployment and legal practice.
An increase in arrests for an offence can reflect more offending, more enforcement, improved detection or changed policy.
Researchers therefore should not use arrest counts as a direct measure of underlying offending without additional evidence.
The interpretation requires a model of how events become police data.
25. Courts operate through legal standards, not criminological prediction
Criminological research can study sentencing patterns, procedural fairness and outcomes.
It should not replace the legal standards governing proof, rights and individual responsibility.
Population-level risk factors are not evidence that a particular accused person committed an offence.
This separation between statistical association and individual legal proof is fundamental.
26. Corrections serves several possible purposes
Custodial and community sanctions can be justified through combinations of punishment, incapacitation, deterrence, rehabilitation and restoration depending on the legal system.
Criminology studies whether these mechanisms operate as intended and at what social cost.
Longer punishment does not automatically produce proportionally stronger deterrence.
Correctional research therefore examines behaviour, institutional conditions, reintegration and reoffending rather than assuming one mechanism dominates every case.
27. Reoffending measurement needs a definition
Recidivism can mean rearrest, reconviction, reincarceration or a new offence over a stated follow-up period.
Different definitions produce different rates.
A programme can appear successful under one measure and less successful under another.
Good reporting therefore states the event, follow-up time, population and comparison group.
28. Rehabilitation targets changeable needs and capabilities
Rehabilitation can address education, employment, substance-use problems, cognitive skills, relationships, housing and other needs linked to reoffending.
Interventions should fit the individual’s assessed needs and risks rather than applying the same programme to everyone.
UN standards and recent UNODC work emphasise rehabilitation, reintegration and evidence-based prison management while respecting human rights. Source: UN work on reducing reoffending.
Support after release can matter because housing, employment and social connection shape the environment into which a person returns.
29. Desistance studies how people stop offending
Desistance is often a process rather than one moment.
People may reduce frequency or seriousness before stopping completely. Identity, relationships, opportunity and institutional support can all matter.
Labelling a person permanently by past offending can undermine reintegration when legitimate roles are difficult to obtain.
Criminology therefore studies change and resilience as seriously as onset and persistence.
30. Criminological research carries ethical responsibility
Research can involve victims, accused persons, prisoners, children and other populations with unequal power or vulnerability.
Confidentiality, informed consent, data security and minimisation of harm are therefore central.
The American Society of Criminology’s ethical code emphasises rights, dignity and worth of research participants, victims, accused persons and convicted persons. Source: ASC Code of Ethical Conduct overview.
Researchers should also consider how findings can be misused to stigmatise communities or justify discrimination.
31. Algorithmic risk assessment raises measurement and fairness questions
Statistical models can estimate probabilities using historical data.
But historical data reflect past policing, prosecution and social conditions as well as behaviour.
A model can therefore reproduce institutional patterns unless the target, inputs and validation are examined carefully.
Accuracy, calibration, group performance, interpretability and the consequences of false positives and false negatives all matter.
32. Predictive accuracy is not moral or legal justification
A model can predict an outcome reasonably well while using variables that are inappropriate for a particular decision.
It can also be useful for resource planning but unsuitable for decisions affecting an individual’s liberty.
The use case determines the ethical and legal threshold.
Criminology should therefore keep prediction separate from causation, responsibility and justice.
33. Criminology uses multiple research methods
Surveys can measure victimisation and attitudes. Administrative data can show institutional events. Interviews and ethnography can reveal lived experience and organisational practice.
Experiments and quasi-experiments can evaluate interventions. Spatial analysis can examine place-based patterns. Longitudinal studies can follow change through time.
Each method has blind spots.
Strong evidence often comes from several methods converging on the same mechanism.
34. Victimisation surveys reveal offences not known to police
The BJS National Crime Victimization Survey has collected information on nonfatal victimisation reported and not reported to police since 1973 in the United States. Source: BJS National Crime Victimization Survey.
The survey illustrates a key research principle: administrative records and population surveys have different observation processes.
Neither is automatically the “real” count.
Understanding the difference between them can reveal reporting behaviour, institutional coverage and measurement error.
35. International comparison requires common definitions
Countries can differ in law, reporting practices, police recording and data systems.
UNODC’s International Classification of Crime for Statistical Purposes and data frameworks seek to improve comparability by organising offences around behavioural concepts rather than only national legal labels.
Comparison still requires attention to coverage and data quality.
A ranking table without methodological context can give false precision.
36. Causal inference asks what would have happened without the intervention
Suppose a city introduces a prevention programme and crime falls. The causal question is whether crime would have fallen anyway.
Randomised experiments can help when feasible and ethical. Natural experiments, matched comparisons, regression discontinuity, interrupted time series and difference-in-differences can also strengthen inference under assumptions.
No statistical method automatically removes bias.
The researcher must explain why the comparison approximates the missing counterfactual.
37. Common misconceptions
“Crime statistics simply count crime.” They count events passing through defined observation systems.
“A high-crime area is full of criminals.” Area-level rates do not describe every resident and can reflect opportunity, land use and institutional patterns.
“One cause explains crime.” Offending is heterogeneous and arises from multiple mechanisms across individuals, situations and structures.
“Harsher punishment always means less crime.” Deterrence depends on perceived certainty, timing, behaviour and context; effects require empirical evaluation.
“Risk factors identify who will offend.” Risk factors shift probabilities and are not deterministic labels for individuals.
38. Learning workshop with worked answers
Question A: Area A has 200 burglaries among 100,000 residents; Area B has 150 among 50,000. Which has the higher population rate? Answer: B, with 300 per 100,000 compared with A’s 200.
Question B: police-recorded assault falls while victimisation-survey assault remains stable. Give two plausible explanations. Answer: reporting or recording may have changed, or the measures may cover different populations or offence definitions. More evidence is required.
Question C: theft falls from 100 to 70 after an intervention, while a comparison site falls from 100 to 85. Can we attribute the entire 30-per-cent fall to the intervention? Answer: no. Background trends suggest part of the decline may have occurred without it.
Question D: an arrest count rises after police increase patrols. Does underlying crime necessarily rise? Answer: no. Detection and enforcement changed as well.
Question E: a risk model predicts well at population level. Can it prove an individual committed a crime? Answer: no. Prediction and individual legal proof are different kinds of inference.
39. A learning progression from counting to causal explanation
Primary learners can begin with fair comparison: counts, rates and why the denominator matters.
Secondary learners can study reporting bias, opportunity, prevention and the difference between correlation and causation.
Advanced learners can add longitudinal data, causal inference, spatial statistics, programme evaluation, criminal justice institutions and ethical analysis.
The strongest assignment is to evaluate one crime-prevention claim using multiple data sources and state what evidence would change the conclusion.
40. Frequently asked questions
Is criminology the same as criminal justice?
No. Criminal justice concerns institutions and legal processes. Criminology studies crime and social responses more broadly, including measurement, theory and prevention.
Why do crime statistics disagree?
Datasets can measure different events through different observation systems. Police reports, victim surveys and court records each answer different questions.
Does poverty cause crime?
There is no simple deterministic relationship. Economic conditions can interact with opportunity, institutions, inequality, social networks and individual factors. Most people experiencing poverty do not commit crime.
Can crime be prevented without policing?
Some prevention acts through design, education, health, social services, community support and product security. Police and criminal justice are one part of a broader prevention system.
Does rehabilitation ignore victims?
It should not. Effective justice systems can protect victims and public safety while also reducing future offending through evidence-based rehabilitation and reintegration.
41. Working glossary
Criminology: study of crime, delinquency, victimisation, causes, consequences and social responses. Victimisation: experience of a criminal event by a person or household under a defined measure. Dark figure: crime not captured by a particular official data system.
Crime rate: offences or victimisations relative to a defined population and time. Hot spot: place where incidents concentrate relative to surrounding areas. Recidivism: repeated criminal-justice event under a stated definition and follow-up period.
Deterrence: reduction of offending because anticipated consequences change behaviour. Incapacitation: prevention of offending through restriction of opportunity to offend. Rehabilitation: interventions intended to change capabilities or conditions associated with future offending. Desistance: process of reducing and ending offending.
42. Evidence and further reading
Primary routes include the American Society of Criminology; the UNODC Data Portal; the UN crime prevention and criminal justice standards; the BJS National Crime Victimization Survey; and the BJS 2026 comparison of victimisation and police-recorded crime measures.
The deeper answer: criminology works by refusing to confuse the event, the record and the explanation
A crime, a police report, an arrest, a conviction and a prison admission are different events in different systems. Criminology connects them without pretending they are interchangeable.
The discipline becomes useful when it measures carefully, tests competing causes, evaluates prevention honestly and remembers that statistical patterns describe probabilities and institutions—not fixed identities of people or communities.
Continue: Law explains legal rules and institutions; Sociology explains social structure and institutions; Psychology explains behaviour, cognition and development. Return to the How X Works Hub for the complete subject map.